Welcome to MilkyWay@home

Posts by Joseph Stateson

1) Message boards : News : Validator Outage (Message 71244)
Posted 3 days ago by ProfileJoseph Stateson
Post:
These "Invalids" are really becoming irritating. I am encountering about 450 per day and each is consuming 1.625 of the normal computer time. Are we making any progress on eliminating the 7 WUs tasks?
I won't even bitch about the "Error While Computing " Tasks. They are not Errors While Computing since no computing is ever done. Actually, they are Initialization Errors. I don't care what they are I just want them to go away.
On the Invalids, I am fairly certain that tasks that end up as 7 WU are sent as 4 WU. The assessment 7 WUs is made by the clients' computers. Maybe what we need is a subroutine in Initialization that tests the WU Count before computation starts and if it is not 4 aborts the run. I don't like creating "workarounds" as fixes but it would be better than we have today.


The separation source file at "milkywayathome_client/separation/separation_main.c"

Has the following code line: "mw_printf("<number_WUs> %d </number_WUs>\n", ap.totalWUs);"

putting the following under it would abort all 7 parameter work units
if (ap.totalWUs == 7)
{
exit(EXIT_FAILURE)
}

This would avoid crunching 7 WU tasks but you would end up with a lot of "error" which might cause the daily quota to be exceeded. The number of GPUs can always be faked to get more but I suspect it is better if the project guru can fix the problem

It has been over a year since I last built the client
https://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=4551#69402

It appears there have been a few changes since then.
https://github.com/Milkyway-at-home/milkywayathome_client
2) Message boards : Number crunching : Finally getting new tasks only seconds after running out. May not be worth the hassle. (Message 71197)
Posted 16 days ago by ProfileJoseph Stateson
Post:
For me this time out script does not work as I can make a WU in less than 92 sec so each time it runs it reports the completed tasks and then does not get anymore.

Also, if I disable the network for 15 min, then report it the same sort of deal.

It seems like its only possible to get task in batches of 300 with the current setup of the project


The only script that I know of that work is listed here
https://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=4532&postid=69286#69286

You can try my program "7.15.0" mentioned further down this thread but you would have to give up 7.16.11 "the latest and greatest" as my app is a mod of 7.14.dontremember

The problem (feature ?) is with the Milkyway server not the SETI client or manager.
3) Message boards : Number crunching : Problem with scheduler request (Message 71188)
Posted 16 days ago by ProfileJoseph Stateson
Post:
It looks like not many users have left. General increase in user count since middle of August. About 20 new users each day according to the stats sites.


I have started moving my "Einstein" racks over here now that the weather has cooled. When SETI was functioning there was no problem mining GRC at Einstein. Ever since you and Steve moved from SETI things have turned down for me there!. l can never catch up with you two over at Einstein!

4) Message boards : Number crunching : Problem with scheduler request (Message 71171)
Posted 18 days ago by ProfileJoseph Stateson
Post:
That problem (or feature) has been around for a long time. I think the only workaround is to specified "no new work" before doing the update and then resume work
5) Message boards : Number crunching : Run Multiple WU's on Your GPU (Message 71158)
Posted 20 days ago by ProfileJoseph Stateson
Post:
No matter how many WU's I run my card won't use more than 210w when crunching, 2 / 4 / 8 units, its a 3090 and under WCG running 8 of their GPU units at the same time it will pull the full 350w,



With the price of graphic cards skyrocketing, I would avoid running any card at its max rating. High temps cause thermal paste to harden making removal non-trivial. Exact OEM fan replacement can be hard to find and one must be creative on occasions. I have a few really weird fan arrangements I can post if anyone interested.


BUT the problem comes down to MW not being able to supply you with enough tasks in one day to keep it going, with their 10 minutes back-off between sending tasks your 3090 will be doing something else for those 10 minutes even more often.


That BOINC app I modded, 7.15.0 fixes the 10 minute wait. The latest official version is 7l.16.11 and I assume the 10 minute problem still exists for that app.

Due to temperatures recently dropping here in texas, I started up a pair of garage "racks" to start crunching on Einstein and WCG. I have a 3rd rack for milkyway but the garage is still too hot to run that one.
6) Message boards : Cafe MilkyWay : Crazy GPU prices (Message 70748)
Posted 21 Apr 2021 by ProfileJoseph Stateson
Post:
I just sent pair of rx560 and a single rx570 to my son in germany to sell for me. I should get 2x what I paid for them. If it works out I will unload some gtx1060s over there. I also have two more rx570 factory refurbished what should pay good
7) Message boards : Number crunching : AMD FirePro S9150 (Message 69762)
Posted 27 Apr 2020 by ProfileJoseph Stateson
Post:
I have a G12 coming this week to try on my S9100 with an Arctic Liquid Freezer 120 that I had on my 2600k CPU before I upgraded to a 3900X and new cooler. Curious as to how it's going to work. Wish I knew if 7950 or 280X coolers fit this card.

I'll have to try that 20q1 Pro driver as well as you seem to have it working well. Getting full speeds with it as well I hope.


non turbo coolers from "parts only" 7950 work fine with S9000 but NOT the 9100 which has a different chip: Hawaii.

The s9100 has a larger rectangular chip, not a diamond shaped chip (the layout on the board). I suspect you will need a copper shim. I do not want to take mine apart but I remember trying to use an Nzxt with a cpu cooler and had to get a copper shim as the chip was recessed. Not positive, maybe it was the s9000 that needed the shim. PM me your address and I can mail you two different shims if location is USA.

I recommend the following, a Dell part that I got from North America for S9050
https://tinyurl.com/yau7lesf

you can google for BG0903-B047-VTL and I think Newegg also carries it. The shroud fits the fins of the cooler perfectly and is easily held on with aluminum tape. Only works on an open air mining rig due to the length. Try to get one that is not from china as not economically returnable if defective. One of two I bought from Newegg was DOA, obviously used. Both were 3rd party from china.
8) Message boards : Number crunching : Finally getting new tasks only seconds after running out. May not be worth the hassle. (Message 69721)
Posted 16 Apr 2020 by ProfileJoseph Stateson
Post:

Actually, mine are Tahiti. Are you thinking of the 290X? Mine are 280X.
And checking on https://www.techpowerup.com/gpu-specs/firepro-s9100.c2636 it's a different version of Hawaii GPU anyway. So perhaps the insides are indeed different.

And S9100! Wow, they cost a lot, even second hand. Two 280X is a fifth of the price of one S9100, and does the same DP and twice the SP.


I have a single 9100 that I got used about 175 as I recall. However, boinc thinks I have all s9100 due to they way they report only one gpu.

The x280 is superior to all my "S's": s9050 (new $69) and s9000 (70 - 90) of which I have a mix and they have only 1792 cores unlike the x280 which has a full 2048 and better FP and DP performance. However, all of my boards have a single 8 pin power connector and run nowhere near the 225watt TP even with 5 concurrent tasks. However, Einstein runs hotter and tasks take longer on these boards than, for example a low power (6 pin) gtx1060

I can do 5 concurrent Milkyway tasks at about 39-41 seconds (click calculate button) each task but a Einstein runs 12 minutes on s9050 and under 10 on gtx1060 but cooler.

Shut down most of my rigs last month when texas weather got hot but a strange cold front in april allowed me to turn some back on.
9) Message boards : Number crunching : Finally getting new tasks only seconds after running out. May not be worth the hassle. (Message 69709)
Posted 14 Apr 2020 by ProfileJoseph Stateson
Post:
I am running Milkyway@home Separation on my GPU....

The "NVIDIA GPU task request deferred for 00:0x:xx" in conjunction with "NVIDIA GPU task request deferral interval for 00:10:00" is getting painfully problematic as it prevents me from downloading new work. Whenever I return tasks to the server this deferral gets reset and my computer gets no new tasks. On my GTX 1660 Ti i have gone to running four tasks simultaneously to increase the time it takes to send back to the server, now at around 12 minutes, still this is not enough as too many results are still returned and the deferral is yet again postponed. I have even tried settings to grab a larger number of tasks, but instead of helping it ended up messing up my other projects.

I don't want to download any special version of boinc as this is not my primary project. Maybe I should be running more than four tasks? ...although the four are already causing havoc with cpu loads... In any case MW's failure to keep my computer continually loaded is a boon to my other projects which are more than happy to fill a 10 minute gap with hours of work.


Yes, I ran Einstein tasks by setting its resources to 0.0. That caused Milkyway to get it full load of 900 work units and allowed Einstein to get a few when Milkyway was taking its 10 minute siesta. However the Einstein ran poorly on my AMD boards and was a waste of resources. Your 1660ti will handle Einstein or gpugrid nicely unlike my s9000 series boards.
10) Message boards : Number crunching : Finally getting new tasks only seconds after running out. May not be worth the hassle. (Message 69705)
Posted 14 Apr 2020 by ProfileJoseph Stateson
Post:
You might want to PM him and ask him to build his client out of the current master branch. You do want the latest branch because one of the fixes that it has is the fix for the "finish file present too long" error.
That really is the only part of the new client that is worth updating for.


Just saw this and no, my fix does not include the "finish file" solution. I will take a look at including that in my app and post in a couple of days hopefully. Be nice if there was a moderator who could make a sticky post for where to download and how to install this Windows & Linux app. Note that there is a script that works with the official release somewhere down this same thread but it is a PITA to implement.
11) Message boards : Number crunching : Finally getting new tasks only seconds after running out. May not be worth the hassle. (Message 69696)
Posted 12 Apr 2020 by ProfileJoseph Stateson
Post:
Seems to be only one developer / moderator here and I see him posting on news only.


I have sent a message to Eric, we shall see.... It can't be that hard to fix it!

There was an offer made months ago to look a the config files but no one replied. I consider "no reply" to be better than the rant at Einstein that an offer to improve the code performance is childish.

Well.... if I had invested a lot of money in some research, I'd not be too keen on others seeing the source code. Mind you, isn't there something called reverse engineering?


Pretty sure the C code that allows users to contribute their processing power is GPL'd. I am thinking that if the moderator pisses off enough volunteers someone will clue in the GPL license enforcement people. No one wants to have their spaghetti code and "FIXME FIXME" comments exposed to the light of day. The ClimateGate programmer code comments such as "[projected temperatures] ... past 1960 ... will be artificially adjusted to look closer to the real temperatures" caused scientists to be more carefully about releasing code to the public.
12) Message boards : Number crunching : Finally getting new tasks only seconds after running out. May not be worth the hassle. (Message 69694)
Posted 12 Apr 2020 by ProfileJoseph Stateson
Post:
Can't somebody give the project admin a shove and just get him to change the config on the server? IT'S SET UP WRONG. I will not be wasting my time messing around with special configurations to get round their problem. If Milkyway doesn't hand out work when it's supposed to, my GPUs run Einstein, which gives piles of work at once, so Milkyway hardly ever gets a shot on the GPUs. FIX IT. NOW.


Seems to be only one developer / moderator here and I see him posting on news only.

There was an offer made months ago to look a the config files but no one replied. I consider "no reply" to be better than the rant at Einstein that an offer to improve the code performance is childish.

I too would do some other project but I have an investment in double precision GPUs that run very poorly on Einstein but on Milkyway can run circles about nVidia boards.
13) Questions and Answers : Windows : Milkyway@Home Uses Only One of Three GPUs (Message 69651)
Posted 5 Apr 2020 by ProfileJoseph Stateson
Post:
Thank you for the response. I use the cc_config use all gpus on all my computers. I checked and it is there in ProgramData file. I tried updating the NVIDIA drivers and it didn't help. So, I.m still chasing my problem.


Possibly the problem is how the server fills the queue. When the last tasks completes there is usually a 10 minute wait before more tasks download. This is a well known problem with 2 solutions as discussed here https://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=4532
However, your problem seems different. I assume you checked the server status and there are jobs?
14) Message boards : Number crunching : running GPU tasks and want to run CPU tasks also? (Message 69595)
Posted 12 Mar 2020 by ProfileJoseph Stateson
Post:
Consider running LHC "atlas" It can consume up to 8 GPUs. It requires VBox which can be downloaded from Oracle. Version 6.x works fine, no need to reinstall "boinc+vbox". Image files run about 1gb so downloads can take a while on slow systems. Be sure to read this post first
https://lhcathome.cern.ch/lhcathome/forum_thread.php?id=5348&postid=41896#41896
The memory requirement of 3gb ram + 0.9gb for each CPU includes virtual ram so a 12gb system can easily handle an 8cpu job.

Warning; my app, the "7.15.0" one that fixes the 10 minute delay does not work when CPU Milkyway tasks are being crunched in addition to GPU ones. One should split up GPU and CPU tasks to different projects anyway.
15) Message boards : Number crunching : Not getting tasks, even though the server has 10000 available. (Message 69594)
Posted 12 Mar 2020 by ProfileJoseph Stateson
Post:

Not getting tasks, even though the server has 10000 available.


Well, my local bank has millions of euros available, but they aint gonna give'm to me.
I guess that's life, usually we dont get what we want, I'm used to it by now, i just ignore it and move on.


Just went through this with LHC. They had plenty of atlas work units, most 1gb in size (VBox image) but only my Linux system could get work. Turned out the checkbox "Run native if available" should have been worded "Don't run anything except native code" After fixing the problem I got 5 huge withdrawals.
The 1gb+ withdrawals from the :"bank" on two system tied up my network for 20 minutes.

Atlas tasks can take up to 8 cpus which is nice to use with Milkyway GPU tasks.
16) Message boards : Number crunching : Delay in getting new work units untill all work units have cleared (Message 69584)
Posted 7 Mar 2020 by ProfileJoseph Stateson
Post:
Should have mentioned this earlier: The Milkyway "fix" I put together here was built with VS2013 and requires VS2013 runtime.
https://www.microsoft.com/en-us/download/details.aspx?id=40784

It will not work with the newer runtimes VS2015..VS2019. Just found out as I tried using my program on a win7 system and it did not work and there were no error messages. If you want to use my program to avoid the 10-20 minute delay then for a new install put in 7.14.2 or the newer 7.16.x and then install that VS2013 runtime. The VS2015..VS2019 runtimes did not work.

The Linux version does not have any special requirements other than setting attributes and ownerships as explained here

https://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=4532&postid=69562#69562

VietOZ has scripts that essentially do the same thing and no need to replace the boinc client.

https://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=4532&postid=69283#69283
17) Message boards : Number crunching : Bogus AMD FX9590 CPU (Message 69580)
Posted 5 Mar 2020 by ProfileJoseph Stateson
Post:
I sympathize with you. Some time ago I picked up an HP Pro 3400 mombo, $20. It would not work with a single gpu that I had and I have a lot of old gpus as well as new ones. No bios upgrade and a search of HP came up with missing links for all driver downloads.
18) Questions and Answers : Unix/Linux : CL_DEVICE_NOT_FOUND / MW_CL_ERROR on two vega64 (Message 69565)
Posted 24 Feb 2020 by ProfileJoseph Stateson
Post:
OS Manjaro, Kernel 5.3 rc2 ROCm 2.6.


I suspect the problem is the ROCm driver. There is a long discussion of this driver problem over at Einstein and they provided an update to their app.

I have no interest in testing this theory out as I lost all but one RX-570 boards when I enabled the ROCm driver on my 18.04 box.
19) Message boards : Number crunching : Finally getting new tasks only seconds after running out. May not be worth the hassle. (Message 69562)
Posted 22 Feb 2020 by ProfileJoseph Stateson
Post:
Hi I see you guys mentioning editing "coproc" or something like that to download 900 WUs at a time instead of 150 - 300? I was wondering how you go about doing that, because I've been searching around for a bit and I can't find any instructions.

Also how do you install the ubuntu version? I'm a complete novice with Linux and you mention you need to "use 0751 on program and 0664 on the xml" but I have no idea what that means.


Milkyway allows 300 downloads per GPU with a max of 900. There is no need to get 900 or 9000. The problem is when you run out of data there is a 10 minutes wait before anymore arrives which is a bummer.

I am not sure if the boinc_linux program will work on your version of Linux. I had forgotten I had even made a Linux version. Before doing anything, make a backup of the boinc stuff you got. I assume you are using 7.2.14. Maybe 7.16.3 will work

download all the linux files to a download directory such as
/home/username/Download

the executable goes to /usr/bin
the config goes to /etc/boinc-client
do something like the following using a terminal window and adding "sudo" in front of commands that complain
and fixing any typos or adding anything I might have left out that is needed.

sudo su
/etc/init.d/boinc-client stop
mkdir Download
cd Download
mkdir mw_fix
cd mw_fix
wget https://github.com/JStateson/MilkywayNewWork/archive/master.zip
unzip ./master.zip
chmod 0755 ./boinc_ubuntu
chown root:root ./boinc_ubuntu
chmod 0644 ./cc_config.xml
mv /usr/bin/boinc /usr/bin/boinc_original
mv /etc/boinc-client/cc_config.xml /etc/boinc-client/cc_config.xml.bu
cp ./boinc_ubuntu /usr/bin/boinc
cp ./cc_config.xml /etc/boinc-client/cc_config.xml

/etc/init.d/boinc-client start

within a 5-8 minutes you should see Milkyway download some addtional work and the number of "waiting" work units should hover just below the maximum (300, 600 or 900). Should never go to 0 again unless the project goes off-line.

in the file cc_config.xml, if you edit it for:
<mw_debug>1</mw_debug>
then look at event viewer for messages about the bug fix for debugging purposes
20) Message boards : Number crunching : AMD FirePro S9150 (Message 69560)
Posted 21 Feb 2020 by ProfileJoseph Stateson
Post:
I just switched to that new 20q1 "Enterprise"

"02/21/2020 09:52 AM 463,754,128 win10-radeon-pro-software-enterprise-20.q1-feb10.exe"

and seems all is working ok for Einstein. Running some tests for that project so will check out Milkyway later

OpenCL: AMD/ATI GPU 0: AMD FirePro S9000 (driver version 3004.4, device version OpenCL 1.2 AMD-APP (3004.4), 6144MB, 6144MB available, 3226 GFLOPS peak)	
OpenCL: AMD/ATI GPU 1: AMD FirePro S9000 (driver version 3004.4, device version OpenCL 1.2 AMD-APP (3004.4), 6144MB, 6144MB available, 3226 GFLOPS peak)	
OpenCL: AMD/ATI GPU 2: AMD FirePro S9000 (driver version 3004.4, device version OpenCL 1.2 AMD-APP (3004.4), 6144MB, 6144MB available, 3226 GFLOPS peak)	
OpenCL: AMD/ATI GPU 3: AMD FirePro S9000 (driver version 3004.4, device version OpenCL 1.2 AMD-APP (3004.4), 6144MB, 6144MB available, 3226 GFLOPS peak)	
OpenCL: AMD/ATI GPU 4: AMD FirePro S9100 (driver version 3004.4, device version OpenCL 2.0 AMD-APP (3004.4), 12288MB, 12288MB available, 4608 GFLOPS peak)	
OpenCL: AMD/ATI GPU 5: AMD FirePro S9000 (driver version 3004.4, device version OpenCL 1.2 AMD-APP (3004.4), 6144MB, 6144MB available, 3226 GFLOPS peak)


I enabled ECC for all boards.

Going to pull the S9100 for a test of the RocM AMD drivers on my Linux platform. Not sure if those drivers work with Milkyway so I am testing on Einstein who have a beta for RocM.


Next 20

©2021 Astroinformatics Group