Posts by shaf*

21) Message boards : Number crunching : Only getting 1 wu per GPU (Message 26101) Posted 20 Jun 2009 by shaf* Post: Still need to know what projects you are running and their resource share and also your preferences set under the BOINC Manager menu Advanced -> Preferences then "Connect about every" and "Additional work buffer". I believe you have a work scheduler issue....so to confirm it I need the above info. Also thanks for the link to the wu, but it has already been deleted off the system.... Thankyou again for your reply... Connect about every 10 days .... Additional work buffer 10 days Maybe my problem resides here ???? Strange really as I haven't changed anything since the last torrent a few months back. Many thanks in advance :) Copying and pasting a recent wu .... Task ID 85578306 Name ps_sgr_214F5_2s_hiw_2525229_1245535487_0 Workunit 83960315 Created 20 Jun 2009 22:04:51 UTC Sent 20 Jun 2009 22:06:29 UTC Received 20 Jun 2009 22:08:50 UTC Server state Over Outcome Success Client state Done Exit status 0 (0x0) Computer ID 55369 Report deadline 23 Jun 2009 22:06:29 UTC CPU time 76.51563 stderr out <core_client_version>6.6.20</core_client_version> <![CDATA[ <stderr_txt> Running Milkyway@home ATI GPU application version 0.19f by Gipsel CPU: Intel(R) Core(TM)2 CPU 6420 @ 2.13GHz (2 cores/threads) 3.19991 GHz (200ms) CAL Runtime: 1.3.158 Found 2 CAL devices Device 0: ATI Radeon HD 4800 (RV770) 512 MB local RAM (remote 28 MB cached + 256 MB uncached) GPU core clock: 655 MHz, memory clock: 250 MHz 640 shader units organized in 8 SIMDs with 16 VLIW units (5-issue), wavefront size 64 threads supporting double precision Device 1: ATI Radeon HD 4800 (RV770) 512 MB local RAM (remote 28 MB cached + 256 MB uncached) GPU core clock: 655 MHz, memory clock: 250 MHz 640 shader units organized in 8 SIMDs with 16 VLIW units (5-issue), wavefront size 64 threads supporting double precision 0 WUs already running on GPU 0 1 WUs already running on GPU 1 Starting WU on GPU 0 main integral, 320 iterations predicted runtime per iteration is 208 ms (33.3333 ms are allowed), dividing each iteration in 7 parts borders of the domains at 0 232 464 688 920 1144 1376 1600 Calculated about 7.40024e+012 floatingpoint ops on GPU, 1.23583e+008 on FPU. Approximate GPU time 76.5156 seconds. probability calculation (stars) Calculated about 1.97172e+009 floatingpoint ops on FPU. WU completed. CPU time: 5.14063 seconds, GPU time: 76.5156 seconds, wall clock time: 77.786 seconds, CPU frequency: 3.19995 GHz </stderr_txt> ]]> Validate state Valid Claimed credit 0.437636898158303 Granted credit 55.516612 application version 0.19
22) Message boards : Number crunching : Only getting 1 wu per GPU (Message 26083) Posted 20 Jun 2009 by shaf* Post: You'll need to tell us more about your system. Which BOINC version are you running? What other projects are you running? What is the resource share for each project? What are the network prefences you have set up? Thanks mate. Here's what I've got. OS Win XP SP3 Catalyst 9.1 Cal version : 1.3.158 E6420 dual core @ 3.4ghz 2gig of RAM 2 x 4830s @ 655/950 run these separately when crunching (Catalyst profile) Crossfire profile only for gaming (whilst I rarely do anyway). GPU-Z shows load on both the cards very well. Detached all boinc projects and the machine is dedicated to Milkyway only... Boinc manager tried 6.6.36 but slipped back to 6.6.20 Run Always Network Always avaiable Using zslip 0.19 ATI post June the 5th astronomy_0.19_ATI_SSE2f.exe My app_info <app_info> <app> <name>milkyway</name> </app> <file_info> <name>astronomy_0.19_ATI_SSE2f.exe</name> <executable/> </file_info> <file_info> <name>brook.dll</name> <executable/> </file_info> <app_version> <app_name>milkyway</app_name> <version_num>19</version_num> <flops>2.0e11</flops> <avg_ncpus>0.1</avg_ncpus> <ncpus>2</ncpus> <max_ncpus>1</max_ncpus> <cmdline>n3 w1.4/cmdline> <file_ref> <file_name>astronomy_0.19_ATI_SSE2f.exe</file_name> <main_program/> </file_ref> <file_ref> <file_name>brook.dll</file_name> </file_ref> </app_version> </app_info> Sample WU > http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=85362441 If I look at the tasks in Boinc Manager I only ever see 2 being worked on ... Never any waiting. Any ideas would be much appreciated. Many thanks !
23) Message boards : Number crunching : Only getting 1 wu per GPU (Message 26053) Posted 19 Jun 2009 by shaf* Post: btw, I'm not getting 6 wu's/cpu on my quady. I'm averaging just over 2.... 2 per core ? That makes 8 available for for your gpus. I get just 2 :-o
24) Message boards : Number crunching : Only getting 1 wu per GPU (Message 25994) Posted 19 Jun 2009 by shaf* Post: Can someone please explain why I'm only getting 1 when everyone else gets 6 ? It's very frustrating :(
25) Message boards : Number crunching : 64 CPUs on a Q6600 ???? (Message 25992) Posted 19 Jun 2009 by shaf* Post: Hi, Anyone have a link to the boincmgr mod ? I too have been contemplating modifying the boincmgr source.
26) Message boards : Number crunching : MW ATI work (Message 25963) Posted 18 Jun 2009 by shaf* Post: No worries... I've got it sorted... Slip back to 6.6.20 and modified app_info, now the GPUs are fully loaded... Next problem is that there's a bit too much heat on the motherboard ;)
27) Message boards : Number crunching : MW ATI work (Message 25956) Posted 18 Jun 2009 by shaf* Post: I don't get it ... http://milkyway.cs.rpi.edu/milkyway/results.php?hostid=65070 This guy is getting 12 WUs to process whereas I'm only getting 2 ! What settings is he using in app_info.xml ? Or should I try using the same client as him 6.6.20 ?
28) Message boards : Number crunching : ATI 4750/4770/4890 support (Message 18058) Posted 9 Apr 2009 by shaf* Post: +3
29) Message boards : Application Code Discussion : Recompiled Linux 32/64 apps (Message 17684) Posted 5 Apr 2009 by shaf* Post: You could use the SSSE3 version (instead of only SSE3) on your Core2. The SSE3 binary retains compatibilty to AMD CPUs and may be a bit slower. OK thanks - I don't think it'll make much difference - I'll give that a try... Seems also that he's using version 0.19.... This is looking more and more like me needing to compile 0.19 from source unless there are any binaries laying around for the lazy ;)
30) Message boards : Application Code Discussion : Recompiled Linux 32/64 apps (Message 17681) Posted 5 Apr 2009 by shaf* Post: I thought the fastest version of Speedimic is quite close. Just looked it up, speedimics own Q9550 (45nm, 2.83GHz) is taking about 1030 seconds for a 27.77 credit WU under Linux. He has another host (Q6600, 65nm, appears to be overclocked to 2.7-2.8GHz from the benchmark values) completing the same tasks in about 1150 seconds under Windows. It is a somehow bad comparison because of the different clockspeed and that the 45nm CPUs should be faster per clock than their 65nm counterparts, but nonetheless it is clear that his Linux version can't be that bad. OK the q6600 is closer to arch than the q9550. He's using SSE4.1 on the q9550. Odd - He has a different client to me on his q6600. Any ideas where I can get it from ?
31) Message boards : Application Code Discussion : Recompiled Linux 32/64 apps (Message 17680) Posted 5 Apr 2009 by shaf* Post: The linux client is simply inefficient compared to win32. I thought the fastest version of Speedimic is quite close. Just looked it up, speedimics own Q9550 (45nm, 2.83GHz) is taking about 1030 seconds for a 27.77 credit WU under Linux. He has another host (Q6600, 65nm, appears to be overclocked to 2.7-2.8GHz from the benchmark values) completing the same tasks in about 1150 seconds under Windows. It is a somehow bad comparison because of the different clockspeed and that the 45nm CPUs should be faster per clock than their 65nm counterparts, but nonetheless it is clear that his Linux version can't be that bad. Here are my tasks .... Laptop win32 : http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=39511370 Linux : http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=39523720 Thanks for the pointers I'll recheck if I'm using the wrong client
32) Message boards : Application Code Discussion : Recompiled Linux 32/64 apps (Message 17668) Posted 5 Apr 2009 by shaf* Post: It might be because Linux manages power differently from Windows, running BOINC applications at a slow CPU frequency in order to save energy. See more details here. HTH Afraid not, speedstep, thermal throttling are all disabled in the BIOS. Solid overclock applied and powernow/acpi daemons are not running once booted. The linux client is simply inefficient compared to win32.
33) Message boards : Application Code Discussion : Recompiled Linux 32/64 apps (Message 17624) Posted 5 Apr 2009 by shaf* Post: Still a little disappointed with the linux apps. I have SSE3 running on an e6400@2.4 test machine ... -rw-r--r-- 1 root root 594 2009-02-18 21:33 app_info.xml -rw-r--r-- 1 root root 35127 2009-01-19 20:07 GPL.txt -rwxr-xr-x 1 root root 1222624 2009-02-18 21:25 milkyway_0.18_sse3_i686-pc-linux-gnu This yield a wu time of 35minutes Whereas the windows op app yields a time of 20minutes on a slower processor.(1.8ghz T2390) The difference seems a lot to me or is this typical ?
34) Message boards : Number crunching : ATI 4750/4770/4890 support (Message 17301) Posted 1 Apr 2009 by shaf* Post: Do we envisage milkyway@home being able to run on these cards as I'm intending to sell up and jump to 47xx.
35) Message boards : Number crunching : 3rd.in - optimized apps (Message 15534) Posted 15 Mar 2009 by shaf* Post: Being at 90% the system is more responsive but to be honest I would prefer to have the gpu crunching 100% with 0.19e:) Sometimes I wonder why I have taken the effort to write a 12kB readme. Is anyone reading it? If you don't know how to change the behaviour after reading the readme (hint: the f option is what you are looking for), you may ask again ;) Thanks
36) Message boards : Number crunching : 3rd.in - optimized apps (Message 15509) Posted 15 Mar 2009 by shaf* Post: By the way, you could put version 0.19e of the ATI GPU app on zslip. The new version plays a bit nicer with the OS trying to react to user input. The system response should be quite a bit improved. Thanks, I've heard good news in that the gpu is now only at 90% load but does that mean that my RAC will go down by 10% ? Being at 90% the system is more responsive but to be honest I would prefer to have the gpu crunching 100% with 0.19e:)

Previous 20