Testing ATI Application Availability

It executes correctly for me by manually downloading the files and manually executing the executable, does that error code mean the wrong dlls were found?

It executes correctly for me by manually downloading the files and manually executing the executable, does that error code mean the wrong dlls were found?

It's one of the codes meaning that the application failed to initialize. That normally happens if a dll is not found. As it was running before with an app_info.xml (Win32_SSE2_ati), I would assume the driver is installed correctly. I also checked the Win32 version you put on the server and it is linked against the correct dlls, so no problem there.

The only explanation I can come up with, is that you maybe forgot to set the destination name to "brook32.dll" as the name on the server is brook32_ati.dll. At least if Michael Phillips didn't messed around with dll files itself.

yes your correct
it's a vista 32 bit with catalyst control 9.12
but what can i do to solve it or is it a server side issue, should i just go back to app_info for now?

If I remove the app_info file will it trash the cached wu's?

yes your correct
it's a vista 32 bit with catalyst control 9.12
but what can i do to solve it or is it a server side issue, should i just go back to app_info for now?

If you feel up to the task for some debugging can you:
1. look in C:\WINDOWS\system32 for the dll files
aticalcl32.dll
aticaldd32.dll
aticalrt32.dll
2. In the BOINC application directory (C:\Documents and Settings\All Users\Application Data\BOINC for me), there should be a directory called 'slots' and within it subdirectories named 0,1,2,etc. Please verify that within the sub directories if there is a program named milkyway_0.21_windows_intelx86__ati13ati.exe that there is also a file named brook32.dll. If no sub directories are in the 'slots' folder please start a WU without the anonymous platform and once it starts immediately suspend activity, after that hopefully one of the 'slots' sub directories stays around.

i have solved the problem and feel a bit daft now
when i removed app_info.xml brook32.dll and astronomy 0.20b app from the boinc app data directory it failed to remove brook.dll file claimed it was still in use. I did this earlier today but forgot about it, now after downloading the new app via boinc it failed due to the old brook32.dll i left behind. a simple reboot and i could remove the old brook32.dll and problem solved.
sorry for wasting your time, should have remembered this earlier.
thanks for the help anyway.
michael P.

It must be the b-1 tag I was using then as I have slowed down quite a bit.

Okay, I will probably go back to the app_info and turn on CPU usage.

It is weird that one little setting accounts for a 40 second improvement.

27214710 25885832 306.03 303.86 213.76 Anonymous platform
27061134 25742898 347.52 345.78 213.76 MilkyWay@Home v0.21 (ati13ati)

This must be the only little thing, ["setting GPU polling behaviour: busy wait (fastest, potentially high CPU load)"]. It is telling the GPU to just go.

---

As a complete novice at all this GPU stuff, I want some really basic questions answered.

I have a Radeon X1300/X1550 series card (DELL 1907FP) identified as device ID 7183 and Catalyst version 9.11 running under Vista Home Premium (32 bit) on an AMD Athlon 64 X2 Dual Core processor 4200+ 2.20 GHz

1. Can I run this new app alongside my other projects and my CPU stuff?
2. If so, what are the steps I need to take?

I have done all I thought was necessary - ticked the use GPU options, etc and reset the project a couple of times to try and get it to download the ATI app, but to no avail. Should I untick the use NVIDIA if present option?

Thanks

Michael

Unfortunately, your video card is sufficiently old enough that it does not have the capability to crunch for any project.

Minimum for Milkyway is a 3800 series, 4770, 4800 series and 5800 series.

---

So why is the app at 0.38 cpu's and taking about 25 seconds longer on my 4870?

So why is the app at 0.38 cpu's and taking about 25 seconds longer on my 4870?

Did you use the b-1 tag in your app_info?

I am back to using a app_info as I did not like that slower crunching, of course I also had to change my prefs to use CPU as it will not get any work when it asks for GPU tasks.

---

So why is the app at 0.38 cpu's and taking about 25 seconds longer on my 4870?

Did you use the b-1 tag in your app_info?

I am back to using a app_info as I did not like that slower crunching, of course I also had to change my prefs to use CPU as it will not get any work when it asks for GPU tasks.

Yes, sux, doesn't it.
And there is no indication whatsoever that this might get fixed anytime (soon? lol). Anyway, it's time for me, to dust off my almost forgotten "non DA power BOINC" CC plans and bring them into existence. The main features I intend to implement are:
- independent resource/project pairing allowing user to specify exactly which resources are used by each attached project
- independent resource shares by resources (now a resource share is global for all types of resources)
- true backup projects (now it is only possible to approximate this functionality using very low resource shares for "would be" backup projects, which in combination with current flaky scheduler - not to mention the most idiotic "workaround" ever, GPU FIFO execution rule - just isn't functioning properly)
- benchmarking only in case of resource change (new cpu, gpu), or in similar relevant cases

I'm still in design/pseudocode phase, so it is still uncertain if this will require writing client/manager from scratch, depending on usability of current code. If anyone is interested and/or willing to help, PM me.

[edit]Made the subject of my reply bold[/edit]
BR,

---

So why is the app at 0.38 cpu's and taking about 25 seconds longer on my 4870?

Did you use the b-1 tag in your app_info?

I am back to using a app_info as I did not like that slower crunching, of course I also had to change my prefs to use CPU as it will not get any work when it asks for GPU tasks.

I also use the b-1 option in the app_info.xml to decrease WU time and have no problem getting WUs with "use CPU" set to no. I notice you're using BOINC v6.10.21. Try BOINC v6.10.25 and see if that fixes the problem. Please post the result.


My cc_config.xml also contains the following:

<cc_config>
<options>
<report_results_immediately>1</report_results_immediately>
<use_all_gpus>1</use_all_gpus>
<zero_debts>1</zero_debts>
<ncpus>-1</ncpus>
</options>
</cc_config>

Maybe the use_all_gpus is having an effect...

Hello!

Host 103082. ATI Radeon HD4890. ATI Catalyst 9.12.

12/01/2010 20:06:37 Starting BOINC client version 6.10.3 for windows_intelx86
...
12/01/2010 20:06:37 Processor: 2 GenuineIntel Intel(R) Core(TM)2 Duo CPU E6550 @ 2.33GHz [x86 Family 6 Model 15 Stepping 11]
12/01/2010 20:06:37 Processor: 4.00 MB cache
12/01/2010 20:06:37 Processor features: fpu tsc pae nx sse sse2 pni mmx
12/01/2010 20:06:37 OS: Microsoft Windows 7: x86 Edition, (06.01.7100.00)
12/01/2010 20:06:37 Memory: 2.00 GB physical, 4.00 GB virtual
12/01/2010 20:06:37 Disk: 32.00 GB total, 8.07 GB free
12/01/2010 20:06:37 Local time is UTC +3 hours
12/01/2010 20:06:41 Can't load library nvcuda.dll
12/01/2010 20:06:41 ATI GPU 0: RV770 (CAL version 1.4.515, 1024MB, 1520GFLOPS)

12/01/2010 20:07:10 Milkyway@home Requesting new tasks
12/01/2010 20:07:21 Milkyway@home Scheduler request completed: got 0 new tasks
12/01/2010 20:07:21 Milkyway@home Message from server: No work sent
12/01/2010 20:07:21 Milkyway@home Message from server: Need libraries named ati* to use ATI GPU
12/01/2010 20:07:21 Milkyway@home Message from server: Your computer has no NVIDIA GPU

Back to app_info...

Hello!

Host 103082. ATI Radeon HD4890. ATI Catalyst 9.12.

12/01/2010 20:06:37 Starting BOINC client version 6.10.3 for windows_intelx86
...
12/01/2010 20:06:37 Processor: 2 GenuineIntel Intel(R) Core(TM)2 Duo CPU E6550 @ 2.33GHz [x86 Family 6 Model 15 Stepping 11]
12/01/2010 20:06:37 Processor: 4.00 MB cache
12/01/2010 20:06:37 Processor features: fpu tsc pae nx sse sse2 pni mmx
12/01/2010 20:06:37 OS: Microsoft Windows 7: x86 Edition, (06.01.7100.00)
12/01/2010 20:06:37 Memory: 2.00 GB physical, 4.00 GB virtual
12/01/2010 20:06:37 Disk: 32.00 GB total, 8.07 GB free
12/01/2010 20:06:37 Local time is UTC +3 hours
12/01/2010 20:06:41 Can't load library nvcuda.dll
12/01/2010 20:06:41 ATI GPU 0: RV770 (CAL version 1.4.515, 1024MB, 1520GFLOPS)

12/01/2010 20:07:10 Milkyway@home Requesting new tasks
12/01/2010 20:07:21 Milkyway@home Scheduler request completed: got 0 new tasks
12/01/2010 20:07:21 Milkyway@home Message from server: No work sent
12/01/2010 20:07:21 Milkyway@home Message from server: Need libraries named ati* to use ATI GPU
12/01/2010 20:07:21 Milkyway@home Message from server: Your computer has no NVIDIA GPU

Back to app_info...

Just a guess, but you're way back on BOINC v6.10.3. BOINC GPU functionality has improved a lot since then. Try v6.10.25 and see if it solves the problem.

BOINC v6.10.25? I see 6.10.18 as recommended and 6.10.24 as development...

I also use the b-1 option in the app_info.xml to decrease WU time and have no problem getting WUs with "use CPU" set to no. I notice you're using BOINC v6.10.21. Try BOINC v6.10.25 and see if that fixes the problem. Please post the result.


My cc_config.xml also contains the following:

<cc_config>
<options>
<report_results_immediately>1</report_results_immediately>
<use_all_gpus>1</use_all_gpus>
<zero_debts>1</zero_debts>
<ncpus>-1</ncpus>
</options>
</cc_config>

Maybe the use_all_gpus is having an effect...

I am already using BOINC 6.10.25. Just to test I set my preferences the same as yours. BOINC only requests GPU but won't download any tasks. This is the same behaviour as previously. I added <use_all_gpus>1</use_all_gpus> to my cc_config.xml file and restarted BOINC. No difference.

With BOINC Manager v6.10.24 works. Thank you! :)

Can I control a CPU Time% that need for GPU-unit and number of concurrent GPU-units?
By default - only 68%, that too small, because with 100% I have a 99% GPU load, but with 68% - 90-95% only (about 6000 CS per day).

---

So why is the app at 0.38 cpu's and taking about 25 seconds longer on my 4870?

Did you use the b-1 tag in your app_info?

I am back to using a app_info as I did not like that slower crunching, of course I also had to change my prefs to use CPU as it will not get any work when it asks for GPU tasks.

Ahh, I had forgotten I had put <cmdline>w0.8 f20</cmdline> in my app_info!