I've updated all of the separation applications to 1.00. For changes people might care about,
- - The old CAL version is gone; it's replaced with the OpenCL application. On AMD/ATI GPUs (older than 79xx) it is using some hackery to use the same IL kernel as before so it should be as fast. However this also means the Radeon 38xx cards aren't supported with new stuff.
- Radeon 79xx stuff should work
- The occasional validate errors from empty / truncated stderr should stop
- AVX will be used if available on Linux and Windows (64 bit only for Windows)
- I've increased the default GPU target frequency so GPU stuff should make things less laggy on average. You can now also configure this with the web preferences now so you don't need to use app_info stuff if you want to play with that.
- Partial workaround with high CPU usage with recent Nvidia drivers.*
As usual post problems you run into here.
* It should cut down on the CPU usage a bit while not sacrificing too much. I would recommend not using it unless you are very unhappy with the CPU usage on Nvidia. There are options to change the polling mode if you want to lower the CPU usage further while not slowing it down. (--gpu-wait-factor (default = 0.75) and --gpu-polling-mode (default = - 2) work similarly to how they did with the old CAL one, but slightly different). With the default of -2 it will use mode -1 unless it is an Nvidia driver newer than the one that introduced the high CPU issue, where it will use mode 0. Mode -1 uses the correct waiting method, mode 0 use the correct waiting method with an initial sleep based on time estimates, and modes > 0 are a polling period in milliseconds. The wait factor is a sort of correction of the time estimate used for the initial wait. The default is 0.75, to wait for 75% of the estimated time before trying to poll.
/ 
Reply
Quote
Hi Matt,
since a couple of minutes I do not get wu's.
Servers are running, I found your post here, resetted the project.
Message log says 'Not reporting or requesting tasks'
08.02.2012 20:39:28 | Milkyway@Home | Sending scheduler request: Requested by user.
08.02.2012 20:39:28 | Milkyway@Home | Not reporting or requesting tasks
08.02.2012 20:39:31 | Milkyway@Home | Scheduler request completed
08.02.2012 07:03:56 | | Libraries: libcurl/7.21.6 OpenSSL/1.0.0d zlib/1.2.5
08.02.2012 07:03:56 | | Data directory: E:\Boinc
08.02.2012 07:03:56 | | Running under account alex
08.02.2012 07:03:56 | | Processor: 9 GenuineIntel Intel(R) Core(TM) i7 CPU 860 @ 2.80GHz [Family 6 Model 30 Stepping 5]
08.02.2012 07:03:56 | | Processor: 256.00 KB cache
08.02.2012 07:03:56 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 sse4_2 syscall nx lm vmx smx tm2 popcnt pbe
08.02.2012 07:03:56 | | OS: Microsoft Windows 7: Professional x64 Edition, Service Pack 1, (06.01.7601.00)
Any hints?
/ Reply
Quote
I've tried it on my other machine, which has latest AMD-drivers and BM 7.0.12.
A single wu was downloaded, it failed with exit code 1
http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=98920851
Albert and SETI Ati-OpenCL wu's work fine there.
hostid=398327
Alexander
/ Reply
Quote
I've tried it on my other machine, which has latest AMD-drivers and BM 7.0.12.Looks like BOINC is trying to run it on a non-double GPU
A single wu was downloaded, it failed with exit code 1
http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=98920851
Albert and SETI Ati-OpenCL wu's work fine there.
hostid=398327
Alexander
/ Reply
Quote
So maybe BM is the source?
I'm running MW for years now, unless my gpu's haven't lost some bits they are DP!
On my A8-system you may be right, the onboard GPU isn't DP, but the installed HD5830 is. Is there a way to dedicate this gpu to MW?
/ Reply
Quote
One of my systems is not getting WUs now either. It did at one point give a message about it being a non-DP GPU which is not the case - it is a 5970, i updated the drivers and copied the app_info from the system which was getting WUs but it still doesn't work.
/ Reply
Quote
So maybe BM is the source?As far as I know BOINC has always had this absolutely horrid defect where the capabilities of the best GPU = the capabilities of ALL the GPUs, which is why you need to create the cc_config.xml file to use all of your GPUs because this stuff can happen. I still can't believe this is a thing. You can block that GPU from a specific project I guess by adding something like this where you have the <use_all_gpus>1</use_all_gpus> enabled.
I'm running MW for years now, unless my gpu's haven't lost some bits they are DP!
On my A8-system you may be right, the onboard GPU isn't DP, but the installed HD5830 is. Is there a way to dedicate this gpu to MW?
<exclude_gpu>
<url>http://milkyway.cs.rpi.edu/milkyway</url>
<device_num>1</device_num>
<type>ati</type>
<app>milkyway</app>
</exclude_gpu>
/ Reply
Quote
Radeon HD 7970 isn't getting any WUs (I reset the project).
I'm also running the BM 7.0.8.
/ Reply
Quote
Albert@home requires a newer BM; one post there says that BM 6.12.34 has faults regarding openCL and ATI. Version 7.0.3 and higher required.
I will wait some time until my running wu's are finished then I will update BM.
/ Reply
Quote
No work here, all done.......
/ Reply
Quote
08.02.2012 21:42:26 | | Starting BOINC client version 7.0.14 for windows_x86_64
08.02.2012 21:42:26 | | log flags: file_xfer, sched_ops, task
08.02.2012 21:42:26 | | Libraries: libcurl/7.21.6 OpenSSL/1.0.0d zlib/1.2.5
08.02.2012 21:42:26 | | Data directory: E:\Boinc
08.02.2012 21:42:26 | | Running under account alex
08.02.2012 21:42:26 | | Processor: 9 GenuineIntel Intel(R) Core(TM) i7 CPU 860 @ 2.80GHz [Family 6 Model 30 Stepping 5]
08.02.2012 21:42:26 | | Processor: 256.00 KB cache
08.02.2012 21:42:26 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 sse4_2 syscall nx lm vmx smx tm2 popcnt pbe
08.02.2012 21:42:26 | | OS: Microsoft Windows 7: Professional x64 Edition, Service Pack 1, (06.01.7601.00)
08.02.2012 21:42:26 | | Memory: 7.99 GB physical, 15.98 GB virtual
08.02.2012 21:42:26 | | Disk: 31.25 GB total, 15.17 GB free
08.02.2012 21:42:26 | | Local time is UTC +1 hours
08.02.2012 21:42:26 | | VirtualBox version: 4.1.8
08.02.2012 21:42:26 | | ATI GPU 0: AMD Radeon HD 6900 series (Cayman) (CAL version 1.4.1546, 2048MB, 2016MB available, 5632 GFLOPS peak)
08.02.2012 21:42:26 | | ATI GPU 1: ATI Radeon HD 5800 series (Cypress) (CAL version 1.4.1546, 1024MB, 991MB available, 4608 GFLOPS peak)
08.02.2012 21:42:26 | | OpenCL: ATI GPU 0: Cayman (driver version CAL 1.4.1546 (VM), device version OpenCL 1.1 AMD-APP-SDK-v2.5 (732.1), 2048MB, 2016MB available)
08.02.2012 21:42:26 | | OpenCL: ATI GPU 1: Cypress (driver version CAL 1.4.1546 (VM), device version OpenCL 1.1 AMD-APP-SDK-v2.5 (732.1), 1024MB, 991MB available)
08.02.2012 21:42:26 | | ATI GPU 0 is OpenCL-capable
08.02.2012 21:42:26 | SETI@home | Found app_info.xml; using anonymous platform
08.02.2012 21:42:26 | | Config: simulate 9 CPUs
08.02.2012 21:42:26 | | Config: use all coprocessors
08.02.2012 21:42:26 | | Version change (6.12.34 -> 7.0.14)
08.02.2012 21:43:24 | Milkyway@Home | Sending scheduler request: Requested by user.
08.02.2012 21:43:24 | Milkyway@Home | Not reporting or requesting tasks
08.02.2012 21:43:25 | Milkyway@Home | Scheduler request completed
My gpu's run idle ..
Edit:
HD5850 & HD6950, both should be openCL-enabled ...
Edit 2:
installed albert, both gpu's are crunching the wu's on this machine
/ Reply
Quote
All my tasks are failing with the new app.
<![CDATA[
<message>
process exited with code 1 (0x1, -255)
</message>
<stderr_txt>
../../projects/milkyway.cs.rpi.edu_milkyway/milkyway_separation_1.00_x86_64-pc-linux-gnu: /lib64/libc.so.6: version `GLIBC_2.7' not found (required by ../../projects/milkyway.cs.rpi.edu_milkyway/milkyway_separation_1.00_x86_64-pc-linux-gnu)
</stderr_txt>
]]>
Did the makefile for the app change, because this was not an issue before.
I realize that my glibc is older, but things worked before, and I am not in a position to update them at this time.
____________
/ Reply
Quote
2/8/2012 11:34:51 AM | Milkyway@Home | Requesting new tasks for CPU and ATI GPU
2/8/2012 11:34:52 AM | Milkyway@Home | Scheduler request completed: got 0 new tasks
2/8/2012 11:34:52 AM | Milkyway@Home | No tasks sent
2/8/2012 11:34:52 AM | Milkyway@Home | Message from server: Host device lacks OpenCL
2/8/2012 11:34:52 AM | Milkyway@Home | Message from server: Your NVIDIA GPU lacks the needed compute capability (1.3, required for double precision math
Seems the NVIDIA and ATI is mixed up...
/ Reply
Quote
2/8/2012 11:34:51 AM | Milkyway@Home | Requesting new tasks for CPU and ATI GPU
2/8/2012 11:34:52 AM | Milkyway@Home | Scheduler request completed: got 0 new tasks
2/8/2012 11:34:52 AM | Milkyway@Home | No tasks sent
2/8/2012 11:34:52 AM | Milkyway@Home | Message from server: Host device lacks OpenCL
2/8/2012 11:34:52 AM | Milkyway@Home | Message from server: Your NVIDIA GPU lacks the needed compute capability (1.3, required for double precision math
Seems the NVIDIA and ATI is mixed up...
I have the same problem.
Just retried now that Matt fixed the ATI preference gone, but did not change...
/ Reply
Quote
Did the makefile for the app change, because this was not an issue before.I statically linked the non-GPU Linux ones last time which isn't really a good thing to do. I rebuilt it and called it 1.01 as a static version for the CPU ones.
I realize that my glibc is older, but things worked before, and I am not in a position to update them at this time.
/ Reply
Quote
I have them all downloaded and will update the app_info versions on my site.
I still don't know if I will be updating from 0.82 yet due to the OpenCL CPU bug that still exists. Especially on a mixed GPU machine.
____________
/ Reply
Quote
Right now I'm suspecting that people aren't getting work because it seems that clients < 7 don't report having OpenCL and I updated the scheduler to check for the new stuff.
/ Reply
Quote
I'm using 6.12.34 and I'm getting work.
/ Reply
Quote
I'm using 6.12.34 and I'm getting work.
That is because you are using an app_info (like I am) and that does not bother with the OpenCL check.
____________
/ Reply
Quote
could you please post a working app_info.xml (win7 x64) ?
thx
/ Reply
Quote
"
Actually, it looks like the project is using the BOINC OpenCL detection, which the 6.12 clients don't do. You need the BOINC 7 client for that, which is still being tested. However, why not just try 7.0.14? Links available in the change log thread.
This isn't the last 7 test version, more will follow.
"
and
8/02/2012 23:18:36 | Milkyway@Home | Sending scheduler request: Project initialization.
8/02/2012 23:18:36 | Milkyway@Home | Requesting new tasks for CPU and ATI GPU
8/02/2012 23:18:38 | Milkyway@Home | Scheduler request completed: got 0 new tasks
8/02/2012 23:18:38 | Milkyway@Home | No tasks sent
8/02/2012 23:18:38 | Milkyway@Home | Message from server: Host device lacks OpenCL
8/02/2012 23:18:38 | Milkyway@Home | Message from server: Your NVIDIA GPU lacks the needed compute capability (1.3, required for double precision math
/ Reply
Quote
why not just try 7.0.14? Links available in the change log thread.
I'm running 7.0.14, no work available for me.
I will try to update CCC from 11.9 to 12.1
/ Reply
Quote
indeed 7.0.14 doesn't work either, it solved the opengl problem but he still thinks i'm running nvidia...
and i'm running 11.8 as newer drivers doesn't seem to work but this seems to be a known problem
8/02/2012 23:42:09 | | No config file found - using defaults
8/02/2012 23:42:09 | | Starting BOINC client version 7.0.14 for windows_x86_64
8/02/2012 23:42:09 | | log flags: file_xfer, sched_ops, task
8/02/2012 23:42:09 | | Libraries: libcurl/7.21.6 OpenSSL/1.0.0d zlib/1.2.5
8/02/2012 23:42:09 | | Data directory: C:\ProgramData\BOINC
8/02/2012 23:42:09 | | Running under account Robin
8/02/2012 23:42:09 | | Processor: 2 GenuineIntel Intel(R) Core(TM)2 Duo CPU E7200 @ 2.53GHz [Family 6 Model 23 Stepping 6]
8/02/2012 23:42:09 | | Processor: 3.00 MB cache
8/02/2012 23:42:09 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 lm tm2 pbe
8/02/2012 23:42:09 | | OS: Microsoft Windows 7: Ultimate x64 Edition, Service Pack 1, (06.01.7601.00)
8/02/2012 23:42:09 | | Memory: 4.00 GB physical, 8.00 GB virtual
8/02/2012 23:42:09 | | Disk: 1.82 TB total, 1.39 TB free
8/02/2012 23:42:09 | | Local time is UTC +1 hours
8/02/2012 23:42:09 | | ATI GPU 0: ATI Radeon HD 4700/4800 (RV740/RV770) (CAL version 1.4.1523, 1024MB, 992MB available, 2400 GFLOPS peak)
8/02/2012 23:42:09 | | OpenCL: ATI GPU 0: ATI RV770 (driver version CAL 1.4.1523, device version OpenCL 1.0 AMD-APP (851.4), 512MB, 992MB available)
8/02/2012 23:42:09 | | ATI GPU 0 is OpenCL-capable
8/02/2012 23:42:09 | | Version change (6.12.34 -> 7.0.14)
8/02/2012 23:42:09 | Milkyway@Home | URL http://milkyway.cs.rpi.edu/milkyway/; Computer ID 397774; resource share 100
8/02/2012 23:42:09 | | Reading preferences override file
8/02/2012 23:42:09 | | Preferences:
8/02/2012 23:42:09 | | max memory usage when active: 2047.65MB
8/02/2012 23:42:09 | | max memory usage when idle: 3685.77MB
8/02/2012 23:42:09 | | max disk usage: 100.00GB
8/02/2012 23:42:09 | | (to change preferences, visit the web site of an attached project, or select Preferences in the Manager)
8/02/2012 23:42:09 | | Not using a proxy
8/02/2012 23:42:11 | | Running CPU benchmarks
8/02/2012 23:42:11 | | Suspending computation - CPU benchmarks in progress
8/02/2012 23:42:42 | | FP benchmark ran only 0.046875 sec; ignoring
8/02/2012 23:43:27 | Milkyway@Home | Resetting project
8/02/2012 23:43:27 | Milkyway@Home | Detaching from project
8/02/2012 23:43:55 | | Fetching configuration file from http://milkyway.cs.rpi.edu/milkyway/get_project_config.php
8/02/2012 23:44:13 | Milkyway@Home | Master file download succeeded
8/02/2012 23:44:18 | Milkyway@Home | Sending scheduler request: Project initialization.
8/02/2012 23:44:18 | Milkyway@Home | Requesting new tasks for CPU and ATI
8/02/2012 23:44:20 | Milkyway@Home | Scheduler request completed: got 0 new tasks
8/02/2012 23:44:20 | Milkyway@Home | No tasks sent
8/02/2012 23:44:20 | Milkyway@Home | Bericht van de server:: Your NVIDIA GPU lacks the needed compute capability (1.3, required for double precision math
8/02/2012 23:44:20 | Milkyway@Home | Er zijn taken beschikbaar voor de CPU, maar worden geblokkeerd door uw voorkeursinstellingen
/ Reply
Quote
Maybe we should ask him
http://milkyway.cs.rpi.edu/milkyway/show_user.php?userid=142006
he has some tasks running right now. but they are shown as
MilkyWay@Home
Anonymous platform (ATI GPU)
and he is running the 6.12.34 client.
I just wrote a pm to him. I will tell you, when i get an answer.
/ Reply
Quote
could you please post a working app_info.xml (win7 x64) ?
thx
This should work.
<app_info>
<app>
<name>milkyway</name>
</app>
<file_info>
<name>milkyway_separation_1.00_windows_x86_64__opencl_amd.exe</name>
<executable/>
</file_info>
<app_version>
<app_name>milkyway</app_name>
<version_num>100</version_num>
<flops>1.0e11</flops>
<avg_ncpus>0.05</avg_ncpus>
<max_ncpus>1</max_ncpus>
<plan_class>ati14ati</plan_class>
<coproc>
<type>ATI</type>
<count>1</count>
</coproc>
<cmdline></cmdline>
<file_ref>
<file_name>milkyway_separation_1.00_windows_x86_64__opencl_amd.exe</file_name>
<main_program/>
</file_ref>
</app_version>
</app_info>
____________
/ Reply
Quote
CRUNCHING AGAIN !!!
THX!
/ Reply
Quote
The app_info.xml is not working for me. The exe-file is missing. In my Project Folder.
Milkyway won't download the master file. And when i hit update, i just get "not requesting or reporting new tasks". Upgraded to 7.08 client. Catalyst 12.1 driver
/ Reply
Quote
The 1.00 opencl version seems to be faster than the 0.82 ATI version by about 8 seconds on my HD5830.
____________
/ Reply
Quote
The app_info.xml is not working for me. The exe-file is missing. In my Project Folder.
Milkyway won't download the master file. And when i hit update, i just get "not requesting or reporting new tasks". Upgraded to 7.08 client. Catalyst 12.1 driver
http://www.arkayn.us/forum/index.php?action=downloads;sa=view;down=59
____________
/ Reply
Quote
The app_info.xml is not working for me. The exe-file is missing. In my Project Folder.
Milkyway won't download the master file. And when i hit update, i just get "not requesting or reporting new tasks". Upgraded to 7.08 client. Catalyst 12.1 driver
http://milkyway.cs.rpi.edu/milkyway/download
Edit: Arkayan was faster ..
/ Reply
Quote
Using 7.0.14 with the above posted ap-info file, detached and reattached and still getting no work.
EDIT: Working. In my haste I misnamed the app_info file. *slaps forehead*
/ Reply
Quote
So, early on after the recent changes, I detached from the project thinking that would help. When I reattach, the GPU isn't recognized (by MW@H) so the BOINC manager doesn't download the new app. An app_info file doesn't do me any good then, right? How do I get going again? Is there a place to manually download the app? I have a Win7-64 host with a 5870 running BOINC 6.12.34.
[edit]I see others beat me too it while I was slowly thinking, composing, editing...<blush>[/edit]
____________
/ Reply
Quote
All steps required (win7 x64):
- stop boinc
- place the app_info.xml into your boinc project milkyway directory
- download the app from Arkayn or from mw as posted some msgs before
- put it also into your boinc project milkyway directory
- restart boinc
it took 3 to 4 minutes on my both machines, but both are running now.
one pc with BM 7.0.12 and CCC 12.1
one pc with BM 7.0.14 and CCC 11.9
/ Reply
Quote
I manually downloaded the milkyway_separation_1.00_windows_x86_64__opencl_amd app and put the app_info.xml file suggested below into the project folder, restarted BOINC, and I still get no work...
2/8/2012 6:47:28 PM | Milkyway@Home | Sending scheduler request: To fetch work.
2/8/2012 6:47:28 PM | Milkyway@Home | Requesting new tasks for ATI GPU
2/8/2012 6:47:29 PM | Milkyway@Home | Scheduler request completed: got 0 new tasks
2/8/2012 6:47:29 PM | Milkyway@Home | No tasks sent
Is BOINC 7 required? The latest ATI driver?
____________
/ Reply
Quote
Right now I'm suspecting that people aren't getting work because it seems that clients < 7 don't report having OpenCL and I updated the scheduler to check for the new stuff.
What combination and version of Windows os, catalyst driver and BOINC manager have you tested the new application on?
/ Reply
Quote
last post from me now.
while figuring out the way it should work, i got a real milkyway open cl task. not getting any new from one of those until now.
Working now with the app_info.xml. but boinc downloaded the exe-file after detach and reattach with placed app_info.xml in the folder. it was gone after de-reatach. put back in the .xml file and now it is workin locally.
I am not happy with it, since it still shows the ati cal 1.4 stuff.
I will be patient for now, sleep, go to work and see what happend until then. I still have some rigs working unmodified, but no tasks there until now.
Give it some time. Milkyway@Home is still a Project in the alpha state.
Edit:
got the real open cl app on 7.08 Boinc client.
7.08 and 6.12.34 are working both with app_info.xml
And you may have to enter some numbers in you milkyway preferences on the milkyway hompage
/ Reply
Quote
It's not the application, it's the project schedulerRight now I'm suspecting that people aren't getting work because it seems that clients < 7 don't report having OpenCL and I updated the scheduler to check for the new stuff.
What combination and version of Windows os, catalyst driver and BOINC manager have you tested the new application on?
/ Reply
Quote
Everything errors out immediately on a Mac mini Core 2 Duo running OS X Leopard (10.5.x) since the update to 1.00. Everything was fine with 0.82.
____________
Plus SETI Classic = 21,082 WUs
/ Reply
Quote
I'm using 6.12.34 and I'm getting work.
That is because you are using an app_info (like I am) and that does not bother with the OpenCL check.
I'm using 6.12.33 (Linux), no app_info file, and I'm getting work. The CPU usage was reduced, but I think I will need an app_info file to play with --gpu-polling-mode to reduce it even more.
In my "ultra-low-end" GT 520 I'm getting this message in results:
Estimated Nvidia GPU GFLOP/s: 104 SP GFLOP/s, 13 DP FLOP/s
Warning: Bizarrely low flops (13). Defaulting to 50
Bizarre ;-)
Maybe this 13 GFLOP/s info is correct and the default of 50 is too high, affecting the estimated time of polling period...
Thanks for this new version, Matt!
/ Reply
Quote
OK - working thru the myriad messages here I tried to get this to work on a Win64 with a 4850 and can't get new work.
I copied out the app file posted here. I manually downloaded the exe. I updated to 7.08, and copied the exe file and html file to the MW project folder. Currently, I am getting - no work available. Perhaps that is actually the case at the moment.
I suppose this sort of 'try this, try that' mode is to be accepted when a project reverts to alpha status after a few weeks of running quite well.
Not having the time to fool with it at the moment, nor the inclination to manually tweak a batch of systems as is currently required, I guess I'll wait for the beta version of the update to be available -- one that would at least allow a detatch/reattach of MW to run instead of the collection of interventions now in place.
Some of this same sort of complexity is going on over with POEM as they went to ATI support -- required 7.08 to run for one. The downside over there is that as they had all that development effort, they lacked the resources to stave off the CreditNew monster -- at least we don't have THAT here.
____________
/ Reply
Quote
In my "ultra-low-end" GT 520 I'm getting this message in results:The AMD drivers have an issue that seems to show up every few driver releases where they are detected as 0 or very low Mhz so that is supposed to catch that breaking; I thought the 50 threshold was a bit lower than any DP GPUs which actually exist.
Estimated Nvidia GPU GFLOP/s: 104 SP GFLOP/s, 13 DP FLOP/s
Warning: Bizarrely low flops (13). Defaulting to 50
Bizarre ;-)
Maybe this 13 GFLOP/s info is correct and the default of 50 is too high, affecting the estimated time of polling period...
/ Reply
Quote
I have upgraded my BOINC version to 6.12.34, inserted the app_info.xml file in the milkyway@home project directory as instructed (downloaded from the link kindly supplied by arkayn), and am still not getting any WU's. If there is anything else that I can check or may have done wrong, suggestions would be greatly appreciated.
Also, I'm not entirely sure what the app_info.xml file I just inserted in the milkyway@home project directory actually does. I understand about downloading the application and the fact that older BOINC versions don't necessarily recognise graphics hardware as appropriately OpenCL capable, but what I'm not clear on is what the app_info.xml file does with regard to correcting this. Will the one app_info.xml file be universal to all machines utilising AMD graphics cards that are OpenCL capable? Does my CPU type, clock, core number, or my PC's operating system not require any tweak to the app_info to make this application run properly?
Thanks to everyone for the support on this forum and I would appreciate any reply I can get, because I'm confused and just trying to figure out if I've done something wrong trying to get work for this project. I just want to help! Milkyway is a great project, happy crunching everyone!
/ Reply
Quote
Everything errors out immediately on a Mac mini Core 2 Duo running OS X Leopard (10.5.x) since the update to 1.00. Everything was fine with 0.82.I can't find other cases of that, but I reverted the OS X ones anyway since I guess nothing relevant for it was changed anyway.
/ Reply
Quote
Burned tons of WUs with
Exit status -1073740777 (0xffffffffc0000417) Unknown error number
Stderr output
<core_client_version>6.10.58</core_client_version>
<![CDATA[
<message>
- exit code -1073740777 (0xc0000417)
</message>
<stderr_txt>
<search_application> milkyway_separation 1.00 Windows x86 double OpenCL </search_application>
Unrecognized XML in project preferences: nvidia_block_amount
Skipping: 128
Skipping: /nvidia_block_amount
Error loading Lua script 'astronomy_parameters.txt': [string "number_parameters: 4..."]:1: '<name>' expected near '4'
Error reading astronomy parameters from file 'astronomy_parameters.txt'
Trying old parameters file
Using SSE3 path
Found 1 platform
Platform 0 information:
Name: AMD Accelerated Parallel Processing
Version: OpenCL 1.1 AMD-APP-SDK-v2.4-rc1 (595.9)
Vendor: Advanced Micro Devices, Inc.
Extensions: cl_khr_icd cl_amd_event_callback cl_amd_offline_devices
Profile: FULL_PROFILE
Using device 0 on platform 0
Found 2 CL devices ??? huh ???
Device 'Cypress' (Advanced Micro Devices, Inc.:0x1002) (CL_DEVICE_TYPE_GPU)
Driver version: CAL 1.4.1332
Version: OpenCL 1.1 AMD-APP-SDK-v2.4-rc1 (595.9)
Compute capability: 0.0
Max compute units: 18
Clock frequency: 775 Mhz
Global mem size: 536870912
Local mem size: 32768
Max const buf size: 65536
Double extension: cl_amd_fp64
</stderr_txt>
]]>
WInXP 32bit, AMD 5830
Back to v0.82 for the moment
/ Reply
Quote
Found 2 CL devices ??? huh ???Your CPU counts a 2nd device on AMD's OpenCL.
WInXP 32bit, AMD 5830Using Catalyst 11.3 it looks like. Try a newer driver. I am sort of aware of a few various crashes with different Catalyst versions on different GPUs. I think 11.3-4 and 11.10-11.12 are the most problematic versions; 12.1 seems to be working most consistently on everything. Around 11.3 the binary format stuff was introduced (which it says it supports) but it's probably crasher around then.
Back to v0.82 for the moment
/ Reply
Quote
FWIW, OSX 10.6.8 Boinc 6.12.35 MWAH 1.00...
Received 7 1.00 WUs, processing normally (4 completed as of now). MWAH will not send any work either automatically or via manual update. Server status indicates everything functioning normally...
Any clues as to why no work issuance?
:)
/ Reply
Quote
Is there a command that can be added to the app-info file that will throttle back the GPU usage. It's now running @ 99% and I really don't want to run it that hard.
Thanks.
____________
/ Reply
Quote
Is there a command that can be added to the app-info file that will throttle back the GPU usage. It's now running @ 99% and I really don't want to run it that hard.You can raise the target frequency with the --gpu-target-frequency argument, or you can do it in the project preferences now which would be less annoying.
Thanks.
/ Reply
Quote
FWIW, OSX 10.6.8 Boinc 6.12.35 MWAH 1.00...
Received 7 1.00 WUs, processing normally (4 completed as of now). MWAH will not send any work either automatically or via manual update. Server status indicates everything functioning normally...
Any clues as to why no work issuance?
:)
And, now all WUs complete, task queue empty & computer idle, log says, "Project has no tasks available", server status as of 2:18:32 UTC, "Tasks ready to send 1,980" and "MilkyWay@Home unsent 1,759 MilkyWay@Home N-Body Simulation unsent 221"
???
/ Reply
Quote
Matt
I have no idea if this is symptomatic of the problem ... however its been stated above that its probably connected with recognising OpenCL capability on some cards. The lines below seem strange as I am sure before it also claimed to see GPU1 as OpenCL capable, whereas now it only claims one is. That ties in with a download I did get at one point - just two WUs on GPU0
09/02/2012 02:32:27 | | ATI GPU 0: AMD Radeon HD 79x0 series (Tahiti) (CAL version 1.4.1658, 3072MB, 3033MB available, 12493 GFLOPS peak)
09/02/2012 02:32:27 | | ATI GPU 1: AMD Radeon HD 79x0 series (Tahiti) (CAL version 1.4.1658, 3072MB, 3033MB available, 12493 GFLOPS peak)
09/02/2012 02:32:27 | | OpenCL: ATI GPU 0: Tahiti (driver version CAL 1.4.1658 (VM), device version OpenCL 1.1 AMD-APP (844.5), 3072MB, 3033MB available)
09/02/2012 02:32:27 | | OpenCL: ATI GPU 1: Tahiti (driver version CAL 1.4.1658 (VM), device version OpenCL 1.1 AMD-APP (844.5), 3072MB, 3033MB available)
09/02/2012 02:32:27 | | ATI GPU 0 is OpenCL-capable
Apologies if old news ..... am aware you will be beavering away.
Regards
Zy
/ Reply
Quote
MattThat stuff isn't in my department. I've been seeing that sort of too, but that's the BOINC client's problem.
I have no idea if this is symptomatic of the problem ... however its been stated above that its probably connected with recognising OpenCL capability on some cards. The lines below seem strange as I am sure before it also claimed to see GPU1 as OpenCL capable, whereas now it only claims one is. That ties in with a download I did get at one point - just two WUs on GPU0
09/02/2012 02:32:27 | | ATI GPU 0: AMD Radeon HD 79x0 series (Tahiti) (CAL version 1.4.1658, 3072MB, 3033MB available, 12493 GFLOPS peak)
09/02/2012 02:32:27 | | ATI GPU 1: AMD Radeon HD 79x0 series (Tahiti) (CAL version 1.4.1658, 3072MB, 3033MB available, 12493 GFLOPS peak)
09/02/2012 02:32:27 | | OpenCL: ATI GPU 0: Tahiti (driver version CAL 1.4.1658 (VM), device version OpenCL 1.1 AMD-APP (844.5), 3072MB, 3033MB available)
09/02/2012 02:32:27 | | OpenCL: ATI GPU 1: Tahiti (driver version CAL 1.4.1658 (VM), device version OpenCL 1.1 AMD-APP (844.5), 3072MB, 3033MB available)
09/02/2012 02:32:27 | | ATI GPU 0 is OpenCL-capable
Apologies if old news ..... am aware you will be beavering away.
Regards
Zy
/ Reply
Quote
Oh derp. Since last time I did this I updated to Lion and I forgot the 10.5 SDK was removed from the new stuff so it turns out I was building against the 10.6 SDK. I rebuilt it as 1.01 for the OS X ones.Everything errors out immediately on a Mac mini Core 2 Duo running OS X Leopard (10.5.x) since the update to 1.00. Everything was fine with 0.82.I can't find other cases of that, but I reverted the OS X ones anyway since I guess nothing relevant for it was changed anyway.
/ Reply
Quote
Hi
I attached Milkyway@home and Milky won't any files. I did check this from boinc programdata milkyway folder. So no work,no exes no nothing.
My win7 64 bit+amd 6950 and 5830 were crunching everything ok before this update to 1.00.
I have Boinc 6.12.34 x64 version and Boinc 7.xx versions won't even start running(no boincmanager window would start even fresh windows 7 instal or deleting files manually after unistal) so it won't solution for this milkyway issue for me atleast.
Any ideas?
____________
My Boinc Blog
/ Reply
Quote
Thanks for the quick action. I will have to wait until work is available to see if it works. My Mac mini Core Duo running Tiger (10.4.x) is still working on several units with the 0.82 version. I hope it continues to work when it gets the 1.01.
____________
Plus SETI Classic = 21,082 WUs
/ Reply
Quote
Anyone actually manage to get the Linux x86_64 opencl app to work? I have an HD 4870 and it looks like I've been upgraded out of the ability to crunch. The 4870 doesn't like the new AMD drivers and it doesn't appear that the new app likes the old one.
This couldn't have waited til after the double credit day?
/ Reply
Quote
Anyone actually manage to get the Linux x86_64 opencl app to work? I have an HD 4870 and it looks like I've been upgraded out of the ability to crunch. The 4870 doesn't like the new AMD drivers and it doesn't appear that the new app likes the old one.What problem are you having? The 4870 drivers were broken with the OpenCL stuff from 11.8 to 11.12 or so. 12.1 works.
This couldn't have waited til after the double credit day?
/ Reply
Quote
What problem are you having? The 4870 drivers were broken with the OpenCL stuff from 11.8 to 11.12 or so. 12.1 works.
I'm currently at 11.8 because anything (stable) after that broke other OpenCL apps. I see that 12.1 is in Gentoo Portage, so I'll try that. Thanks for the info.
/ Reply
Quote
It's not the application, it's the project schedulerRight now I'm suspecting that people aren't getting work because it seems that clients < 7 don't report having OpenCL and I updated the scheduler to check for the new stuff.
What combination and version of Windows os, catalyst driver and BOINC manager have you tested the new application on?
Hmm... ok. What combination and version of Windows os, catalyst driver and BOINC manager have you tested the separation applications 1.00 against? I want to know caus I'm not getting any WUs. I gues that you have tested against some combination of OS & drivers or are you just make an update and hope for the best? ;)
/ Reply
Quote
It's not the application, it's the project schedulerRight now I'm suspecting that people aren't getting work because it seems that clients < 7 don't report having OpenCL and I updated the scheduler to check for the new stuff.
What combination and version of Windows os, catalyst driver and BOINC manager have you tested the new application on?
Hmm... ok. What combination and version of Windows os, catalyst driver and BOINC manager have you tested the separation applications to 1.00 against? I want to know caus I'm not getting any WUs. I gues that you have tested against some combination of OS & drivers or are you just make an update and hope for the best? ;)
Right now, there is no work available. I am doing Collatz.
____________
/ Reply
Quote
I'm getting "Server Error recompile needed", the first time was a couple of mins ago.
/ Reply
Quote
Seeing it as well, most likely we will have to wait until the morning as it is 1AM in New York.
____________
/ Reply
Quote
New ati_openCL project startet:
donate@home
http://donateathome.org
I found that a couple of minutes ago, they do medical science.
There is also albert@home, the Einstein beta test app, they do have fine running ati openCL apps.
http://albert.phys.uwm.edu/
For all the guys who do not want to search just new numbers.
/ Reply
Quote
This whole time I have gotten GPU work on occasion, but it's like 3 tasks at a time and then it stops.
/ Reply
Quote
Oh thats the ground my 3850 AGP get no work anymore :/ *sign*
____________
DSKAG Austria Research Team: http://www.research.dskag.at
/ Reply
Quote
New ati_openCL project startet:
donate@home
http://donateathome.org
I found that a couple of minutes ago, they do medical science.
There is also albert@home, the Einstein beta test app, they do have fine running ati openCL apps.
http://albert.phys.uwm.edu/
For all the guys who do not want to search just new numbers.
Sorry, my mistake, donate@home is crunching for bitcoin ...
/ Reply
Quote
I'm getting "Server Error recompile needed", the first time was a couple of mins ago.
Getting the exact same thing here.
/ Reply
Quote
Are you still getting that? That should only have been happening for a few minutes a while ago.
/ Reply
Quote
Its ok now Matt - and supply is steady. V1 appears to be running ok on mine. I had 3 computation errors early on, but nothing since.
So far so good :)
Regards
Zy
/ Reply
Quote
Whats the cheapest lowest card i can get with pcie working for MW (must not be a new one)? Will try to get one over ebay or so then.
____________
DSKAG Austria Research Team: http://www.research.dskag.at
/ Reply
Quote
Watch out for the needs of OpenCL - you will need to check 4XXX will take an OpenCL app, I'm fairly certain they will, but I dont know for certain if all the 4XXX variants do or not - and you need to ensure what you use is OpenCL capable via its driver set.
CAL is now dead, and apps will be OpenCL as they get redeveloped, all Projects, not just here.
Regards
Zy
/ Reply
Quote
here are the current errors i'm getting:
Stderr output
<core_client_version>6.12.34</core_client_version>
<![CDATA[
<message>
Incorrect function. (0x1) - exit code 1 (0x1)
</message>
<stderr_txt>
<search_application> milkyway_separation 1.00 Windows x86 double OpenCL </search_application>
Unrecognized XML in project preferences: max_gfx_cpu_pct
Skipping: 0
Skipping: /max_gfx_cpu_pct
Unrecognized XML in project preferences: apps_selected
Skipping: app_id
Skipping: /apps_selected
Error loading Lua script 'astronomy_parameters.txt': [string "number_parameters: 4..."]:1: '<name>' expected near '4'
Error reading astronomy parameters from file 'astronomy_parameters.txt'
Trying old parameters file
Using SSE3 path
Found 1 platform
Platform 0 information:
Name: ATI Stream
Version: OpenCL 1.1 ATI-Stream-v2.3 (451)
Vendor: Advanced Micro Devices, Inc.
Extensions: cl_khr_icd cl_amd_event_callback cl_amd_offline_devices
Profile: FULL_PROFILE
Using device 1 on platform 0
Found 2 CL devices
Device 'AMD Phenom(tm) II X6 1090T Processor' (AuthenticAMD:0x1002) (CL_DEVICE_TYPE_CPU)
Driver version: 2.0
Version: OpenCL 1.1 ATI-Stream-v2.3 (451)
Compute capability: 0.0
Max compute units: 6
Clock frequency: 3712 Mhz
Global mem size: 1073741824
Local mem size: 32768
Max const buf size: 65536
Double extension: cl_amd_fp64
Build log:
--------------------------------------------------------------------------------
C:\DOCUME~1\Eric\LOCALS~1\Temp\OCL3B.tmp.cl(201): error: invalid unroll factor
#pragma unroll NSTREAM
^
C:\DOCUME~1\Eric\LOCALS~1\Temp\OCL3B.tmp.cl(243): error: invalid unroll factor
#pragma unroll NSTREAM
^
C:\DOCUME~1\Eric\LOCALS~1\Temp\OCL3B.tmp.cl(272): error: invalid unroll factor
#pragma unroll NSTREAM
^
C:\DOCUME~1\Eric\LOCALS~1\Temp\OCL3B.tmp.cl(279): error: invalid unroll factor
#pragma unroll NSTREAM
^
C:\DOCUME~1\Eric\LOCALS~1\Temp\OCL3B.tmp.cl(287): error: invalid unroll factor
#pragma unroll NSTREAM
^
5 errors detected in the compilation of "C:\DOCUME~1\Eric\LOCALS~1\Temp\OCL3B.tmp.cl".
--------------------------------------------------------------------------------
clBuildProgram: Build failure (-11): CL_BUILD_PROGRAM_FAILURE
Error building program from source (-11): CL_BUILD_PROGRAM_FAILURE
Error creating integral program from source
Failed to calculate likelihood
<background_integral> 1.#QNAN0000000000 </background_integral>
<stream_integral> 1.#QNAN0000000000 1.#QNAN0000000000 </stream_integral>
<background_likelihood> 1.#QNAN0000000000 </background_likelihood>
<stream_only_likelihood> 1.#QNAN0000000000 1.#QNAN0000000000 </stream_only_likelihood>
<search_likelihood> 1.#QNAN0000000000 </search_likelihood>
03:08:21 (564): called boinc_finish
</stderr_txt>
]]>
platform: WinXP 32-bit, BOINC v6.12.34, AMD HD 5870
____________
/ Reply
Quote
Double check your GPU clock setting, because of this inside the stderr ....
Clock frequency: 3712 Mhz
I cant believe its that ..... but .... if it is, the card has not long left in this world..... :)
Must be a symptom of your error and falsely reported to Stderr .... but if its not .... gulp ... that needs changing fast.
Regards
Zy
/ Reply
Quote
here are the current errors i'm getting:First it looks like your Catalyst drivers are really old. The minimum I have the scheduler requiring is ~11.4. Second it's actually trying to run it on the CPU. You're using app_info, are you giving it a --device argument?
/ Reply
Quote
Watch out for the needs of OpenCL - you will need to check 4XXX will take an OpenCL app, I'm fairly certain they will, but I dont know for certain if all the 4XXX variants do or not - and you need to ensure what you use is OpenCL capable via its driver set.R700 stuff (4770 and 4850, 4870) support OpenCL but just barely. There are various limitations and they're missing features nbody uses.
CAL is now dead, and apps will be OpenCL as they get redeveloped, all Projects, not just here.
Regards
Zy
/ Reply
Quote
New version runs fine for me! Run-times about 8 s faster than 0.82 separation! Thanks Matt!
System
Win 7 64 Bit
6.12.34
2x 5850
app_info with two WU per GPU
____________
/ Reply
Quote
I've gotten it to send out ATI work again by temporarily undeprecating the old 0.82 CAL stuff. This also seems to have had the side effect of it more frequently sending out the OpenCL 1.0 AMD ones. I seem to be getting a majority of tasks as the old one, but I guess that's good enough to go to sleep for now.
/ Reply
Quote
Now i detach & retach the project.i get the old 0.82ati.
It runs dry and no scheduler request.
What can help?
Greetings.
/ Reply
Quote
Now i detach & retach the project.i get the old 0.82ati.
It runs dry and no scheduler request.
What can help?
Greetings.
Same problems for me.
I detach and retach the project repeatedly without any success.
One workunit and after nothing.
I have to detach and retach the project to have a new one workunit.
I have a 5870 ATI GPU on Gentoo Linux 64 bits system with a 12.1 Catalyst driver and a BOINC client version 7.0.14 .
____________
/ Reply
Quote
R700 stuff (4770 and 4850, 4870) support OpenCL but just barely. There are various limitations and they're missing features nbody uses.
So this means 4XXX users can't upgrade to the new app and are stuck with 0.82 and Cat 11.10? (since CAL is broken in later versions)
/ Reply
Quote
Thanks ever so much for needlessly eliminating support for an older ATI GPU that was producing credible work results every day. As that was the primary GPU we've had available, I guess we'll be turning to other projects whose administrators aren't as short-sighted.
/ Reply
Quote
I'm using Nvidia but CPU usage is still quite high.
This is my app_info.xml
<app>
<name>milkyway</name>
</app>
<file_info>
<name>milkyway_separation_1.00_windows_x86_64__opencl_nvidia.exe</name>
<executable />
</file_info>
<app_version>
<app_name>milkyway</app_name>
<version_num>100</version_num>
<flops>1.0e11</flops>
<avg_ncpus>0.05</avg_ncpus>
<max_ncpus>1</max_ncpus>
<plan_class>cuda</plan_class>
<coproc>
<type>CUDA</type>
<count>.5</count>
</coproc>
<cmdline></cmdline>
<file_ref>
<file_name>milkyway_separation_1.00_windows_x86_64__opencl_nvidia.exe</file_name>
<main_program/>
</file_ref>
</app_version>
</app_info>
Where do I put --gpu-wait-factor and --gpu-polling-mode, and how to lower CPU usage even more?
edit: And do I need to change the <type> tag?
Anthony.
____________
The longer I live, the more reasons I develop for wanting to die.
/ Reply
Quote
However this also means the Radeon 38xx cards aren't supported with new stuff.
Does that mean, that it's time to find a new project for my HD3850, or is it still OK to run the 0.82 with app_info file? All results validate so far...
____________
.
/ Reply
Quote
So now,i change the driver from 11.9 to 12.1 CCC.
I get work now 0.82ATI and the scheduler request
runs.That looks good.:-))
Greetings.
/ Reply
Quote
Got my 7970 up and running again with 34s (15s faster than 0.82) WU's. :)
Boinc v6.12.34(x64)
GFX driver: 11.12 (8.921.2)
(Thanks arkayn)
http://www.arkayn.us/forum/index.php?action=downloads;sa=view;down=59
____________
/ Reply
Quote
CAL 12.1, BOINC 6.10.60, 2x HD 6950, no app_info
I'm getting tons of work (mixed type, both v1.00 & v0.82). Everything is ok,
although the 1.00 wus are a little bit slower, say ~72sec instead of ~68s), is it normal?
/ Reply
Quote
yay! Work again for the old 3850 AGP :D
How long you plan to run temporary 0.82 reactivation? ^^
____________
DSKAG Austria Research Team: http://www.research.dskag.at
/ Reply
Quote
FWIW, OSX 10.6.8 Boinc 6.12.35 MWAH 1.00...
Received 7 1.00 WUs, processing normally (4 completed as of now). MWAH will not send any work either automatically or via manual update. Server status indicates everything functioning normally...
Any clues as to why no work issuance?
:)
And, now all WUs complete, task queue empty & computer idle, log says, "Project has no tasks available", server status as of 2:18:32 UTC, "Tasks ready to send 1,980" and "MilkyWay@Home unsent 1,759 MilkyWay@Home N-Body Simulation unsent 221"
???
A.M. Update:
1.01's in the queue and processing normally as CPU WUs, also, return of NBody 0.84's to queue. Thanks! Looks like the Valentine's Day 2-for-1 sale will help make up for last night's loss.
Still no love for the 2GB HD6970M or Sandy Bridge i7 3,4GHz 2600k HD Graphics 3000 for GPU processing?
Best, :)
/ Reply
Quote
What problem are you having? The 4870 drivers were broken with the OpenCL stuff from 11.8 to 11.12 or so. 12.1 works.
I'm currently at 11.8 because anything (stable) after that broke other OpenCL apps. I see that 12.1 is in Gentoo Portage, so I'll try that. Thanks for the info.
I installed 12.1 and now Sep. 1.00 is working on my 4870 with Gentoo x86_64. Ironically, now the old app is failing.
My task times went from 2:38 to 2:28.
/ Reply
Quote
Version 1.01 appears to be running properly on Mac mini Core 2 Duo under Leopard and also Mac mini Core Duo under Tiger. Won't know for certain until something finishes, goes in, and validates but it looks promising. I'll report back when I know more. Thanks again - you take amazingly good care of your volunteers.
____________
Plus SETI Classic = 21,082 WUs
/ Reply
Quote
Is it the app or the boinc client causing my starting "Time Remaining" to stay at 3 hours?
/ Reply
Quote
First it looks like your Catalyst drivers are really old. The minimum I have the scheduler requiring is ~11.4. Second it's actually trying to run it on the CPU. You're using app_info, are you giving it a --device argument?
actually Matt, my first separation v1.00 task ran without an app_info.xml file. i was using an app_info.xml file when i was running v0.82, but only to allow 2 tasks to run simultaneously, and it contained no --device argument in it b/c i didn't need such an argument to prevent tasks from running on anything other than my HD 5870 GPU. but that app_info.xml file got deleted along with everything else in the MW@H project folder when i detached from the project (i forgot to save a copy of it). when i reattached, i allowed that first v1.00 task to run without an app_info.xml. when i saw it error out right away, i set "no new tasks," copied the app_info.xml posted by Arkayn earlier in this thread, and changed all instances of "milkyway_separation_1.00_windows_x86_64__opencl_amd.exe" to "milkyway_separation_1.00_windows_intelx86__opencl_amd.exe" to match the executable in my project folder. i left count at 1, as its best to run 1 task at a time until i've got the problems worked out. so that's why it appears that most of my v1.00 errors were on an anonymous platform.
at any rate, i'd like to try a --device argument by itself first to see if that does anything for me. i know the GPU drivers are old, but 11.2 has worked extremely well for me, so i'd like to use a driver update as a second option...that is, unless you're confident that the drivers are part of my problem. if not, what --device argument must i include in the app_info.xml file? and where exactly do i insert it into the file?
TIA,
Eric
____________
/ Reply
Quote
yay! Work again for the old 3850 AGP :DI patched the scheduler so only R600/Radeon 38xx stuff will get the 0.82 for now.
How long you plan to run temporary 0.82 reactivation? ^^
/ Reply
Quote
Hmm... ok. What combination and version of Windows os, catalyst driver and BOINC manager have you tested the separation applications 1.00 against? I want to know caus I'm not getting any WUs. I gues that you have tested against some combination of OS & drivers or are you just make an update and hope for the best? ;)On Nvidia it should require somewhere around 270.xx, so I have the scheduler require 280.xx+ for that. On AMD, Catalyst between 11.4 and 11.9 should work, 11.9-11.12 was broken for both the CAL version and OpenCL on R770, 12.1 works for the OpenCL one on everything I tried as well as the special 7900 driver.
/ Reply
Quote
Thanks ever so much for needlessly eliminating support for an older ATI GPU that was producing credible work results every day. As that was the primary GPU we've had available, I guess we'll be turning to other projects whose administrators aren't as short-sighted.There isn't really anything I can do when AMD doesn't support it. It doesn't support OpenCL at all, and CAL was deprecated a few releases ago, and will not be supported by the update that should be coming next month.
/ Reply
Quote
I will leave the 0.82 ATI apps available on my site as well for those people who need it for older cards.
____________
/ Reply
Quote
yay! Work again for the old 3850 AGP :DI patched the scheduler so only R600/Radeon 38xx stuff will get the 0.82 for now.
How long you plan to run temporary 0.82 reactivation? ^^
Thx :) For how long? i read about the update coming next month. Or do you ment ATI Drivers with that?
Cos i must calculate then for a new cheap little BOINC Machine with PCIe slot very soon. (or not?) ^^
____________
DSKAG Austria Research Team: http://www.research.dskag.at
/ Reply
Quote
at any rate, i'd like to try a --device argument by itself first to see if that does anything for me. i know the GPU drivers are old, but 11.2 has worked extremely well for me, so i'd like to use a driver update as a second option...that is, unless you're confident that the drivers are part of my problem. if not, what --device argument must i include in the app_info.xml file? and where exactly do i insert it into the file?
TIA,
Eric
The first thing is this error:
C:\DOCUME~1\Eric\LOCALS~1\Temp\OCL45.tmp.cl(201): error: invalid unroll factor
I remember this was a bug in the first few releases after they added loop unroll; it didn't work with preprocessor defined things. It was fixed in 11.3 or 11.4.
The second thing is it's trying to use the CPU (device 1, the last device given is the CPU on AMD's). That works, but it isn't really what you're trying to do.
The device selection uses a --device argument first if given, otherwise it uses the device index from the BOINC app_init_data which happens on new enough clients (> 6.13.1), otherwise it defaults to using device 0. It's getting a 1 from one of those places and since you're using app_info I figure it's coming from there.
/ Reply
Quote
Thx :) For how long? i read about the update coming next month. Or do you ment ATI Drivers with that?Not really sure. Right now workflow seems working as expected; it seems the scheduler mostly stops both types of ATI work unless I un-deprecate those so I'll have to figure that out.
Cos i must calculate then for a new cheap little BOINC Machine with PCIe slot very soon. (or not?) ^^
I would like to get people off of those somewhat quickly since there's stuff to hopefully reduce the number of incidents where ATI results doesn't validate against CPU results.
The last SDK release notes said "CAL will no longer be supported as of version 2.7" which is expected in March and coincides with the Catalyst release.
/ Reply
Quote
Error while computing: http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=99460786
I have Intel OpenCL runtime installed. It seems (->log) that exec selects wrong platform.
/ Reply
Quote
The first thing is this error:
C:\DOCUME~1\Eric\LOCALS~1\Temp\OCL45.tmp.cl(201): error: invalid unroll factor
I remember this was a bug in the first few releases after they added loop unroll; it didn't work with preprocessor defined things. It was fixed in 11.3 or 11.4.
hmm...well then it does appear that the current driver i'm using is at least part of the problem, so i'll look into updating it when i get home.
The second thing is it's trying to use the CPU (device 1, the last device given is the CPU on AMD's). That works, but it isn't really what you're trying to do.
The device selection uses a --device argument first if given, otherwise it uses the device index from the BOINC app_init_data which happens on new enough clients (> 6.13.1), otherwise it defaults to using device 0. It's getting a 1 from one of those places and since you're using app_info I figure it's coming from there.
here's my app_info.xml - as far as i can tell, there isn't a line of code specifically forcing tasks to run on device 1 (the CPU):
<app_info>
<app>
<name>milkyway</name>
</app>
<file_info>
<name>milkyway_separation_1.00_windows_intelx86__opencl_amd.exe</name>
<executable/>
</file_info>
<app_version>
<app_name>milkyway</app_name>
<version_num>100</version_num>
<flops>1.0e11</flops>
<avg_ncpus>0.05</avg_ncpus>
<max_ncpus>1</max_ncpus>
<plan_class>ati14ati</plan_class>
<coproc>
<type>ATI</type>
<count>1</count>
</coproc>
<cmdline></cmdline>
<file_ref>
<file_name>milkyway_separation_1.00_windows_intelx86__opencl_amd.exe</file_name>
<main_program/>
</file_ref>
</app_version>
</app_info>
...so despite the errors being caused by an out-of-date driver, i still don't understand why tasks are trying to run on device 1.
____________
/ Reply
Quote
Error while computing: http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=99460786What happens with a newer (You're using 7.0.8, the current 7 one is 7.0.14) or older BOINC (6.12)?
I have Intel OpenCL runtime installed. It seems (->log) that exec selects wrong platform.
/ Reply
Quote
Hello,
all new tasks got calculation errors.
Stderr Ausgabe
<core_client_version>6.12.33</core_client_version>
<![CDATA[
<message>
process exited with code 127 (0x7f, -129)
</message>
<stderr_txt>
../../projects/milkyway.cs.rpi.edu_milkyway/milkyway_separation_1.00_x86_64-pc-linux-gnu__opencl_amd: error while loading shared libraries: libOpenCL.so.1: cannot open shared object file: No such file or directory
</stderr_txt>
]]>
I got only tasks with V1.00 no V1.01.
Do I need new drivers?
My machine:
Do 09 Feb 2012 17:26:23 CET | | Starting BOINC client version 6.12.33 for x86_64-pc-linux-gnu
Do 09 Feb 2012 17:26:23 CET | | Libraries: libcurl/7.19.7 OpenSSL/0.9.8k zlib/1.2.3.3 libidn/1.15
Do 09 Feb 2012 17:26:23 CET | | Processor: 4 AuthenticAMD AMD Athlon(tm) II X4 605e Processor [Family 16 Model 5 Stepping 2]
Do 09 Feb 2012 17:26:23 CET | | Processor: 512.00 KB cache
Do 09 Feb 2012 17:26:23 CET | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm 3dnowext 3dnow constant_tsc rep_good nonstop_tsc extd_apicid pni monitor cx16 popcnt lahf
Do 09 Feb 2012 17:26:23 CET | | OS: Linux: 2.6.32-38-generic
Do 09 Feb 2012 17:26:23 CET | | Memory: 3.87 GB physical, 9.90 GB virtual
Do 09 Feb 2012 17:26:23 CET | | Disk: 283.67 GB total, 247.76 GB free
Do 09 Feb 2012 17:26:23 CET | | Local time is UTC +1 hours
Do 09 Feb 2012 17:26:23 CET | | ATI GPU 0: ATI Radeon HD 5800 series (Cypress) (CAL version 1.4.1417, 1024MB, 1792 GFLOPS peak)
Thanks for any help
Lennart
/ Reply
Quote
I got only tasks with V1.00 no V1.01.1.01 is just the statically linked 1.00 CPU version. Your Catalyst driver is new enough that it should include libOpenCL. Is it actually not present on your system or in some weird place? What distro and how did you install the drivers?
Do I need new drivers?
/ Reply
Quote
As a point of information my HD 4770 (11.6, BOINC 6.12.34, winXP32, AMD64 3800+) has been very unhappy with the change to OpenCL. Host 179083 ( http://milkyway.cs.rpi.edu/milkyway/show_host_detail.php?hostid=179083 ) gets OpenCL work, but promptly errors out with -1073740777 (unknown error). I detached and tried again with the same result.
I'm reluctant to install 12.1 as that may break Moo. It's my understanding that CAL support has been removed in this release. I'm open to suggestions.
Rick
/ Reply
Quote
Version 1.01 is running properly on Mac mini Core 2 Duo under Leopard and also on Mac mini Core Duo under Tiger. Units have finished, reported, and validated for both systems. Thanks again.
____________
Plus SETI Classic = 21,082 WUs
/ Reply
Quote
Which are the best drivers to use? I have tried 12.1 and 11.9 and keep getting BSODs with both. I had 10.5 before I think but that was too old and wouldn't work with the new app. I have a 5970 on Win7 64
/ Reply
Quote
Rick, from what I've seen, 12.1 doesn't break Moo, but it DOES break Collatz.
It does seem that as the various projects engage in a 'my update is better than yours' war, that it has become more difficult to work with multiple GPU projects. Sort of an Anti-BOINC philosophy. All in all, I think the jury is in on the alpha product Separation 1.00. We had a good run after the hardware upgrades were deployed at the beginning of the month. But now it seems that MW has found a new way to drop down on my GPU project list.
This workstation is running 7.08 (beta, but both MW and POEM seem to compel that change), and 12.1 on an XP workstation along with a 4850. No joy from MW -- either they don't have work, or the work doled out is in hold due to the high number of computation errors kicked up by folks attempting to get 1.0 to work with various hardware and software versions. I generated a ton of comp errors with an earlier version of BOINC and an earlier version of the ATI driver myself.
So, for now, I'll stay with older versions of ATI since I hate punishing the one long term reliable ATI GPU project out there (Collatz) due to trouble generated by other projects.
Edit,I reverted to ATI 11.10 - solved the Collatz problem - not getting work from MW right now, but it seems to me that is more the project side problem then end users there. I figure that MW folks will eventually figure out an application combination that 'regular people' can work with. I think the current 'try this, try that' *requirement* and report back, given the variables that are out there, is simply too onerous for those with other lives to lead.
For me, my GPU hardware is pretty much HD 4850 -- good price/performance and power requirement point for me. My OS environment is XP, Win7 32, and Win 7 64. The ATI drivers range from 11.4 to 11.10. The BM versions range from 6.10.58 to 6.12.34 to 7.0.08.
The 7.10.08 turned out to be a requirement for POEM ATI GPU support. That is one of the 'downsides' there (forcing a beta manager), though 7.0.08 isn't too bad. The other downside over in POEM land is that they went with CreditNew -- which results in strange (and typically low credit) results with limited standardization in terms of actual calculations performed. But that's a rant for another venue.
For now, it looks like MW will go in stasis for me -- attached but not able to get work.
As a point of information my HD 4770 (11.6, BOINC 6.12.34, winXP32, AMD64 3800+) has been very unhappy with the change to OpenCL. Host 179083 ( http://milkyway.cs.rpi.edu/milkyway/show_host_detail.php?hostid=179083 ) gets OpenCL work, but promptly errors out with -1073740777 (unknown error). I detached and tried again with the same result.
I'm reluctant to install 12.1 as that may break Moo. It's my understanding that CAL support has been removed in this release. I'm open to suggestions.
Rick
/ Reply
Quote
For my system the following is working fine:
one pc with BM 7.0.12 and CCC 12.1, AMD A8 + HD5830
one pc with BM 7.0.14 and CCC 11.9 Intel i7 + HD5870 + HD6950
all running win7 x64
/ Reply
Quote
Rick, from what I've seen, 12.1 doesn't break Moo, but it DOES break Collatz.
No, it does not.
This is my A8 system, running win7 x64, BM 7.0.12 and CCC 12.1
GPU 0 is the HD5830
GPU 1 is on-chip HD6550
/ Reply
Quote
Still can't get any WUs for my XP 32 bit box if I use the app_info I was using before to reduce the GPU usage. Worked fine before. Anyone have a suggestion. File used is below.
<app_info>
<app><!-- GPU app v0.82 for MW Separation -->
<name>milkyway</name>
</app>
<file_info>
<name>milkyway_separation_0.82_windows_intelx86__ati14.exe</name>
<executable/>
</file_info>
<app_version>
<app_name>milkyway</app_name>
<version_num>82</version_num>
<plan_class>ati14</plan_class>
<flops>1.0e11</flops>
<avg_ncpus>0.05</avg_ncpus>
<max_ncpus>1</max_ncpus>
<coproc>
<type>ATI</type>
<count>1.0</count>
</coproc>
<cmdline>--gpu-target-frequency 80 --gpu-polling-mode 16 --process-priority 1</cmdline>
<file_ref>
<file_name>milkyway_separation_0.82_windows_intelx86__ati14.exe</file_name>
<main_program/>
</file_ref>
</app_version>
</app_info>
/ Reply
Quote
2/9/2012 10:57:53 PM | Milkyway@Home | Sending scheduler request: To fetch work.
2/9/2012 10:57:53 PM | Milkyway@Home | Requesting new tasks for ATI
2/9/2012 10:57:55 PM | Milkyway@Home | Scheduler request completed: got 0 new tasks
2/9/2012 10:57:55 PM | Milkyway@Home | No tasks sent
2/9/2012 10:57:55 PM | Milkyway@Home | Message from server: GPU lacks necessary double precision extension
which extension are you looking for?
My GPU (cayman) still does not have cl_khr_fp64 (promised by amd since launch, still not delivered to VLIW4). We're stuck on cl_amd_fp64. Can you please accept it in your app or it's not valid for you?
Thanks a lot!
/ Reply
Quote
which extension are you looking for?It looks for cl_khr_fp64 or cl_amd_fp64, if you have a client new enough to report OpenCL information back. It looks like some of the 7.0.x aren't sending back the extensions properly.
My GPU (cayman) still does not have cl_khr_fp64 (promised by amd since launch, still not delivered to VLIW4). We're stuck on cl_amd_fp64. Can you please accept it in your app or it's not valid for you?
Thanks a lot!
/ Reply
Quote
Started to get the several errors today on 3 different systems with GPUs.
Suggestions?
-------------------------------------
http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=99668860
Coprocessors AMD ATI Radeon HD 5800 series (Cypress) (2048MB) driver: Operating System Microsoft Windows XP
Using SSE3 path
Error getting number of platform (-1001): CL_PLATFORM_NOT_FOUND_KHR
Failed to get information about device
Error getting device and context (1): MW_CL_ERROR
Failed to calculate likelihood
-------------------------------------
http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=99641528
Coprocessors AMD AMD Radeon HD 6900 series (Cayman) (2048MB) driver: 1.4.1385
Operating System Microsoft Windows XP
Unrecognized XML in project preferences: max_gfx_cpu_pct
Skipping: 90
Skipping: /max_gfx_cpu_pct
Error loading Lua script 'astronomy_parameters.txt': [string "number_parameters: 4..."]:1: '<name>' expected near '4'
Error reading astronomy parameters from file 'astronomy_parameters.txt'
Trying old parameters file
Using SSE3 path
Error getting number of platform (-1001): CL_PLATFORM_NOT_FOUND_KHR
Failed to get information about device
Error getting device and context (1): MW_CL_ERROR
Failed to calculate likelihood
-------------------------------------
http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=99668972
Coprocessors AMD AMD Radeon HD 6900 series (Cayman) (2048MB) driver: 1.4.1385
Operating System Microsoft Windows 7
Unrecognized XML in project preferences: max_gfx_cpu_pct
Skipping: 90
Skipping: /max_gfx_cpu_pct
Error loading Lua script 'astronomy_parameters.txt': [string "number_parameters: 4..."]:1: '<name>' expected near '4'
Error reading astronomy parameters from file 'astronomy_parameters.txt'
Trying old parameters file
Using SSE4.1 path
Found 1 platform
/ Reply
Quote
Well it did for me -- HD 4850, Win XP, BM 7.0.08.
The thing is, right now we are solving for five variables (AMD video card, OS, BM, AMD driver, and GPU application). Makes it difficult to troubleshoot.
What I think we can state is that earlier (pre 1.0 deploy) these issues were seemingly quiet.
I figure I will things settle out perhaps waiting on more serial number 01a experiences to result in a version stability scenario. So I'll watch and wait -- and while that goes on, let Collatz, Moo, and to a lesser degree POEM run ATI GPU traffic.
Rick, from what I've seen, 12.1 doesn't break Moo, but it DOES break Collatz.
No, it does not.
This is my A8 system, running win7 x64, BM 7.0.12 and CCC 12.1
GPU 0 is the HD5830
GPU 1 is on-chip HD6550
____________
/ Reply
Quote
Still can't get any WUs for my XP 32 bit box if I use the app_info I was using before to reduce the GPU usage. Worked fine before. Anyone have a suggestion. File used is below.
<app_info>
<app><!-- GPU app v0.82 for MW Separation -->
<name>milkyway</name>
</app>
<file_info>
<name>milkyway_separation_0.82_windows_intelx86__ati14.exe</name>
<executable/>
</file_info>
<app_version>
<app_name>milkyway</app_name>
<version_num>82</version_num>
<plan_class>ati14</plan_class>
<flops>1.0e11</flops>
<avg_ncpus>0.05</avg_ncpus>
<max_ncpus>1</max_ncpus>
<coproc>
<type>ATI</type>
<count>1.0</count>
</coproc>
<cmdline>--gpu-target-frequency 80 --gpu-polling-mode 16 --process-priority 1</cmdline>
<file_ref>
<file_name>milkyway_separation_0.82_windows_intelx86__ati14.exe</file_name>
<main_program/>
</file_ref>
</app_version>
</app_info>
Never mind. I figured it out for XP 32 bit. Can someone help me with an app_info command that will throttle down the GPU on Win7 64 bit?
Thanks in advance.
/ Reply
Quote
The thing is, right now we are solving for five variables (AMD video card, OS, BM, AMD driver, and GPU application). Makes it difficult to troubleshoot.
You are absolutely right, and Matt and the Team are doing a wonderful job with the implementation of a very new technology. It's really impressive how many problem have been solved within a couple of hours.
CONGRATULATION, Matt & Team!
/ Reply
Quote
Radeon 38xx cards aren't supported with new stuff
Recently bought a new HD3850 AGP. Cost me quite a lot, also time & effort.
Not amused.
(FWIW I just received & processed a few 0.82 tasks. I think I'll keep this PC open for MW just to check.)
/ Reply
Quote
Using Catalyst 11.3 it looks like. Try a newer driver. I am sort of aware of a few various crashes with different Catalyst versions on different GPUs. I think 11.3-4 and 11.10-11.12 are the most problematic versions; 12.1 seems to be working most consistently on everything. Around 11.3 the binary format stuff was introduced (which it says it supports) but it's probably crasher around then.
Upgraded from 11.3 to 11.9 (2-3% slower on mw w/ CAL, system more sluggish w/ other ATI apps too) and still getting
Exit status -1073740777 (0xffffffffc0000417) Unknown error number
when trying v1.00 OpenCL.
Thought 11.9 should be working for XP 32bit with HD5850?
/ Reply
Quote
Rick, from what I've seen, 12.1 doesn't break Moo, but it DOES break Collatz.
Thanks Barry. Because I run both Moo and Collatz on this host I'll stand pat for now. I'm 100% Moo at the moment, but I'll try something different when I see something change at MW.
I'm the kinda guy that likes to push the do-not-push button (retired engineer...sigh), but I guess that I'll wait for the dust to settle here. Is anyone settling the dust?
Rick
/ Reply
Quote
Thought 11.9 should be working for XP 32bit with HD5850?
12.1 is working for me on XP 32 bit with a 5830 but I'm using the 0.82 separation app not the 1.00.
____________
/ Reply
Quote
FYI, 12.1 drivers don't break Collatz - I tried them. However, I quickly moved up to 12.2 (in pre-release) because of issues with the video converter (nothing to do with BOINC). 12.2 works fine with MW and Collatz - both CAL and OpenCL are working. CAUTION - since this is pre-release, I'm not recommending anyone move to Catalyst 12.2 (which they now call AMD Vision) - your risk & peril.
Equipment
AMD Phenom 9850 4x, Radeon HD5830 on Win 7 PRO 64
____________
Me: Go forth, GPU, plumb the depths of the galactic void to forever advance MilkyWay's lofty endeavour!
GPU: Last thing I need is more hot air...
Me: Shed the light of knowledge! Engage on the starry ways!
GPU: Warp core collapse in 10 seconds...
/ Reply
Quote
For Nvidia it looks like running 2 WU's on the same GPU uses a lot of the CPU's resources, while running 1 WU (still with an app_info.xml) uses a lot less. But running without an app_info.xml uses even less CPU resources.
Edit: I'm using driver 290.53.
Anthony.
____________
The longer I live, the more reasons I develop for wanting to die.
/ Reply
Quote
Radeon 38xx cards aren't supported with new stuff
Recently bought a new HD3850 AGP. Cost me quite a lot, also time & effort.
Not amused.
(FWIW I just received & processed a few 0.82 tasks. I think I'll keep this PC open for MW just to check.)
Currently still happily processing lots more new 0.82 tasks, aborting 1.00 tasks manually.
Did someone suddenly think "DUH! Way too much hassle dude. FIXITNOW. Generate new, old-style 0.82 tasks because everyone can crunch THEM. Then we might, sometime much later look at the project participants' reaction (AKA flak on the forum) to decide what damage was caused by abandoning CAL ... Nota Bene:- AFTER the event, BEFORE asking the obvious question.
Ray
FWIW I have used only one apostrophe above and it is in the correct place according to Word 2000. Flay me if Word is wrong ;-}
/ Reply
Quote
Just as a follow up note, I had the same result on a Win7 64 bit configuration -- and since (for whatever reason) with that combination I did not see any MW workunits come in, I reverted -- Boinc 7.08, HD 4850, Win7 64 bit, Catylist 11.4. I had issues with 11.10 at the OS level. That combination seems to work fine with Collatz, Moo, and even POEM. MW -- nope, it doesn't like the Catalyst version (or so it reports).
I have three or four workstations which for some reason download a BUNCH of MW .82's to chew on -- I am letting them process (temporarily suspending Collatz and Moo on those systems). I mean well over 100 workunits. I suspect they came through when Matt put .82's back in the queue but for some reason they weren't under the 25 work unit limit control. I figure, no big deal, since those are just about the ONLY systems in my farm that can currently get and process MW work, I figure at the moment that balances things out a bit.
I am still hoping that some brilliant mind figures out the right combination of variables to work like *normal* as the current, try this, try that clearly is trying one thing === *patience*. So for now, I'll wait on the right combination being confirmed that works for configurations like mine (and doesn't mess up Collatz, Moo or POEM at the same time).
Rick, from what I've seen, 12.1 doesn't break Moo, but it DOES break Collatz.
Thanks Barry. Because I run both Moo and Collatz on this host I'll stand pat for now. I'm 100% Moo at the moment, but I'll try something different when I see something change at MW.
I'm the kinda guy that likes to push the do-not-push button (retired engineer...sigh), but I guess that I'll wait for the dust to settle here. Is anyone settling the dust?
Rick
/ Reply
Quote
Right -- but it did for me -- two different systems -- and one not all that much different from yours -- Phenom II 925, Win 7 64 bit, 4850 and BOINC 7.0.08.
FYI, 12.1 drivers don't break Collatz - I tried them. However, I quickly moved up to 12.2 (in pre-release) because of issues with the video converter (nothing to do with BOINC). 12.2 works fine with MW and Collatz - both CAL and OpenCL are working. CAUTION - since this is pre-release, I'm not recommending anyone move to Catalyst 12.2 (which they now call AMD Vision) - your risk & peril.
Equipment
AMD Phenom 9850 4x, Radeon HD5830 on Win 7 PRO 64
/ Reply
Quote
I'm using Nvidia but CPU usage is still quite high.In the <cmdline></cmdline>
Where do I put --gpu-wait-factor and --gpu-polling-mode, and how to lower CPU usage even more?
edit: And do I need to change the <type> tag?No
/ Reply
Quote
Started to get the several errors today on 3 different systems with GPUs.I've seen this problem mentioned before with a few different causes. The Catalyst installer/uninstaller at some points didn't clean up the ICD entries in the registry properly. Try removing the drivers and installing newer ones.
Suggestions?
-------------------------------------
http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=99668860
Coprocessors AMD ATI Radeon HD 5800 series (Cypress) (2048MB) driver: Operating System Microsoft Windows XP
Using SSE3 path
Error getting number of platform (-1001): CL_PLATFORM_NOT_FOUND_KHR
/ Reply
Quote
which extension are you looking for?It looks for cl_khr_fp64 or cl_amd_fp64, if you have a client new enough to report OpenCL information back. It looks like some of the 7.0.x aren't sending back the extensions properly.
My GPU (cayman) still does not have cl_khr_fp64 (promised by amd since launch, still not delivered to VLIW4). We're stuck on cl_amd_fp64. Can you please accept it in your app or it's not valid for you?
Thanks a lot!
I was on 7.0.12, upgraded to 7.0.14, still the same problem.
Maybe we need to open a ticket at boinc/trac ?
I don't want to roll back to 6.12 series, because I would loose all the WU in progress and because I would like to test also the boinc/vbox project which requires 7.x BM...
Anyway, thanks for your hard work. All the troubles that will be fixed here will be useful also for all the other projects.
/ Reply
Quote
I updated the GPU stuff to 1.02. This should fix the multiple vendor selection for some people, and hopefully works around the common error on older Nvidia drivers with the nonsense "call to 'mad' is ambiguous" error.
The opencl_amd is now known as opencl_amd_ati because I'm pretty sure the lack of the substring 'ati' in it was the cause of the no-ATI work problem yesterday (I hate the plan class system so, so much).
The scheduler should be enforcing more correct driver limit ranges now also.
g.g. for Radeon 4xxx: more broken than I thought. It only works (and you should only get it) if your catalyst version is between 11.7 and 11.9 or >= 12.1
/ Reply
Quote
I was on 7.0.12, upgraded to 7.0.14, still the same problem.I'm not sure. I've been mutilating the scheduler trying to quickly dodge driver version issues so it's quite likely not working, though it does seem to be working for me using 7.0.14
Maybe we need to open a ticket at boinc/trac ?
/ Reply
Quote
I was on 7.0.12, upgraded to 7.0.14, still the same problem.I'm not sure. I've been mutilating the scheduler trying to quickly dodge driver version issues so it's quite likely not working, though it does seem to be working for me using 7.0.14
Maybe we need to open a ticket at boinc/trac ?
I'm on 12.1 since they went out.
I just upgraded to 7.0.15 but still got the same message that I lack the necessary extension. I wrote something here, but I don't know if it will be read by boinc devs http://boinc.berkeley.edu/dev/forum_thread.php?id=7273
/ Reply
Quote
It was scheduler problem. I don't have any pure cl_amd_fp64 systems so it was working for me. It should start working for you now.I was on 7.0.12, upgraded to 7.0.14, still the same problem.I'm not sure. I've been mutilating the scheduler trying to quickly dodge driver version issues so it's quite likely not working, though it does seem to be working for me using 7.0.14
Maybe we need to open a ticket at boinc/trac ?
I'm on 12.1 since they went out.
I just upgraded to 7.0.15 but still got the same message that I lack the necessary extension. I wrote something here, but I don't know if it will be read by boinc devs http://boinc.berkeley.edu/dev/forum_thread.php?id=7273
/ Reply
Quote
Right -- but it did for me -- two different systems -- and one not all that much different from yours -- Phenom II 925, Win 7 64 bit, 4850 and BOINC 7.0.08.
FYI, 12.1 drivers don't break Collatz - I tried them.
Equipment
AMD Phenom 9850 4x, Radeon HD5830 on Win 7 PRO 64
The important difference is the GPU. CAL (which is used by Collatz) is only broken since 11.11 for HD4XXX-GPUs. Newer GPUs (like HD5XXXX) are working.
So far the 1.00 OpenCL app has worked well for my HD4870 using Catalyst 12.1 :)
Thank you very much Matt.
/ Reply
Quote
It was scheduler problem. I don't have any pure cl_amd_fp64 systems so it was working for me. It should start working for you now.I was on 7.0.12, upgraded to 7.0.14, still the same problem.I'm not sure. I've been mutilating the scheduler trying to quickly dodge driver version issues so it's quite likely not working, though it does seem to be working for me using 7.0.14
Maybe we need to open a ticket at boinc/trac ?
I'm on 12.1 since they went out.
I just upgraded to 7.0.15 but still got the same message that I lack the necessary extension. I wrote something here, but I don't know if it will be read by boinc devs http://boinc.berkeley.edu/dev/forum_thread.php?id=7273
yes you're right!
Thanks a lot man!
/ Reply
Quote
Thank you matt.
I also try the 7.0.12 version of BOINC client which seems to work more better than the 7.0.14.
____________
/ Reply
Quote
Hello,
Opposite to my first post above, I have encountered problems now. I have very high cpu-usage with 1.00 and 1.02 app and two 5850 GPU.
I faced the problem after upgrading an old ATI-beta-driver from December, which worked well for playing Skyrim, but shows intermittent BOD. With this driver Milkyway runs with very low cpu-usage very fine. I deinstalled the beta driver, cleaned with driver-sweeper as I always do when installing new ATI-driver, reboot, install new driver (first attempt 12.2 preview), reboot, starting boinc. Result: Milkyway is starting with high cpu-usage (M@H wants one core).
I tried several drivers installed in the above scheme. 11.11c, 12.1, reverting to old (previous functioning) beta-driver, 12.2 preview. Tried app_info w/o cmdline and without app_info. Changed crossfire enabled/disabled. Still the same result M@H wants a whole core.
example wu: http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=100185622
Collatz and Moo are running fine with every driver.
Boinc 6.12.34 (upgrading to 7.0.x shows no effect)
Win 7 64 Bit
Do you need further information?
I hope for suggestions from you.
Thanks!
Nowi
____________
/ Reply
Quote
I got only tasks with V1.00 no V1.01.1.01 is just the statically linked 1.00 CPU version. Your Catalyst driver is new enough that it should include libOpenCL. Is it actually not present on your system or in some weird place? What distro and how did you install the drivers?
Do I need new drivers?
I use ubuntu 10.04, driver was installed by the ati-installer.
Now I read, openCL came with ubuntu 10.10.
I tried a distri-upgrade but it doesn't work.
So I have to make a complete new installation. But in the moment I have no time for that.
So I first try to go back to V0.82 by app_info.
/ Reply
Quote
Hello,
Opposite to my first post above, I have encountered problems now. I have very high cpu-usage with 1.00 and 1.02 app and two 5850 GPU.
I faced the problem after upgrading an old ATI-beta-driver from December, which worked well for playing Skyrim, but shows intermittent BOD. With this driver Milkyway runs with very low cpu-usage very fine. I deinstalled the beta driver, cleaned with driver-sweeper as I always do when installing new ATI-driver, reboot, install new driver (first attempt 12.2 preview), reboot, starting boinc. Result: Milkyway is starting with high cpu-usage (M@H wants one core).
I tried several drivers installed in the above scheme. 11.11c, 12.1, reverting to old (previous functioning) beta-driver, 12.2 preview. Tried app_info w/o cmdline and without app_info. Changed crossfire enabled/disabled. Still the same result M@H wants a whole core.
example wu: http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=100185622
Collatz and Moo are running fine with every driver.
Boinc 6.12.34 (upgrading to 7.0.x shows no effect)
Win 7 64 Bit
Do you need further information?
I hope for suggestions from you.
Thanks!
Nowi
I believe you are accurately describing the actions of the 12 series drivers from AMD, they want more cpu to make gaming better on their video output. Trying going back to the 11.9 series of AMD drivers and see if that is better.
BTW upgrading to the 7 series of Boinc means you have a whole other set of problems with your cache levels! And NO you cannot downgrade without wiping out all your existing units
/ Reply
Quote
I have been running Milkyway@home projects on my ATI 4850. Are these cards no longer supported?
/ Reply
Quote
I have been running Milkyway@home projects on my ATI 4850. Are these cards no longer supported?They are
/ Reply
Quote
Opposite to my first post above, I have encountered problems now. I have very high cpu-usage with 1.00 and 1.02 app and two 5850 GPU.Do you actually have high CPU usage, or is BOINC just telling you it is using 0.98 CPUs? I see that one one of my systems (Using BOINC 7) and not on another (Using 6.12)
/ Reply
Quote
No, it is really high cpu-usage according to the windows task-manager which shows usage up to 25 % on my quad-core for each milkyway process. Boinc shows 0.05 as expected. Please show also in my example wu in the original post. The cpu-time is very high!
____________
/ Reply
Quote
I'm using Nvidia but CPU usage is still quite high.
Where do I put --gpu-wait-factor and --gpu-polling-mode, and how to lower CPU usage even more?
In the <cmdline></cmdline.
I found out where to put it, but what numbers should I use to lower CPU usage?
BTW upgrading to the 7 series of Boinc means you have a whole other set of problems with your cache levels! And NO you cannot downgrade without wiping out all your existing units
Can I upgrade to BOINC 7 without wiping out all of mine existing/cached/running WU's?
Anthony.
____________
The longer I live, the more reasons I develop for wanting to die.
/ Reply
Quote
I have not had a problem upgrading, its downgrading back to 6.X.X thats the problem because 7.X.X deals with files differently, and when thats presented to 6.X.X it dont like it and wont play. Going up to V7, BOINC V7 does all the conversions necessary, dont happen the other way as BOINC V6 knows nothing about it.
.... but so far, going up to 7.X.X has been trouble free for me
Regards
Zy
/ Reply
Quote
I found out where to put it, but what numbers should I use to lower CPU usage?Raising the GPU wait factor (--gpu-wait-factor ) and/or changing the polling mode (--gpu-polling-mode ). The default wait factor is 0.75, to wait for 75% of the estimated kernel run times before waiting. Raising it towards 1 should lower CPU usage. If you want to play with that try raising the wait factor first.
Can I upgrade to BOINC 7 without wiping out all of mine existing/cached/running WU's?I think I saw that it wipes it if the version changes, so no.
/ Reply
Quote
I found out where to put it, but what numbers should I use to lower CPU usage?Raising the GPU wait factor (--gpu-wait-factor ) and/or changing the polling mode (--gpu-polling-mode ). The default wait factor is 0.75, to wait for 75% of the estimated kernel run times before waiting. Raising it towards 1 should lower CPU usage. If you want to play with that try raising the wait factor first.
Can I upgrade to BOINC 7 without wiping out all of mine existing/cached/running WU's?I think I saw that it wipes it if the version changes, so no.
No, it is fine going up to version 7 from version 6, it is only a problem if you want to go back to version 6.
____________
/ Reply
Quote
No, it is really high cpu-usage according to the windows task-manager which shows usage up to 25 % on my quad-core for each milkyway process. Boinc shows 0.05 as expected. Please show also in my example wu in the original post. The cpu-time is very high!I know the AMD drivers had the high CPU problem in 11.7 and 11.8 but I guess that could be different on different hardware. Does it happen with the current 12.1 drivers for you?
/ Reply
Quote
Matt,
I have two computers with ATI GPU's
Q9400 - ATI5850 recently doing poem OPEN_cl app
17-920 - ATI4850 recently doing Milkys CAL app
The i7-920 -ATI4850 stopped getting milky so I went to the forum and read the latest posts.
Consequently I upgraded Boinc to 7.0.12 and the ATI CCC to 11.9 I still could not get the new WU for Open_cl
I switched the q9400 - ATI5850 from POEM to Milky By by suspending poem and hitting update.
The new app was downloaded to:
C:\Documents and Settings\All Users\Application Data\BOINC\projects\milkyway.cs.rpi.edu_milkyway
APP:milkyway_separation_1.02_windows_intelx86__opencl_amd_ati
and the app started immediately
the Q9400 has boinc 7.0.12 and ATI CCC 11.9 identical to the setup to the i7-920
Confused , I restarted the i7-920 and updated the project and tried again (no luck)
I reset the project (no luck)
I restarted (no luck)
I removed milky and re-installed using boinc manager and updated the project (no luck)
C:\Documents and Settings\All Users\Application Data\BOINC\projects\milkyway.cs.rpi.edu_milkyway directory is now empty.
Messages in boinc read:
Host Project Date Message
xp-pro-i7 Milkyway@Home 11/02/2012 1:48:27 AM Sending scheduler request: To fetch work.
xp-pro-i7 Milkyway@Home 11/02/2012 1:48:27 AM Requesting new tasks for ATI
xp-pro-i7 Milkyway@Home 11/02/2012 1:48:29 AM Scheduler request completed: got 0 new tasks
xp-pro-i7 Milkyway@Home 11/02/2012 1:48:29 AM No tasks sent
xp-pro-i7 Milkyway@Home 11/02/2012 1:48:29 AM Tasks for CPU are available, but your preferences are set to not accept them
On restart the boinc manager reports this:
Host Project Date Message
xp-pro-i7 --- 11/02/2012 12:54:44 AM Starting BOINC client version 7.0.12 for windows_intelx86
xp-pro-i7 --- 11/02/2012 12:54:44 AM log flags: file_xfer, sched_ops, task
xp-pro-i7 --- 11/02/2012 12:54:44 AM Libraries: libcurl/7.21.6 OpenSSL/1.0.0d zlib/1.2.5
xp-pro-i7 --- 11/02/2012 12:54:44 AM Data directory: C:\Documents and Settings\All Users\Application Data\BOINC
xp-pro-i7 --- 11/02/2012 12:54:44 AM Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 sse4_2 nx lm vmx tm2 popcnt pbe
xp-pro-i7 --- 11/02/2012 12:54:44 AM ATI GPU 0: ATI Radeon HD 4700/4800 (RV740/RV770) (CAL version 1.4.1546, 512MB, 471MB available, 2000 GFLOPS peak)
xp-pro-i7 --- 11/02/2012 12:54:44 AM OpenCL: ATI GPU 0: ATI RV770 (driver version CAL 1.4.1546, device version OpenCL 1.0 AMD-APP-SDK-v2.5 (732.1), 256MB, 471MB available)
xp-pro-i7 --- 11/02/2012 12:54:44 AM ATI GPU 0 is OpenCL-capable
From this end it looks like the 4850's are not being recognised as supportable.
EDIT:
I found a difference in open cl versions
OpenCL: ATI GPU 0: Cypress (driver version CAL 1.4.1546, device version OpenCL 1.1 AMD-APP-SDK-v2.5 (732.1), 512MB, 945MB available)
/ Reply
Quote
No, it is really high cpu-usage according to the windows task-manager which shows usage up to 25 % on my quad-core for each milkyway process. Boinc shows 0.05 as expected. Please show also in my example wu in the original post. The cpu-time is very high!I know the AMD drivers had the high CPU problem in 11.7 and 11.8 but I guess that could be different on different hardware. Does it happen with the current 12.1 drivers for you?
I tried 11.11c, 12.1, 12.2 preview. Every driver showed the high cpu-usage.
____________
/ Reply
Quote
Where can I download the latest Boinc 7 version? The Boinc site only show the 7.0.8 version?
____________
The longer I live, the more reasons I develop for wanting to die.
/ Reply
Quote
You can find the version here: http://boinc.berkeley.edu/dl/?C=M;O=D
____________
/ Reply
Quote
I was getting high CPU usage (one CPU core per WU) with 11.9 drivers but it is ok with 12.1, I just hope it doesn't BSOD again like it did before with 12.1
/ Reply
Quote
I'am fucked up with all the mW versions, why not keep it to simple 1.
Some host got work and some not, all running on same old boinc and almost same GPU!
Please admins, put a email to all of us next time before upgrading:
We are upgrading next week separation to 1.00, please upgrade
your boinc client to x.xx and min gpu driver to x.xx.
/ Reply
Quote
It depends on the driver version and GPU. The most problematic GPU are Radeon 4xxx series. There are only 4 driver releases in the past year that I'm certain work (11.7, 11.8, 11.9 and 12.1+) so unless you're using one of those you won't get work for it.
/ Reply
Quote
Indeed it does -- also the version of Boinc Manager.
With 11.8 (which also works with Collatz -- a critical constraint for me), and Vista 32 bit and Windows 7 64 bit, it does work -- with one of the 6.x Boinc managers. With 11.8 and XP 32 bit and BM 7.0.8 it does not get work, go figure.
It depends on the driver version and GPU. The most problematic GPU are Radeon 4xxx series. There are only 4 driver releases in the past year that I'm certain work (11.7, 11.8, 11.9 and 12.1+) so unless you're using one of those you won't get work for it.
/ Reply
Quote
With 11.8 (which also works with Collatz -- a critical constraint for me), and Vista 32 bit and Windows 7 64 bit, it does work -- with one of the 6.x Boinc managers. With 11.8 and XP 32 bit and BM 7.0.8 it does not get work, go figure.
that's my dilemma too - i need to be setup so that both Milkyway and Collatz will run...and unfortunately my platform is WinXP 32-bit like yours. on the other hand, i'm running a different driver and BOINC version, so i suppose there's a chance it'll work if i try it. i just haven't been able to make the time to play around with it all. i'm thinking i'll tackle it sometime next week after the Valentine's Day bonus.
____________
/ Reply
Quote
Indeed it does -- also the version of Boinc Manager.
With 11.8 (which also works with Collatz -- a critical constraint for me), and Vista 32 bit and Windows 7 64 bit, it does work -- with one of the 6.x Boinc managers. With 11.8 and XP 32 bit and BM 7.0.8 it does not get work, go figure.
today matt implemented a fix for the scheduler. I think that now it should work also in your "11.8 with BM 7.0.8" configuration, because you're in what he called and discarded previously (wrongly) a "pure cl_amd_fp64" extension situation
/ Reply
Quote
Those having apparent work-fetch problems with 7.X.X .... at present it works in a different way, I suspect - only that, no definitive knowledge - that its being designed to reduce cache sizes at the user end, particularly massive days long ones due to the problems that can give servers.
At present, the cache will reduce down without being refreshed. At the point when its almost at dry, it will then download what you set in the cache preferences all the other supply constraints being equal. Often had 200 WU downloads on other projects in one lump after apparent famine. Its feast or famine and will not play unless the cache is near zero.
Work in progress .... will they change that .... I have no idea, but meanwhile be aware of the behaviour - its easy to get the perception it will never refill, it will (usually), but it does it at the last minute.
Regards
Zy
/ Reply
Quote
I saw that and was hopeful, but...
I tried this on a Win64 system with an HD 4850 and BM 7.0.8
1) Uninstalled Catalyst 11.4
2) Rebooted, installed Catalyst 11.8
3) Rebooted, removed MW from BM 7.08
4) Reattached to MW
See messages below:
2/10/2012 1:57:54 PM | Milkyway@Home | Master file download succeeded
2/10/2012 1:57:59 PM | Milkyway@Home | Sending scheduler request: Project initialization.
2/10/2012 1:57:59 PM | Milkyway@Home | Requesting new tasks for CPU and ATI
2/10/2012 1:58:00 PM | Milkyway@Home | Scheduler request completed: got 0 new tasks
2/10/2012 1:58:00 PM | Milkyway@Home | No tasks sent
2/10/2012 1:58:00 PM | Milkyway@Home | Message from server: Catalyst driver version is not OK for OpenCL application with this GPU
2/10/2012 1:58:00 PM | Milkyway@Home | Tasks for CPU are available, but your preferences are set to not accept them
2/10/2012 1:58:00 PM | Milkyway@Home | New computer location: home
Indeed it does -- also the version of Boinc Manager.
With 11.8 (which also works with Collatz -- a critical constraint for me), and Vista 32 bit and Windows 7 64 bit, it does work -- with one of the 6.x Boinc managers. With 11.8 and XP 32 bit and BM 7.0.8 it does not get work, go figure.
today matt implemented a fix for the scheduler. I think that now it should work also in your "11.8 with BM 7.0.8" configuration, because you're in what he called and discarded previously (wrongly) a "pure cl_amd_fp64" extension situation
/ Reply
Quote
The other thing to note, with the earlier BM (6.10.58 or 6.12.32) which do get work with the 11.8 Catalyst -- they get .82 separation work units -- which process fine.
I think the 7.0.8 BM 'wants' the 1.0 separation work units with OpenCL, but MW doesn't 'like' this.
The deal here which might be interesting, POEM GPU *requires* the BOINC 7.x series (and seems picky about sub versions there), and runs an OpenCL application there.
So at the moment, I have an either/or -- POEM GPU or MW GPU. I don't run much for POEM GPU (it is new for those guys and so has some issues AND is a CreditNew (bad, bad) house these days.
Also, as long as .82 separation works with work units I will be OK for now on all the 6.10.58 and 6.12.32 BM + 11.8 Catalyst combinations I have deployed. When .82 no longer works, unless things are fixed, I will be 'MW-less'.
____________
/ Reply
Quote
....When .82 no longer works, unless things are fixed, I will be 'MW-less'.
That will be increasingly common story with 4XXX as the year moves on as 4XXX have limited OpenCL support. Matt has done some sterling tweeks in the background to get this going here for 4XXX, but the longer term prognosis re 4XXX and OpenCL across BOINC is one of gradual degrading as full blown OpenCL takes hold across BOINC, as a consequence of the dropping of CAL by AMD.
Not happening "by yesterday", but it will happen, and its irreversable
Regards
Zy
/ Reply
Quote
The other thing to note, with the earlier BM (6.10.58 or 6.12.32) which do get work with the 11.8 Catalyst -- they get .82 separation work units -- which process fine.
I think the 7.0.8 BM 'wants' the 1.0 separation work units with OpenCL, but MW doesn't 'like' this.
The deal here which might be interesting, POEM GPU *requires* the BOINC 7.x series (and seems picky about sub versions there), and runs an OpenCL application there.
So at the moment, I have an either/or -- POEM GPU or MW GPU. I don't run much for POEM GPU (it is new for those guys and so has some issues AND is a CreditNew (bad, bad) house these days.
Also, as long as .82 separation works with work units I will be OK for now on all the 6.10.58 and 6.12.32 BM + 11.8 Catalyst combinations I have deployed. When .82 no longer works, unless things are fixed, I will be 'MW-less'.
The other thing to note, with the earlier BM (6.10.58 or 6.12.32) which do get work with the 11.8 Catalyst -- they get .82 separation work units -- which process fine.
Nopem, do not, errors out running on xp 32......
/ Reply
Quote
Fair enough, when it happens, I'll process elsewhere -- new double precision AMD cards are quite an expense, and also, from what I've seen require more power as well. I'm not inclined to rebuild a batch of systems for that purpose. When other projects also move to OpenCL -- I'll probably shut down my BOINC work as I simply can't justify that sort of investment in time for the projects.
....When .82 no longer works, unless things are fixed, I will be 'MW-less'.
That will be increasingly common story with 4XXX as the year moves on as 4XXX have limited OpenCL support. Matt has done some sterling tweeks in the background to get this going here for 4XXX, but the longer term prognosis re 4XXX and OpenCL across BOINC is one of gradual degrading as full blown OpenCL takes hold across BOINC, as a consequence of the dropping of CAL by AMD.
Not happening "by yesterday", but it will happen, and its irreversable
Regards
Zy
____________
/ Reply
Quote
Works OK here for me as I reported -- no big deal though, a lot of variables in the mix (I run 4850's).
The other thing to note, with the earlier BM (6.10.58 or 6.12.32) which do get work with the 11.8 Catalyst -- they get .82 separation work units -- which process fine.
Nopem, do not, errors out running on xp 32......
/ Reply
Quote
BarryAZ can lik his ass on mw if he like to do so!!!!!!!!!!
I am PUT ME MYSELF OUT from mv............
/ Reply
Quote
Prices are bound to come down, so dont be too depressed, there's no way they will stay this high for good. Also another new series like 8XXX are probably 12-18 months away, that will also help depress prices of 5XXX and 6XXX both of which have full OpenCL support.
Personal guess is it becomes a real drama circa 18months from now when all have gone OpenCL (as such), but could be that future series will knock down 5XXX and 6XXX into more affordable scenarios by then. Difficult to say, but on balance I reckon it will trend that way with second prices of 5XXX and 6XXX.
Regards
Zy
/ Reply
Quote
Just upgraded to BOINC 7.0.15 but it's not requesting any tasks, for any of my projects except SETI.
EDIT: I'm running dry on all projects.
Anthony.
____________
The longer I live, the more reasons I develop for wanting to die.
/ Reply
Quote
Those having apparent work-fetch problems with 7.X.X .... at present it works in a different way, I suspect - only that, no definitive knowledge - that its being designed to reduce cache sizes at the user end, particularly massive days long ones due to the problems that can give servers.
It certainly works in a different way but it's nothing to do with reducing cache sizes. The two cache settings now have different meanings in 7.x.x. The 'Connect Every ...' (CE) setting should now be regarded as the 'low water' mark. Many people would have this at zero (from past behaviour) and if you do, that is why BOINC wont get work until you actually run out. The 'Extra Days' (ED) setting should now be regarded as an increment to define a 'high water' mark.
So, if you want BOINC to maintain a cache between X and X.01 days, you would set CE=X and ED=0.01.
It's as simple as that. Of course, it's not really documented properly anywhere and if you gripe about that you would probably be told that nothing is guaranteed with 'bleeding edge' versions so why are you using one ... Projects will continue to try to use new features that aren't ready for prime time so you can't really blame BOINC entirely for all the teething problems. It might have been better for volunteers if the new MW 1.0x app had been released in a testing environment similar to the way that the new Einstein CL app is being tested in the Albert@Home test project. I know that requires a lot more resources at the project end but it's a lot more 'friendly' for the long suffering volunteers.
At present, the cache will reduce down without being refreshed. At the point when its almost at dry, it will then download what you set in the cache preferences all the other supply constraints being equal. Often had 200 WU downloads on other projects in one lump after apparent famine. Its feast or famine and will not play unless the cache is near zero.
This is exactly what to expect if you are using 7.x.x with 'old style' cache settings - something like CE=0 and ED=3.0 for example.
____________
Cheers,
Gary.
/ Reply
Quote
Could be -- I've over a dozen 4850 systems in play -- then again, it might encourage me to simply shut down a batch of systems and reduce the electric bill. I tend to do that in the summer anyway.
Prices are bound to come down, so dont be too depressed, there's no way they will stay this high for good. Also another new series like 8XXX are probably 12-18 months away, that will also help depress prices of 5XXX and 6XXX both of which have full OpenCL support.
Personal guess is it becomes a real drama circa 18months from now when all have gone OpenCL (as such), but could be that future series will knock down 5XXX and 6XXX into more affordable scenarios by then. Difficult to say, but on balance I reckon it will trend that way with second prices of 5XXX and 6XXX.
Regards
Zy
____________
/ Reply
Quote
Yes I figured that out a while ago -- though I think it is not quite so much 'as simple as that' -- there are (as it always the case) some other variables in play.
It's as simple as that. Of course, it's not really documented properly anywhere and if you gripe about that you would probably be told that nothing is guaranteed with 'bleeding edge' versions so why are you using one ... Projects will continue to try to use new features that aren't ready for prime time so you can't really blame BOINC entirely for all the teething problems. It might have been better for volunteers if the new MW 1.0x app had been released in a testing environment similar to the way that the new Einstein CL app is being tested in the Albert@Home test project. I know that requires a lot more resources at the project end but it's a lot more 'friendly' for the long suffering volunteers.
/ Reply
Quote
This is exactly what to expect if you are using 7.x.x with 'old style' cache settings - something like CE=0 and ED=3.0 for example.
The new cache settings in BOINC 7.0.xx are a real PITA but if you set connect everyday to 1.00 and additional cache to 0.50 you should get work.
XP 32 and WIN7 64 both crunching away. It's the connect everyday setting that is the most important with the newer versions.
____________
/ Reply
Quote
New 1.02 running fine on my HD6970 with Win7 64 catalyst 12.1 (preview) and BM 6.10.60
All wu's running under 4 seconds of CPU time.
GPU time average 60 seconds/wu
I use Efmer priority to keep it on 'high'.
/ Reply
Quote
So I first try to go back to V0.82 by app_info.
Done that. All works fine.
I'm back and crunching.
Please keep it alive.
/ Reply
Quote
I too have gotten work from MW and from SETI, but not from GPUGrid. Normally I have 2 WU's on my system (per GPU), but now I only have one crunching and it won't request a new one. I'm running BOINC 7.0.15. I'm running GPUGrid on 1 GPU.
I know this is the MW forum but maybe there's someone who could help. I also posted this on the GPUGrid forum, but I don't know how long it will take them to response, and this thread here is going like a train.
Anthony.
____________
The longer I live, the more reasons I develop for wanting to die.
/ Reply
Quote
Hello,
I am not getting new work, this is the driver:
2/11/2012 1:09:02 AM ATI GPU 0: ATI Radeon HD5800 series (Cypress) (CAL version 1.4.1332, 1024MB, 2720 GFLOPS peak)
That should work though?
Where can I find new CAL drivers?
Thanks.
____________
Greetings from,
TJ
/ Reply
Quote
TJ:
Look at the lines following this part - they should show something like this:
2012-02-10 07:22:48 | | ATI GPU 0: ATI Radeon HD 5800 series (Cypress) (CAL version 1.4.1683, 1024MB, 991MB available, 3584 GFLOPS peak)
2012-02-10 07:22:48 | | OpenCL: ATI GPU 0: Cypress (driver version CAL 1.4.1683 (VM), device version OpenCL 1.1 AMD-APP (873.1), 1024MB)
2012-02-10 07:22:48 | | ATI GPU is OpenCL-capable
You need OpenCL to run MW on a 5800, unless you use some of the hacks described above...
____________
Me: Go forth, GPU, plumb the depths of the galactic void to forever advance MilkyWay's lofty endeavour!
GPU: Last thing I need is more hot air...
Me: Shed the light of knowledge! Engage on the starry ways!
GPU: Warp core collapse in 10 seconds...
/ Reply
Quote
Hi Maac59,
Thanks for your quick response. I don't see any of those messages.
I know that I was crunching happily last week with MW, and now nothing works on the 2 ATI cards.
I got this message all the time:
2/11/2012 1:11:12 AM Milkyway@Home Message from server: Catalyst driver version is not OK for OpenCL application with this GPU
Where can I find the OpenCL?
Thanks.
____________
Greetings from,
TJ
/ Reply
Quote
Hi Maac59,
Thanks for your quick response. I don't see any of those messages.
I know that I was crunching happily last week with MW, and now nothing works on the 2 ATI cards.
I got this message all the time:
2/11/2012 1:11:12 AM Milkyway@Home Message from server: Catalyst driver version is not OK for OpenCL application with this GPU
Where can I find the OpenCL?
Thanks.
http://support.amd.com/us/Pages/AMDSupportHub.aspx
____________
/ Reply
Quote
You probably need to update your Catalyst drivers to get the latest OpenCL package. As for the solutions I was referring to, they deal with using the app_info file in convincing MW to keep using the old CAL application. Do a search on app_info on this page.
/ Reply
Quote
Deleted
/ Reply
Quote
This host (AMD 1090t, Win7-64, HD 5870, BOINC 6.12.34) is back in business after upgrading to Cat 12.1. I don't know if any of the 11.x versions would have worked -- I went straight up from 10.10. No other changes were made. WU's now complete 4-5% faster (2-3 sec) on average.
____________
/ Reply
Quote
- The old CAL version is gone; it's replaced with the OpenCL application
Question: Has the nVidia application also switched from CUDA to OpenCL now? The application name suggests it.
Seems to work fine on a GTX 560 Ti w/ Win7 64bit and BOINC 6.12.34 (x64), first result is valid. The ATI onboard GPU (SP) is not causing problems, server tells me: "An ATI GPU supporting double precision math is required" and "ATI GPU R600 (R38xx) does not support OpenCL".
<core_client_version>6.12.34</core_client_version>
<![CDATA[
<stderr_txt>
<search_application> milkyway_separation 1.02 Windows x86_64 double OpenCL </search_application>
Guessing preferred OpenCL vendor 'NVIDIA Corporation'
Error loading Lua script 'astronomy_parameters.txt': [string "number_parameters: 4..."]:1: '<name>' expected near '4'
Error reading astronomy parameters from file 'astronomy_parameters.txt'
Trying old parameters file
Using SSE3 path
Found 1 platform
Platform 0 information:
Name: NVIDIA CUDA
Version: OpenCL 1.1 CUDA 4.1.1
Vendor: NVIDIA Corporation
Extensions: cl_khr_byte_addressable_store cl_khr_icd cl_khr_gl_sharing cl_nv_d3d9_sharing cl_nv_d3d10_sharing cl_khr_d3d10_sharing cl_nv_d3d11_sharing cl_nv_compiler_options cl_nv_device_attribute_query cl_nv_pragma_unroll
Profile: FULL_PROFILE
Using device 0 on platform 0
Found 1 CL device
Device 'GeForce GTX 560 Ti' (NVIDIA Corporation:0x10de) (CL_DEVICE_TYPE_GPU)
Driver version: 285.62
Version: OpenCL 1.1 CUDA
Compute capability: 2.1
Max compute units: 8
Clock frequency: 1645 Mhz
Global mem size: 2147483648
Local mem size: 49152
Max const buf size: 65536
Double extension: cl_khr_fp64
Build log:
--------------------------------------------------------------------------------
ptxas info : Compiling entry function 'probabilities' for 'sm_21'
ptxas info : Function properties for probabilities
80 bytes stack frame, 76 bytes spill stores, 76 bytes spill loads
ptxas info : Used 62 registers, 100 bytes cmem[0], 56 bytes cmem[16]
--------------------------------------------------------------------------------
Build log:
--------------------------------------------------------------------------------
ptxas info : Compiling entry function 'summarization' for 'sm_21'
ptxas info : Function properties for summarization
0 bytes stack frame, 0 bytes spill stores, 0 bytes spill loads
ptxas info : Used 13 registers, 2048+0 bytes smem, 48 bytes cmem[0]
--------------------------------------------------------------------------------
Estimated Nvidia GPU GFLOP/s: 842 SP GFLOP/s, 105 DP FLOP/s
Using a target frequency of 60.0
Using a block size of 4096 with 17 blocks/chunk
Using clWaitForEvents() for polling with initial wait of 12 ms (mode 0)
Range: { nu_steps = 640, mu_steps = 1600, r_steps = 1400 }
Iteration area: 2240000
Chunk estimate: 32
Num chunks: 33
Chunk size: 69632
Added area: 57856
Effective area: 2297856
Initial wait: 12 ms
Integration time: 542.787896 s. Average time per iteration = 848.106088 ms
Integral 0 time = 546.308173 s
Running likelihood with 109999 stars
Likelihood time = 2.219000 s
<background_integral> 0.000133824638386 </background_integral>
<stream_integral> 6.443820669765729 875.872833232415470 </stream_integral>
<background_likelihood> -2.940328087823295 </background_likelihood>
<stream_only_likelihood> -83.771552699247522 -7.323208286845183 </stream_only_likelihood>
<search_likelihood> -2.935410274019145 </search_likelihood>
11:38:38 (3120): called boinc_finish
</stderr_txt>
]]>
/ Reply
Quote
Thanks Arkayn and Maac59 for your suggestions and link.
It is working smoothly again under Win7-64ultimate, BOINC 6.10.58 and CAL 12.1.
____________
Greetings from,
TJ
/ Reply
Quote
Could be -- I've over a dozen 4850 systems in play -- then again, it might encourage me to simply shut down a batch of systems and reduce the electric bill. I tend to do that in the summer anyway.
Prices are bound to come down, so dont be too depressed, there's no way they will stay this high for good. Also another new series like 8XXX are probably 12-18 months away, that will also help depress prices of 5XXX and 6XXX both of which have full OpenCL support.
Personal guess is it becomes a real drama circa 18months from now when all have gone OpenCL (as such), but could be that future series will knock down 5XXX and 6XXX into more affordable scenarios by then. Difficult to say, but on balance I reckon it will trend that way with second prices of 5XXX and 6XXX.
Regards
Zy
Barry maybe EBay will become your friend and help you unload those cards and get newer, faster ones! Just remember EVERYONE on EBay is NOT honest!!!
/ Reply
Quote
This host (AMD 1090t, Win7-64, HD 5870, BOINC 6.12.34) is back in business after upgrading to Cat 12.1. I don't know if any of the 11.x versions would have worked -- I went straight up from 10.10. No other changes were made. WU's now complete 4-5% faster (2-3 sec) on average.
I am using 11.9 on all of my pc's and I am getting work on two of them! I only have 5870's on my gpu's here at MW, so that is the same but the Boinc version is different and so is whether they are 32 or 64 bit. I re-enabled work so I will wait and see if it picks some up later. I also did a reset on one but nothing was happening, so I left it alone and turned off the monitor but the pc is still running.
/ Reply
Quote
I found that once I had everything set to go, I needed to just leave it alone. The WUs came in one at a time and it took a few minutes to get the next one. I had to force myself not to babysit and manually update Boinc trying to get more work every time it ran out. I let it go overnight with no other GPU projects running. By morning it was running just as before.
Win 7-64
Boinc 7.0.8(x64)
Catalyst 11.9
HD5870 GPU speed = 900MHZ - memory speed = 300MHz
OpenCL app from Arkayn
With an app_info.xml instruction to run two WUs at a time they complete in ~2:12 minutes with approx 5 sec CPU time.
/ Reply
Quote
The Nvidia app has always been OpenCL. We used to be a ATI only project.
____________
/ Reply
Quote
This host (AMD 1090t, Win7-64, HD 5870, BOINC 6.12.34) is back in business after upgrading to Cat 12.1...
For completeness, I might maybe should have added that I'm using the stock 1.02 from the project, no app_info file.
____________
/ Reply
Quote
For completeness, I might maybe should have added that I'm using the stock 1.02 from the project, no app_info file.
the app_info.xml file really isn't necessary unless you either want to change the number of simultaneously running MW@H tasks to something greater than 1 or improve GUI lag/set process priority/etc. via command line parameters.
____________
/ Reply
Quote
I said 2 days ago:
As a point of information my HD 4770 (11.6, BOINC 6.12.34, winXP32, AMD64 3800+) has been very unhappy with the change to OpenCL. Host 179083 ( http://milkyway.cs.rpi.edu/milkyway/show_host_detail.php?hostid=179083 ) gets OpenCL work, but promptly errors out with -1073740777 (unknown error). I detached and tried again with the same result.
Upgrading from Cat 11.6 to Cat 12.1 made MW work without trashing anything. Of course Matt has made a number of changes in the meantime. Moo works too. I wasn't sure Cat 12.1 would support an (old) HD 4770 adequately. I guess it does.
Performance is very near what it was with the CAL app. I've seen a decrease of about 4 seconds from 192 seconds total time, but an increase of 3 seconds (from 6 to 9 seconds) of CPU time. I'd have never noticed if I wasn't looking.
An appropriate classic quote of wisdom:
"Thank god Mr. Spock, you've fixed it"
"There was no deity involved Mr. Scott. It was my cross circuiting from A to B that did it"
/ Reply
Quote
For completeness, I might maybe should have added that I'm using the stock 1.02 from the project, no app_info file.
the app_info.xml file really isn't necessary unless you either want to change the number of simultaneously running MW@H tasks to something greater than 1 or improve GUI lag/set process priority/etc. via command line parameters.
What command would be added to the the app-info file to throttle back GPU usage? It's currently at 99% and I'd like to pull it back to 80-85%.
Win7 64 bit. Radeon 5830.
____________
/ Reply
Quote
For completeness, I might maybe should have added that I'm using the stock 1.02 from the project, no app_info file.
the app_info.xml file really isn't necessary unless you either want to change the number of simultaneously running MW@H tasks to something greater than 1 or improve GUI lag/set process priority/etc. via command line parameters.
What command would be added to the the app-info file to throttle back GPU usage? It's currently at 99% and I'd like to pull it back to 80-85%.
Win7 64 bit. Radeon 5830.
sorry...i've only done it successfully on WinXP 32-bit like you. i have yet to tackle the issue on Win7 x64, let alone install it on one of my machines...but hopefully i'll be getting a licensed copy from work soon...
____________
/ Reply
Quote
What command would be added to the the app-info file to throttle back GPU usage? It's currently at 99% and I'd like to pull it back to 80-85%.
Win7 64 bit. Radeon 5830.
With the new app you don't need the app_info-file to control GPU load - it's now only necessary if you want to run multiple instances on one GPU. If you want to use it anyway the command line parameter apparently wasn't changed from the old app:
--gpu-target-frequency XX
Raising XX lessens the GPU load and increases the run time (default is now 60).
With the new app it's easier controlled in the Milkyway@home preferences: Frequency (in Hz) that should try to complete individual work chunks
____________
Starfire
/ Reply
Quote
Thanks
____________
/ Reply
Quote
Thanks, how do you think I was able to afford the dozen or so 4850's -- I well know about ebay -- and look closely at listings there.
Most of the 4850's I got were fine, but a couple of them needed replacement heatsink and fan combinations -- which I keep around just in case.
I also got one 4870 from NewEgg -- but for whatever reason, that puppy runs GPU BOINC applications VERY slow. I just swapped out a re-fanned 4850 there.
In any event, after my summer system shutdowns are over this coming fall, I may look into getting the next series up. What 5xxx AMD cards run double precision -- I know their numbering scheme got a bit different (ie a 5850 might well not be double precision).
Barry maybe EBay will become your friend and help you unload those cards and get newer, faster ones! Just remember EVERYONE on EBay is NOT honest!!!
/ Reply
Quote
I've got an AMD FX-8150 CPU, is there anything I need to modify to receive the new WU's?
- AVX will be used if available on Linux
I'm using BOINC 7.0.15 in OpenSUSE 12.1
/ Reply
Quote
5850 is double precision.
____________
/ Reply
Quote
In any event, after my summer system shutdowns are over this coming fall, I may look into getting the next series up. What 5xxx AMD cards run double precision -- I know their numbering scheme got a bit different (ie a 5850 might well not be double precision).
The 58xx series does have double precision, it is the 68xx series that does not have it.
We are still waiting to see what happens with the 78xx series.
____________
/ Reply
Quote
My ATI 4850 is not processing the new work units. In an attempt to make the suggested upgrade, my 4850 is now not even processing the Collatz projects that it was processing fine. My machine is running an ATI 3300, 4850 and Nvidia 550. I am afraid I am returning more errors now than work.
/ Reply
Quote
Mark, the combination I've found that works for me with 4850's.
1) AMD video1 1.8
2) BOINC 6.10.58
This works with both Collatz and MW -- as well as MooWrapper.
I've used this with XP, Win7 32 and Win7 64 bit setups.
Also, after the software versions are in shape, I'd detatch and reattach the projects.
My ATI 4850 is not processing the new work units. In an attempt to make the suggested upgrade, my 4850 is now not even processing the Collatz projects that it was processing fine. My machine is running an ATI 3300, 4850 and Nvidia 550. I am afraid I am returning more errors now than work.
/ Reply
Quote
Thanks - good to know for when I make the move. Does the 5850 require a higher power supply profile than the 4850?
The 58xx series does have double precision, it is the 68xx series that does not have it.
We are still waiting to see what happens with the 78xx series.
/ Reply
Quote
I know my 5830 needs 2 pci-e power connectors, but I think the 4830 only needs one.
____________
/ Reply
Quote
The Guru3D 5850 review:
http://www.guru3d.com/article/radeon-hd-5850-review-crossfire/
Regards
Zy
/ Reply
Quote
Right, the 4870 requires two leads, the 4850's only one.
I know my 5830 needs 2 pci-e power connectors, but I think the 4830 only needs one.
/ Reply
Quote
What command would be added to the the app-info file to throttle back GPU usage? It's currently at 99% and I'd like to pull it back to 80-85%.
Win7 64 bit. Radeon 5830.
With the new app you don't need the app_info-file to control GPU load - it's now only necessary if you want to run multiple instances on one GPU. If you want to use it anyway the command line parameter apparently wasn't changed from the old app:
--gpu-target-frequency XX
Raising XX lessens the GPU load and increases the run time (default is now 60).
With the new app it's easier controlled in the Milkyway@home preferences: Frequency (in Hz) that should try to complete individual work chunks
For some reason the frequency setting in the M@H preferences is ignored. The GPU runs @ 99% no matter what the setting. I tried a target frequency in the app-info file with separation 1.02 and it caused the GPU frequency and power draw to be all over the place. Had to switch back to 0.82 to get things running smoothly again. Only thing I can figure is it must be some kind of a hardware thing on my setup.
____________
/ Reply
Quote
I found that once I had everything set to go, I needed to just leave it alone. The WUs came in one at a time and it took a few minutes to get the next one. I had to force myself not to babysit and manually update Boinc trying to get more work every time it ran out. I let it go overnight with no other GPU projects running. By morning it was running just as before.
That worked for me too, all 4 of my 5870's now have lots of work!
____________
/ Reply
Quote
Results
Config:
HD6950, unlocked shaders (HD6970), 900 MHz GPU, 625 MHz memory, Win 7 64, i7 2600K @ 4.0 GHz, BOINC 6.12.34
Before:
0.82 CAL + app_info, 1 WU at a time, parameters: --process-priority 3 --gpu-disable-checkpointing
Times: 53 - 54s GPU, 2.4s CPU
Now:
1.02 OpenCL, no app_info, using AVX code path
default: 56s GPU, 3.3s CPU
setting refrseh rate to 10 Hz: 55s GPU, ~2.9s
setting refrseh rate to 1* Hz: 55s GPU, ~2.9s
That's about 3.5% slower. Previously the error rate was ~0.7%, so if these were fixed now (can't tell yet) overall throughput would be less than 3% lower. That's about 7800 credits/day less.. but running without an app_info if nicer.
MrS
(*) No worries, the IGP is driving my display ;)
____________
Scanning for our furry friends since Jan 2002
/ Reply
Quote
BOINC 6.10.58 appears to be for Mac OS X. Is it safe to run in windows?
/ Reply
Quote
Been running 6.10.58 for a long time on Win before I switched to 6.12.
MrS
____________
Scanning for our furry friends since Jan 2002
/ Reply
Quote
For some reason the frequency setting in the M@H preferences is ignored. The GPU runs @ 99% no matter what the setting. I tried a target frequency in the app-info file with separation 1.02 and it caused the GPU frequency and power draw to be all over the place. Had to switch back to 0.82 to get things running smoothly again. Only thing I can figure is it must be some kind of a hardware thing on my setup.
Strange
I'm running now without app_info and the default frequency of 60Hz -> GPU load 82-85%. Before I removed the app_info I was running with "--gpu-target-frequency 40". I left it in when I switched from 0.82 to 1.00. According to the results it was used and run times where lower than now (with 60Hz).
Maybe the varying GPUs react differently to the setting.
Edit: A quick test changing the web preferences to 40Hz: result
The setting change was recognized by the app -> GPU load 90-91%.
____________
Starfire
/ Reply
Quote
Thanks, I'll give it a try.
/ Reply
Quote
Why is your CPU load so low? I'm leaving one logical core free on my i7 and that nails GPU utilization at 99%. Otherwise performance suffered too much for my taste (0.82).
MrS
____________
Scanning for our furry friends since Jan 2002
/ Reply
Quote
Ok, it has been a long road switching from mw v0.82 with cat 11.3 to mw v1.02 with cat 12.1. Done several tests on the way.
(The driver updates gave several hickups, needing manual cleanup after uninstall and cleaner program before installing the next version.)
System is WinXP 32bit with HD5850 @775MHz
(last tweaking of app settings only days ago)
v0.82 CAL, cat 11.3: ~84.5s
v0.82 CAL, cat 11.8: ~84.5s (reasonable higher system kernel times, less responsive)
v0.82 CAL, cat 11.9: ~92.5s (system kernel times and response a little better)
v0.82 CAL, cat 12.1: ~84.5s (system kernel times high, system response bad)
v0.82 CAL, cat 12.1: ~85.3s (relaxed polling to make system responsive again)
For mw v0.82 CAL, cat 12.1 is slightly slower than 11.3 but far better than 11.9.
For collatz and moo I have the impression cat 12.1 is a little faster than 11.3 but no numbers to verify it. Cat 11.8 and 11.9 did _not_ impress there too.
v1.02 OpenCL with default params showed high system kernel use, polling 0 dropped gpu use to 90%, polling 1 to ~83%. So I am using polling 0 with a far reduced wait factor for now. Still thinking there must be a wait factor + polling > 0 but couldn't find one without loosing bad on gpu load. Need to play more with the params to get a better understanding how they work together in this new version. The actual setting has to do for now.
v1.02 OpenCL, cat 11.9: ~81.3s (cpu time includes system kernel times here!)
v1.02 OpenCL, cat 12.1: ~80.9s (system kernel times hidden again)
Cat 11.9 showed a high system kernel use, that was 'just' to get under control with setting command line params. With cat 12.1 and the same settings the system kernel times are roughly cut into half, so it gives a little more comfort related to system response. I think I got pretty close (within less than 1s/WU) to the best time possible before system response goes downhill again.
/ Reply
Quote
Along these lines: I discovered that while futzing around with Catalyst and boinc 7.08 that my online boinc account thinks I have 40 wu's in progress that I don't have in boinc manager. Pretty sure that's why I'm not getting wu's. Any solution besides waiting till 2/23 for the units to time out?
/ Reply
Quote
Along these lines: I discovered that while futzing around with Catalyst and boinc 7.08 that my online boinc account thinks I have 40 wu's in progress that I don't have in boinc manager. Pretty sure that's why I'm not getting wu's. Any solution besides waiting till 2/23 for the units to time out?
Detach and reattach would be the quickest method.
____________
/ Reply
Quote
My error rate is down to ~0.1% from a starting value of 0.7 - 0.8%.
Good job here:
The occasional validate errors from empty / truncated stderr should stop
Edit: using the same command line parameters as with 0.82 I'm still seeing the same crunching speed as reported above.
MrS
____________
Scanning for our furry friends since Jan 2002
/ Reply
Quote
I've updated all of the separation applications to 1.00. For changes people might care about,- The old CAL version is gone; it's replaced with the OpenCL application. On AMD/ATI GPUs (older than 79xx) it is using some hackery to use the same IL kernel as before so it should be as fast. However this also means the Radeon 38xx cards aren't supported with new stuff.
- Radeon 79xx stuff should work
- The occasional validate errors from empty / truncated stderr should stop
- AVX will be used if available on Linux and Windows (64 bit only for Windows)
- I've increased the default GPU target frequency so GPU stuff should make things less laggy on average. You can now also configure this with the web preferences now so you don't need to use app_info stuff if you want to play with that.
- Partial workaround with high CPU usage with recent Nvidia drivers.*
As usual post problems you run into here.
* It should cut down on the CPU usage a bit while not sacrificing too much. I would recommend not using it unless you are very unhappy with the CPU usage on Nvidia. There are options to change the polling mode if you want to lower the CPU usage further while not slowing it down. (--gpu-wait-factor (default = 0.75) and --gpu-polling-mode (default = - 2) work similarly to how they did with the old CAL one, but slightly different). With the default of -2 it will use mode -1 unless it is an Nvidia driver newer than the one that introduced the high CPU issue, where it will use mode 0. Mode -1 uses the correct waiting method, mode 0 use the correct waiting method with an initial sleep based on time estimates, and modes > 0 are a polling period in milliseconds. The wait factor is a sort of correction of the time estimate used for the initial wait. The default is 0.75, to wait for 75% of the estimated time before trying to poll.
Well Ok I see someone else has gone all Beta or maybe that should be Alpha, I'm getting 97-98% cpu use(295) and about 78%(285) while doing a gpu app. Help. I run Windows 7 x64b sp2 w/16GB ram, My driver is 275.50 x64(If I use an older gpu driver then Boinc doesn't see any usable gpu cards, 280.xx produce reboot bsods and that leaves 290.53, 291.51 causes problems someone said over at S@H and it's been reported to Nvidia), I have a GTX285 and a GTX295 in this PC and no other PCs. I use Boinc 6.10.58 x64.
Frequency is set at 120, tried 60 that didn't seem to help or am I missing something?
Maximum CPU % for graphics 0 ... 100: 0
____________
12/21/2012? Bah Humbug!
/ Reply
Quote
[quoteWell Ok I see someone else has gone all Beta or maybe that should be Alpha, I'm getting 97-98% cpu use(295) and about 78%(285) while doing a gpu app.[/quote]It's yet another Nvidia driver problem introduced in 270.something
/ Reply
Quote
Well Ok I see someone else has gone all Beta or maybe that should be Alpha, I'm getting 97-98% cpu use(295) and about 78%(285) while doing a gpu app.It's yet another Nvidia driver problem introduced in 270.something
Sorry, but My experience with Open CL is all beta so far, so 27x.xx is a problem, would 290.53 do better? I won't touch 28x.xx as that only gave Me reboot Bsods. I've tried 266.58 with Boinc not seeing any cards. An AP Nvidia app over at S@H is only usable with 266.58 or earlier and using It with 27x.xx or newer caused driver crashes as It was only certified to with pre 27x.xx drivers so far and it's an Open CL app too. Otherwise CAL is all that will work for Me and I think You've possibly abandoned that, unless I'm wrong on that, Open CL seems to have some teething problems and the latest 295.51 beta driver is not good as it seems to have some serious problems I've read, I think Arky can confirm this.
____________
12/21/2012? Bah Humbug!
/ Reply
Quote
I haven't tried 290 anything yet
/ Reply
Quote
Running 2 at a time on both cards with 290.53, I averaged around 790 seconds or so.
http://milkyway.cs.rpi.edu/milkyway/results.php?hostid=344435
That is 1 GTX460/768 and 1 GTX560/1024.
____________
/ Reply
Quote
Running 2 at a time on both cards with 290.53, I averaged around 790 seconds or so.
http://milkyway.cs.rpi.edu/milkyway/results.php?hostid=344435
That is 1 GTX460/768 and 1 GTX560/1024.
When I get My GTX460 2win card, or maybe I'll go to 290.53 later on as it's on the hdd here(before the card gets here), I'm moving away from pre fermi chips, I have one more GTX460 to win today, It's a PNY card w/1GB and It's a reference board, which means unlike the 2win card, the PNY can be water cooled, aquacomputers said if there was enough demand for a water block they'd make one for the EVGA GTX460 2win card.
____________
12/21/2012? Bah Humbug!
/ Reply
Quote
Update: error rate down to 0 out of 3350+ with the new app. Yay!
MrS
____________
Scanning for our furry friends since Jan 2002
/ Reply
Quote
I've updated all of the separation applications to 1.00. For changes people might care about,- AVX will be used if available on Linux and Windows (64 bit only for Windows)
Apparently no AVX with Bulldozer and Windows 8 Developer Preview x64. Anyone using an Intel CPU w/ Windows 8 Dev. Preview x64 or a Bulldozer w/ Windows 8 Consumer Preview x64? Is AVX used then?
I compared my PC with two of the top 20 computers, all three currently running MilkyWay@Home v1.02 (opencl_amd_ati):
My PC #447441, AMD FX-8120, Windows 8 DP x64, BOINC 7.0.25 (x64), AMD Radeon HD 79x0 series (Tahiti) (3072MB) driver: 1.4.1720 -> Using SSE4.1 path
#365115, Intel i7-3960X, Windows 7 Prof. x64 SP1, BOINC 7.0.27 (x64), [4] AMD Radeon HD 6900 series (Cayman) (2048MB) driver: 1.4.1720 -> Using AVX path
#340810, AMD FX-8120, Windows 7 Home Premium x64 SP1, BOINC 6.12.34 (x64), [4] AMD Radeon HD 6900 series (Cayman) (2048MB) driver: 1.4.1607 -> Using AVX path
Regards
/ Reply
Quote
computation error 5/31/12 9:43.21 through 5/31/12 11;14/38
____________
computation error
5/31/12 21:43:27 through
5/31/12 23:14:38
/ Reply
Quote