Herein lies my install log. Read em and laugh.

I had an asus en9600gso I bought for folding@home, but alas, they have their issues. So I decided to try gpugrid.net

So, I take a gigabyte ga-m57sli-s4, 2GB of ram, said 9600gso and an arbitrary hard disc and install xp pro sp3, the latest cuda and the latest boinc.

First of all, I noticed that the latest cuda is called "NVIDIADisplayWin2K(177_84)Int" instead of the last one I installed for FAH which was "177.35.geforce_winxp_32bit_english_whql". I got it from the same place I got the last, so I figured it was ok.

Next, I discovered I could not find the boinc requested at berekeley, so used the direct download.

Ok, so everything installs ok, and I reboot. It asks me to select a project, and as usual, I hit cancel (I prefer to let bam manage my projects).

boinc then hangs. Deader than a doornail. System is sluggish as sin. Rebooting requires me to terminate boinc.

No worries, I reboot and instead of using bam, I connect to gpugrid.net It connects just fine, then same result. Computer is slugish, boinc won't open.

Ok, so I crawl slowly into c:\documents and settings\all users\boinc\application data\boinc and grab stdoutdae.txt

StartServiceCtrlDispatcher being called.
This may take several seconds.  Please wait.
14-Sep-2008 19:32:50 [---] Starting BOINC client version 6.3.10 for windows_intelx86
14-Sep-2008 19:32:50 [---] This a development version of BOINC and may not function properly
14-Sep-2008 19:32:50 [---] log flags: task, file_xfer, sched_ops
14-Sep-2008 19:32:50 [---] Libraries: libcurl/7.18.0 OpenSSL/0.9.8e zlib/1.2.3
14-Sep-2008 19:32:50 [---] Running as a daemon
14-Sep-2008 19:32:50 [---] Data directory: C:\Documents and Settings\All Users\Application Data\BOINC
14-Sep-2008 19:32:50 [---] Running under account boinc_master
14-Sep-2008 19:32:50 [---] Processor: 2 AuthenticAMD AMD Athlon(tm) 64 X2 Dual Core Processor 4400+ [x86 Family 15 Model 107 Stepping 1]
14-Sep-2008 19:32:50 [---] Processor features: fpu tsc pae nx sse sse2 3dnow mmx
14-Sep-2008 19:32:50 [---] OS: Microsoft Windows XP: Professional x86 Editon, Service Pack 3, (05.01.2600.00)
14-Sep-2008 19:32:50 [---] Memory: 2.00 GB physical, 3.85 GB virtual
14-Sep-2008 19:32:50 [---] Disk: 37.27 GB total, 33.80 GB free
14-Sep-2008 19:32:50 [---] Local time is UTC -7 hours
14-Sep-2008 19:32:50 [---] Not using a proxy
14-Sep-2008 19:32:52 [---] CUDA devices found
14-Sep-2008 19:32:52 [---] Coprocessor: GeForce 9600 GSO (1)
14-Sep-2008 19:32:52 [---] No general preferences found - using BOINC defaults
14-Sep-2008 19:32:52 [---] Preferences limit memory usage when active to 1023.74MB
14-Sep-2008 19:32:52 [---] Preferences limit memory usage when idle to 1842.73MB
14-Sep-2008 19:32:52 [---] Preferences limit disk usage to 9.31GB
14-Sep-2008 19:32:52 [---] This computer is not attached to any projects
14-Sep-2008 19:32:52 [---] Visit http://boinc.berkeley.edu for instructions
[09/14/08 19:34:35] TRACE [1340]: ***** Console Event Detected *****

[09/14/08 19:34:35] TRACE [1340]: Event: CTRL-LOGOFF Event

[09/14/08 19:35:15] TRACE [1868]: ***** Console Event Detected *****

[09/14/08 19:35:15] TRACE [1868]: Event: CTRL-LOGOFF Event

[09/14/08 19:35:16] TRACE [236]: ***** Console Event Detected *****

[09/14/08 19:35:16] TRACE [236]: Event: CTRL-CLOSE or CTRL-SHUTDOWN Event

14-Sep-2008 19:35:17 [---] Exit requested by user

StartServiceCtrlDispatcher being called.
This may take several seconds.  Please wait.
14-Sep-2008 19:36:03 [---] Starting BOINC client version 6.3.10 for windows_intelx86
14-Sep-2008 19:36:03 [---] This a development version of BOINC and may not function properly
14-Sep-2008 19:36:03 [---] log flags: task, file_xfer, sched_ops
14-Sep-2008 19:36:03 [---] Libraries: libcurl/7.18.0 OpenSSL/0.9.8e zlib/1.2.3
14-Sep-2008 19:36:03 [---] Running as a daemon
14-Sep-2008 19:36:03 [---] Data directory: C:\Documents and Settings\All Users\Application Data\BOINC
14-Sep-2008 19:36:03 [---] Running under account boinc_master
14-Sep-2008 19:36:03 [---] Processor: 2 AuthenticAMD AMD Athlon(tm) 64 X2 Dual Core Processor 4400+ [x86 Family 15 Model 107 Stepping 1]
14-Sep-2008 19:36:03 [---] Processor features: fpu tsc pae nx sse sse2 3dnow mmx
14-Sep-2008 19:36:03 [---] OS: Microsoft Windows XP: Professional x86 Editon, Service Pack 3, (05.01.2600.00)
14-Sep-2008 19:36:03 [---] Memory: 2.00 GB physical, 3.85 GB virtual
14-Sep-2008 19:36:03 [---] Disk: 37.27 GB total, 33.78 GB free
14-Sep-2008 19:36:03 [---] Local time is UTC -7 hours
14-Sep-2008 19:36:03 [---] Not using a proxy
14-Sep-2008 19:36:03 [---] CUDA devices found
14-Sep-2008 19:36:03 [---] Coprocessor: GeForce 9600 GSO (1)
14-Sep-2008 19:36:03 [---] No general preferences found - using BOINC defaults
14-Sep-2008 19:36:03 [---] Preferences limit memory usage when active to 1023.74MB
14-Sep-2008 19:36:03 [---] Preferences limit memory usage when idle to 1842.73MB
14-Sep-2008 19:36:03 [---] Preferences limit disk usage to 9.31GB
14-Sep-2008 19:36:03 [---] This computer is not attached to any projects
14-Sep-2008 19:36:03 [---] Visit http://boinc.berkeley.edu for instructions
[09/14/08 19:36:53] TRACE [428]: ***** Console Event Detected *****

[09/14/08 19:36:53] TRACE [428]: Event: CTRL-LOGOFF Event

[09/14/08 19:37:04] TRACE [1224]: ***** Console Event Detected *****

[09/14/08 19:37:04] TRACE [1224]: Event: CTRL-LOGOFF Event

[09/14/08 19:37:06] TRACE [1560]: ***** Console Event Detected *****

[09/14/08 19:37:06] TRACE [1560]: Event: CTRL-CLOSE or CTRL-SHUTDOWN Event

14-Sep-2008 19:37:06 [---] Exit requested by user

StartServiceCtrlDispatcher being called.
This may take several seconds.  Please wait.
14-Sep-2008 19:37:54 [---] Starting BOINC client version 6.3.10 for windows_intelx86
14-Sep-2008 19:37:54 [---] This a development version of BOINC and may not function properly
14-Sep-2008 19:37:54 [---] log flags: task, file_xfer, sched_ops
14-Sep-2008 19:37:54 [---] Libraries: libcurl/7.18.0 OpenSSL/0.9.8e zlib/1.2.3
14-Sep-2008 19:37:54 [---] Running as a daemon
14-Sep-2008 19:37:54 [---] Data directory: C:\Documents and Settings\All Users\Application Data\BOINC
14-Sep-2008 19:37:54 [---] Running under account boinc_master
14-Sep-2008 19:37:54 [---] Processor: 2 AuthenticAMD AMD Athlon(tm) 64 X2 Dual Core Processor 4400+ [x86 Family 15 Model 107 Stepping 1]
14-Sep-2008 19:37:54 [---] Processor features: fpu tsc pae nx sse sse2 3dnow mmx
14-Sep-2008 19:37:54 [---] OS: Microsoft Windows XP: Professional x86 Editon, Service Pack 3, (05.01.2600.00)
14-Sep-2008 19:37:54 [---] Memory: 2.00 GB physical, 3.85 GB virtual
14-Sep-2008 19:37:54 [---] Disk: 37.27 GB total, 33.77 GB free
14-Sep-2008 19:37:54 [---] Local time is UTC -7 hours
14-Sep-2008 19:37:54 [---] Not using a proxy
14-Sep-2008 19:37:56 [---] CUDA devices found
14-Sep-2008 19:37:56 [---] Coprocessor: GeForce 9600 GSO (1)
14-Sep-2008 19:37:56 [---] No general preferences found - using BOINC defaults
14-Sep-2008 19:37:56 [---] Preferences limit memory usage when active to 1023.74MB
14-Sep-2008 19:37:56 [---] Preferences limit memory usage when idle to 1842.73MB
14-Sep-2008 19:37:56 [---] Preferences limit disk usage to 9.31GB
14-Sep-2008 19:37:56 [---] This computer is not attached to any projects
14-Sep-2008 19:37:56 [---] Visit http://boinc.berkeley.edu for instructions
14-Sep-2008 19:43:58 [---] Fetching configuration file from http://www.gpugrid.net/get_project_config.php
14-Sep-2008 19:44:16 [---] Running CPU benchmarks
14-Sep-2008 19:44:16 [---] Suspending computation - running CPU benchmarks
14-Sep-2008 19:44:21 [PS3GRID] Master file download succeeded
14-Sep-2008 19:44:26 [PS3GRID] Sending scheduler request: Project initialization.  Requesting 1 seconds of work, reporting 0 completed tasks
14-Sep-2008 19:44:31 [PS3GRID] Scheduler request completed: got 1 new tasks
14-Sep-2008 19:44:31 [PS3GRID] You used the wrong URL for this project
14-Sep-2008 19:44:31 [PS3GRID] The correct URL is http://www.ps3grid.net/
14-Sep-2008 19:44:31 [PS3GRID] Using the wrong URL can cause problems in some cases.
14-Sep-2008 19:44:31 [PS3GRID] When convenient, detach this project, then reattach to http://www.ps3grid.net/
14-Sep-2008 19:44:33 [PS3GRID] Started download of acemd_6.45_windows_intelx86__cuda.exe
14-Sep-2008 19:44:33 [PS3GRID] Started download of tcl85.dll
14-Sep-2008 19:44:47 [---] Benchmark results:
14-Sep-2008 19:44:47 [---]    Number of CPUs: 2
14-Sep-2008 19:44:47 [---]    2372 floating point MIPS (Whetstone) per CPU
14-Sep-2008 19:44:47 [---]    4365 integer MIPS (Dhrystone) per CPU
14-Sep-2008 19:44:48 [---] Resuming computation
14-Sep-2008 19:44:54 [PS3GRID] Finished download of acemd_6.45_windows_intelx86__cuda.exe
14-Sep-2008 19:44:54 [PS3GRID] Finished download of tcl85.dll
14-Sep-2008 19:44:54 [PS3GRID] Started download of cudart.dll
14-Sep-2008 19:44:54 [PS3GRID] Started download of iNH8602-GPUTEST3-3-LICENSE
14-Sep-2008 19:44:57 [PS3GRID] Finished download of iNH8602-GPUTEST3-3-LICENSE
14-Sep-2008 19:44:57 [PS3GRID] Started download of iNH8602-GPUTEST3-3-COPYRIGHT
14-Sep-2008 19:44:58 [PS3GRID] Finished download of iNH8602-GPUTEST3-3-COPYRIGHT
14-Sep-2008 19:44:58 [PS3GRID] Started download of iNH8602-GPUTEST3-3-iNH8602-GPUTEST3-2-10-acemd_1
14-Sep-2008 19:45:02 [PS3GRID] Finished download of cudart.dll
14-Sep-2008 19:45:02 [PS3GRID] Started download of iNH8602-GPUTEST3-3-iNH8602-GPUTEST3-2-10-acemd_2
14-Sep-2008 19:45:07 [PS3GRID] Sending scheduler request: To fetch work.  Requesting 35176 seconds of work, reporting 0 completed tasks
14-Sep-2008 19:45:12 [PS3GRID] Scheduler request completed: got 1 new tasks
14-Sep-2008 19:45:12 [PS3GRID] You used the wrong URL for this project
14-Sep-2008 19:45:12 [PS3GRID] The correct URL is http://www.ps3grid.net/
14-Sep-2008 19:45:12 [PS3GRID] Using the wrong URL can cause problems in some cases.
14-Sep-2008 19:45:12 [PS3GRID] When convenient, detach this project, then reattach to http://www.ps3grid.net/
14-Sep-2008 19:45:19 [PS3GRID] Finished download of iNH8602-GPUTEST3-3-iNH8602-GPUTEST3-2-10-acemd_1
14-Sep-2008 19:45:19 [PS3GRID] Started download of iNH8602-GPUTEST3-3-iNH8602-GPUTEST3-2-10-acemd_3
14-Sep-2008 19:45:20 [PS3GRID] Finished download of iNH8602-GPUTEST3-3-iNH8602-GPUTEST3-2-10-acemd_2
14-Sep-2008 19:45:20 [PS3GRID] Finished download of iNH8602-GPUTEST3-3-iNH8602-GPUTEST3-2-10-acemd_3
14-Sep-2008 19:45:20 [PS3GRID] Started download of iNH8602-GPUTEST3-3-grama.ionized.pdb
14-Sep-2008 19:45:20 [PS3GRID] Started download of iNH8602-GPUTEST3-3-grama.ionized.psf
14-Sep-2008 19:45:47 [PS3GRID] Sending scheduler request: To fetch work.  Requesting 30240 seconds of work, reporting 0 completed tasks
14-Sep-2008 19:45:52 [PS3GRID] Scheduler request completed: got 0 new tasks
14-Sep-2008 19:45:52 [PS3GRID] You used the wrong URL for this project
14-Sep-2008 19:45:52 [PS3GRID] The correct URL is http://www.ps3grid.net/
14-Sep-2008 19:45:52 [PS3GRID] Using the wrong URL can cause problems in some cases.
14-Sep-2008 19:45:52 [PS3GRID] When convenient, detach this project, then reattach to http://www.ps3grid.net/
14-Sep-2008 19:45:52 [PS3GRID] Message from server: No work sent
14-Sep-2008 19:45:52 [PS3GRID] Message from server: (reached per-CPU limit of 1 tasks)
14-Sep-2008 19:46:42 [PS3GRID] Finished download of iNH8602-GPUTEST3-3-grama.ionized.pdb
14-Sep-2008 19:46:42 [PS3GRID] Started download of iNH8602-GPUTEST3-3-parameters
14-Sep-2008 19:46:51 [PS3GRID] Finished download of iNH8602-GPUTEST3-3-parameters
14-Sep-2008 19:46:51 [PS3GRID] Started download of iNH8602-GPUTEST3-3-GPUTEST3
14-Sep-2008 19:46:52 [PS3GRID] Finished download of iNH8602-GPUTEST3-3-GPUTEST3
14-Sep-2008 19:46:52 [PS3GRID] Started download of logops3grid.png
14-Sep-2008 19:46:52 [PS3GRID] Sending scheduler request: To fetch work.  Requesting 30240 seconds of work, reporting 0 completed tasks
14-Sep-2008 19:46:53 [PS3GRID] Finished download of logops3grid.png
14-Sep-2008 19:46:53 [PS3GRID] Started download of project_1.png
14-Sep-2008 19:46:56 [PS3GRID] Finished download of project_1.png
14-Sep-2008 19:46:56 [PS3GRID] Started download of project_2.png
14-Sep-2008 19:46:57 [PS3GRID] Scheduler request completed: got 0 new tasks
14-Sep-2008 19:46:57 [PS3GRID] You used the wrong URL for this project
14-Sep-2008 19:46:57 [PS3GRID] The correct URL is http://www.ps3grid.net/
14-Sep-2008 19:46:57 [PS3GRID] Using the wrong URL can cause problems in some cases.
14-Sep-2008 19:46:57 [PS3GRID] When convenient, detach this project, then reattach to http://www.ps3grid.net/
14-Sep-2008 19:46:57 [PS3GRID] Message from server: No work sent
14-Sep-2008 19:46:57 [PS3GRID] Message from server: (reached per-CPU limit of 1 tasks)
14-Sep-2008 19:46:59 [PS3GRID] Finished download of project_2.png
14-Sep-2008 19:46:59 [PS3GRID] Started download of project_3.png
14-Sep-2008 19:47:01 [PS3GRID] Finished download of project_3.png
14-Sep-2008 19:47:01 [PS3GRID] Started download of pH12961-GPUTEST3-3-LICENSE
14-Sep-2008 19:47:02 [PS3GRID] Finished download of pH12961-GPUTEST3-3-LICENSE
14-Sep-2008 19:47:02 [PS3GRID] Started download of pH12961-GPUTEST3-3-COPYRIGHT
14-Sep-2008 19:47:07 [PS3GRID] Finished download of pH12961-GPUTEST3-3-COPYRIGHT
14-Sep-2008 19:47:07 [PS3GRID] Started download of pH12961-GPUTEST3-3-pH12961-GPUTEST3-2-10-acemd_1
14-Sep-2008 19:47:08 [PS3GRID] Finished download of iNH8602-GPUTEST3-3-grama.ionized.psf
14-Sep-2008 19:47:08 [PS3GRID] Started download of pH12961-GPUTEST3-3-pH12961-GPUTEST3-2-10-acemd_2
14-Sep-2008 19:47:09 [PS3GRID] Starting iNH8602-GPUTEST3-3-10-acemd_0
14-Sep-2008 19:47:09 [PS3GRID] Starting task iNH8602-GPUTEST3-3-10-acemd_0 using acemd version 645
14-Sep-2008 19:47:23 [PS3GRID] Finished download of pH12961-GPUTEST3-3-pH12961-GPUTEST3-2-10-acemd_1
14-Sep-2008 19:47:23 [PS3GRID] Finished download of pH12961-GPUTEST3-3-pH12961-GPUTEST3-2-10-acemd_2
14-Sep-2008 19:47:23 [PS3GRID] Started download of pH12961-GPUTEST3-3-pH12961-GPUTEST3-2-10-acemd_3
14-Sep-2008 19:47:23 [PS3GRID] Started download of pH12961-GPUTEST3-3-grama.ionized.pdb
14-Sep-2008 19:47:25 [PS3GRID] Finished download of pH12961-GPUTEST3-3-pH12961-GPUTEST3-2-10-acemd_3
14-Sep-2008 19:47:25 [PS3GRID] Started download of pH12961-GPUTEST3-3-grama.ionized.psf
14-Sep-2008 19:47:58 [PS3GRID] Sending scheduler request: To fetch work.  Requesting 30240 seconds of work, reporting 0 completed tasks
14-Sep-2008 19:48:03 [PS3GRID] Scheduler request completed: got 0 new tasks
14-Sep-2008 19:48:03 [PS3GRID] You used the wrong URL for this project
14-Sep-2008 19:48:03 [PS3GRID] The correct URL is http://www.ps3grid.net/
14-Sep-2008 19:48:03 [PS3GRID] Using the wrong URL can cause problems in some cases.
14-Sep-2008 19:48:03 [PS3GRID] When convenient, detach this project, then reattach to http://www.ps3grid.net/
14-Sep-2008 19:48:03 [PS3GRID] Message from server: No work sent
14-Sep-2008 19:48:03 [PS3GRID] Message from server: (reached per-CPU limit of 1 tasks)
14-Sep-2008 19:48:04 [PS3GRID] Finished download of pH12961-GPUTEST3-3-grama.ionized.pdb
14-Sep-2008 19:48:04 [PS3GRID] Started download of pH12961-GPUTEST3-3-parameters
14-Sep-2008 19:48:09 [PS3GRID] Finished download of pH12961-GPUTEST3-3-parameters
14-Sep-2008 19:48:09 [PS3GRID] Started download of pH12961-GPUTEST3-3-GPUTEST3
14-Sep-2008 19:48:10 [PS3GRID] Finished download of pH12961-GPUTEST3-3-GPUTEST3
14-Sep-2008 19:48:45 [PS3GRID] Finished download of pH12961-GPUTEST3-3-grama.ionized.psf
14-Sep-2008 19:49:03 [PS3GRID] Sending scheduler request: To fetch work.  Requesting 30240 seconds of work, reporting 0 completed tasks
14-Sep-2008 19:49:08 [PS3GRID] Scheduler request completed: got 0 new tasks
14-Sep-2008 19:49:08 [PS3GRID] You used the wrong URL for this project
14-Sep-2008 19:49:08 [PS3GRID] The correct URL is http://www.ps3grid.net/
14-Sep-2008 19:49:08 [PS3GRID] Using the wrong URL can cause problems in some cases.
14-Sep-2008 19:49:08 [PS3GRID] When convenient, detach this project, then reattach to http://www.ps3grid.net/
14-Sep-2008 19:49:08 [PS3GRID] Message from server: No work sent
14-Sep-2008 19:49:08 [PS3GRID] Message from server: (reached per-CPU limit of 1 tasks)
14-Sep-2008 19:50:08 [PS3GRID] Sending scheduler request: To fetch work.  Requesting 30240 seconds of work, reporting 0 completed tasks
14-Sep-2008 19:50:13 [PS3GRID] Scheduler request completed: got 0 new tasks
14-Sep-2008 19:50:13 [PS3GRID] You used the wrong URL for this project
14-Sep-2008 19:50:13 [PS3GRID] The correct URL is http://www.ps3grid.net/
14-Sep-2008 19:50:13 [PS3GRID] Using the wrong URL can cause problems in some cases.
14-Sep-2008 19:50:13 [PS3GRID] When convenient, detach this project, then reattach to http://www.ps3grid.net/
14-Sep-2008 19:50:13 [PS3GRID] Message from server: No work sent
14-Sep-2008 19:50:13 [PS3GRID] Message from server: (reached per-CPU limit of 1 tasks)

It may actually be doing something. It says it is restarting the task when I reboot the machine (trying to debug a broken boinc design). But, the error messages are not reassuring, and having no control over the machine is even less reassuring.

I may update this if I ever get it to work ..

Update:

I deleted both "client_auth.xml" and "gui_rpc_auth.cfg" and rebooted and BOINC was able to regain local control of the machine. It says it has two tasks and is working on one of them. I loaded a second BOINC project to keep the other core of the X2 happy and it too seems to run.

The machine is worthless for anything else. It takes 2, 3 or 4 clicks for the action to be registered. But, I built that machine for DC, so I don't mind.

Getting remote control of the machine is still on my list, as I usually manage my grove remotely, but for now, I will let it run.

There is nothing unusual in the logs (not at least that I can see anyway). The initial console errors are from your manual interruption to attach via bam and should be transitory. The messages about not receiving work are normal, sent because you have reached your quota. The message about the url being incorrect is a problem with the automatic attachment via boinc manager - it should not in reality be a problem right now.

Nothing stands out to suggest why the machine is unresponsive. I can only suggest that you follow the normal debug routes for unresponsive machines - looking at what is eating the CPU cycles, how much memory is in use, disk I/O, AV scanning etc...

What happens if you suspend PS3grid and crunch two CPU work units: is it still just as sluggish? If you shutdown Boinc and Boinc Manager completely, does the machine spring into life?

---

Next, I discovered I could not find the boinc requested at berekeley, so used the direct download.

Links to download the required BOINC clients are on the front page of this project under Join with your Nvidia graphics card (beta)

Everything seems to be running fine now. I am finishing units in 72000 seconds on a card that the throughput thread should take 74000 seconds. I have one core of an x2-4400 feeding the GPU and the other running a different BOINC project. I don't know why it was so obnoxious initially starting, but it it is working now, so that's all that is important. I plan on sticking a second card in that machine, which will eat both cores until the BOINC update.

What I tried to say in my initial post was that I could not find the latest BOINC beta at the Berkeley site. I did download it using the link on this projects home page. I was crawling around there to see if there was something newer..

Anyway, this project seems to scale nicely with the number of shaders * clock, and does not have the issues that "the other project" does with stock CUDA hardware. I think I'll stick around a while. I just hope I can figure out how to enable remote management in BOINC6 as I am rather fond of it.

You can find all versions here:

all versions

---

What I tried to say in my initial post was that I could not find the latest BOINC beta at the Berkeley site. I did download it using the link on this projects home page. I was crawling around there to see if there was something newer..

Actually there is no BOINC beta. There are alpha testing versions and then when enough testing is done by the alpha testers and they reach a stable point a version goes public.

You won't actually find 6.3.xx listed there, except on the direct download page, as it is basically pre-alpha, the 6.3.xx clients are only for testing the GPU functions at this time. It is not posted as not even the regular alpha testers are testing features in it, they are still working on 6.2.xx clients. 6.3.xx is not really newer then 6.2.xx, its only different in that the scheduler needed to be changed for co-processors and it was easier for the berkeley developers to make a separate branch while the testing goes on. At some point everything will get merged, that may be like 6.4.xx

I have been crunching along for almost two weeks now so thought I would mix things up a little. I picked up a new video card for my desktop to free up a second, identical, en9600gso/top, to the one in the GPUGRID machine. So, I swap them over last night with few supprizes.

Just like F@H, I needed to re-run the nvidia install program before GPUz would recognize the second card, and BOINC was only seeing 1 coprocessor. So I run setup and it tells me that I am running something newer. No biggie, just do your thing. And it deinstalls the existing driver and reboots. This, of course, causes BOINC to crash burn and die when it automatically restarts. No worries, this was expected. Finish loading the new drivers, reboot and both GPUz and BOINC see two nice shiny 9600GSO's. But, BOINC is only using one of them because I had a CPU project running, so I flag that one "no more work" and go to bed.

This afternoon (err, morning) I get up to two PS3GRID happily crunching.

My log file is filled with "reached CPU limit of 1 task" which is normal, but what is different (well, actually, NOT different) is the number of tasks.

Before the upgrade, my x2-4400 with a single 9600gso would have two PS3GRID WU's. One was crunching, the other was in await. I like that so there is always one ready to go when the first finishes. CPU utilization was 0 so I figured I would try the ncpu's trick. Creating a cc_config.xml file and reloading advanced preferences had it spot the change and re-run benchmarks. I re-enabled the CPU project and it grabbed a pile of work for it. I am now crunching two CPU units and two GPU units on my dual core x2.

But ... I really want to have "one wu in reserve". Not necessarily two, but one, so new work can start immediately instead of having to wait. It used to do that before, but doesn't do it now that I have two GPU's, the same as the number of CPU's in the system.

Anyway, just checking in and saying "looking good folks". I'll update you once I get a weeks stats on the impact of a second card and of using both cores for CPU tasks. I was going to set <ncpus> to 3, and will do that if I see any performance penalty by letting both cores do nqueens. My only concern would be that the machine might choose to run two instances of nqueens and leave one GPU idle. Does anyone know if that is likely?

But ... I really want to have "one wu in reserve". Not necessarily two, but one, so new work can start immediately instead of having to wait. It used to do that before, but doesn't do it now that I have two GPU's, the same as the number of CPU's in the system.

This will be fixed when it is not going to be necessary to use ncpus.

gdf

I'll update you once I get a weeks stats on the impact of a second card and of using both cores for CPU tasks. I was going to set <ncpus> to 3, and will do that if I see any performance penalty by letting both cores do nqueens.

It looks like I took a 10% hit by adding the second card AND setting ncpus to 4.
It was taking a fairly consistent 84 msec per step with just one card in the system and one core dedicated to feed it.

It is now running 92 msec per step on each card.

I will drop ncpus to 3 this weekend which "should" reserve one cpu core to drive both GPU's and see if that gets me back to where I was.

I will drop ncpus to 3 this weekend which "should" reserve one cpu core to drive both GPU's and see if that gets me back to where I was.

Definitely try it.

MrS

---

Scanning for our furry friends since Jan 2002