Message boards :
Number crunching :
v0.18/v0.19 issues here
Message board moderation
Previous · 1 · 2 · 3 · 4 · 5 · 6 · Next
Author | Message |
---|---|
![]() Send message Joined: 12 Apr 08 Posts: 621 Credit: 161,934,067 RAC: 0 ![]() ![]() |
Now it is working fine with 6.6.28. Um, can you show us the cc_config... for 6.5.0 there should be no need to set GPU numbers at all. The only thing you should set, and set to 1, is the flag to use all GPUs ... |
Send message Joined: 7 Jul 09 Posts: 11 Credit: 4,716 RAC: 0 ![]() ![]() |
boinc 6.6.36 http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=97072588 Fonction incorrecte. (0x1) - exit code 1 (0x1) sorry 1) boinc.exe quit 2) ctrl+alt+suppr => boinc & milkyway_19...exe |
Send message Joined: 18 Jul 09 Posts: 7 Credit: 2,373,140 RAC: 0 ![]() ![]() |
Any idea why the optimized Zslip app. for SSE would be about 10% slower than stock app. on an Athlon XP2800, Win. 2000 computer? I checked it 3 times and it is slower. |
Send message Joined: 26 Jul 08 Posts: 627 Credit: 94,940,203 RAC: 0 ![]() ![]() |
Any idea why the optimized Zslip app. for SSE would be about 10% slower than stock app. on an Athlon XP2800, Win. 2000 computer? It was built with Microsoft Visual Studio 2003. This version was done for testing in the first instance. SSE is only useful for single precision, so it doesn't provide any benefit for MW (using double precision calculations) and one sees the differences between the different compilers/switches used for that and the stock app. You should try the x87 version (built with an Intel Compiler), maybe that one is a bit faster than the stock app. Actually, your observation is a clear proof that the stock app is quite good actually (what some people still don't want to believe). The project implemented a lot of the code changes from the early optimized apps so the actual difference of the code is really small now. The factor ~2 (a bit more on recent CPUs, a bit less on older ones) advantage one sees with the SSE2+ versions is caused mainly by the compiler making full use of the SSE2 units (auto vectorization kicking in) and the much more efficient flat register file of SSEx instead of the stack based approach of the x87 FPU. |
![]() Send message Joined: 19 Feb 08 Posts: 350 Credit: 141,284,369 RAC: 0 ![]() ![]() |
Hi, for some reasons i had to reinstall my XP-pro. I've installed all available updates for windows, .net and ATI-software. The first application running on the new system was BOINC. I installed 0.19f, did the copy and renaming in the \system32 directory. Unfortunately my display-driver says sometimes, that it was necessary to reset the driver, which means, that standard BOINC-project resume ther work, but MW stops processing. It is necessary to stop and restart BOINC. It definitely did not happen on the old installation, the main difference is the presence of all updates. Hardware is the same (exept another HD). CCC is version 2009.0702.1239.20840, Boinc is version 6.6.36 and .net is 2.0.50727.3082. The correct message from the driver is: VPU Recover has reset your graphics accelerator as it was no longer responding to graphics driver commands. |
![]() ![]() Send message Joined: 14 Feb 09 Posts: 999 Credit: 74,932,619 RAC: 0 ![]() ![]() |
I am guessing that you installed Catalyst 9.7, we have seen issues with both 9.6 and 9.7 on VPU recovery. So far the safest bet is either 8.12 or 9.5 as both of them seem to keep running with no problem. ![]() |
![]() Send message Joined: 19 Feb 08 Posts: 350 Credit: 141,284,369 RAC: 0 ![]() ![]() |
Hi, yes, you're right. I tried to downgrade to 9.5, but this ugly message came up again. Then i removed all ATI-Software from my PC, removed the graphics-drivers, restarted my system, used CCleaner to remove all the forgotten entries in the registry (there where more than 40 entries pointing to ati-software - what a wonderful uninstaller from ATI! ), restarted again and installed the 9.1 catalyst, which needs no renaming of these three files. And yes, i have also removed the three copied and renamed files. BUT THIS UGLY MESSAGE COMES UP AGAIN !!!! Is there anything else that can be done except reinstalling windows? |
![]() ![]() Send message Joined: 9 Feb 09 Posts: 166 Credit: 27,520,813 RAC: 0 ![]() ![]() |
Werkstatt the problem is ccleaner dont remove residu from system maps I used driver cleaner in the paste but sadly they gone greedy now and is payed Thats the only tool which really cleans all crap drivers keep on your system But also a common mistake people make is not let the uninstaller do its work good, if ati stuff is there select uninstall driver and then restart. If you can find another good free driver uninstall program we would like to know that as well Its new, its relative fast... my new bicycle |
![]() Send message Joined: 19 Feb 08 Posts: 350 Credit: 141,284,369 RAC: 0 ![]() ![]() |
Sorry to tell you that. I used the ATI uninstall utility, which lets you decide to only remove CCC or ompletely remove all ATI-Software including the drivers. I did it with driver 9.7 and with 9.5. Yesterday i decided to reinstall my XP. I installed my old 3850 Graphic card and used the old 8.3 CCC, which was working before the new installation. I installed the Win_32 SSE2 in the Milkyway directory (latest 0.19 SSE2f Version). I do no longer get the VPU-Reset messages, but all WU report calculation errors. The difference to the previous system is: a different hard disk, a mainboard bios update and the windows updates from the last 14 months. I often had troubles with different programs after updating windows; this is why the updates are under suspicion. |
![]() ![]() Send message Joined: 14 Feb 09 Posts: 999 Credit: 74,932,619 RAC: 0 ![]() ![]() |
The minimum CCC version is 8.12, all others are too old to work with the CAL version included. ![]() |
![]() Send message Joined: 19 Feb 08 Posts: 350 Credit: 141,284,369 RAC: 0 ![]() ![]() |
Thanks, now it starts crunching. But the VPU Recover-Message is back! |
![]() Send message Joined: 19 Feb 08 Posts: 350 Credit: 141,284,369 RAC: 0 ![]() ![]() |
Hi, i think, i've got it! During a virussscan i retried to run milkyway .. and it worked! So it seems, heavy system usage solves the problem. I added a w1.1 in the XML-setup file and it crunches since one hour. hopepully longer ... |
Send message Joined: 15 Aug 09 Posts: 9 Credit: 318,640 RAC: 0 ![]() ![]() |
The minimum CCC version is 8.12, all others are too old to work with the CAL version included. I started using CCC 9.8 yesterday and it seems to be working great, using CMDLINE option "f20 n2". This is using a 3780 under Vista Ultimate X64. I found CCC 9.8 on the Guru3D.com enthusiast site. Still not available on the ATI support site. n2n |
![]() ![]() Send message Joined: 9 Feb 09 Posts: 166 Credit: 27,520,813 RAC: 0 ![]() ![]() |
Indeed on any system the 8.12 version is the best for running gipsels MW app except when running vista 64 there the 9.1 or 9.7 driver seems to run fine Its new, its relative fast... my new bicycle |
![]() Send message Joined: 19 Feb 08 Posts: 350 Credit: 141,284,369 RAC: 0 ![]() ![]() |
If you can find another good free driver uninstall program we would like to know that as well I've found this: http://www.guru3d.com search for driver sweeper. Remove drivers with onboard tools, restart in safe mode and use driver sweeper. Regards |
![]() Send message Joined: 19 Feb 08 Posts: 350 Credit: 141,284,369 RAC: 0 ![]() ![]() |
Hi, this is a good source for drivers! I did the same one hour ago and it seems to run fine witch standard settings. With f20 it hangs within a minute. Q9300, 3850, XP-Pro, all Updates, no tuning. Alexander |
Send message Joined: 18 Aug 09 Posts: 116 Credit: 20,355,729 RAC: 3,527 ![]() ![]() ![]() |
I just joined this project and in checking my first batch of tasks I see this message: <core_client_version>6.4.7</core_client_version> <![CDATA[ <stderr_txt> Unrecognized XML in parse_init_data_file: computation_deadline Skipping: 1250882825.136000 Skipping: /computation_deadline Unrecognized XML in GLOBAL_PREFS::parse_override: mod_time Skipping: /mod_time Unrecognized XML in GLOBAL_PREFS::parse_override: max_ncpus_pct Skipping: 100.000000 Skipping: /max_ncpus_pct </stderr_txt> ]]> one of the tasks can be found here the task completes normally, so now credit issues or anything significant that I can tell. |
Send message Joined: 4 Dec 07 Posts: 1 Credit: 8,200,729 RAC: 0 ![]() ![]() |
Its great that you make the CUDA app available to us lazy ones that don't want to do the manual installation. However, I got these error messages, and the app didn't run... Please guide me how I can solve this. mån 24 aug 2009 19.13.27|Milkyway@home|[error] Signature verification error for milkyway_0.19_x86_64-pc-linux-gnu__cuda mån 24 aug 2009 19.13.27|Milkyway@home|[error] Checksum or signature error for milkyway_0.19_x86_64-pc-linux-gnu__cuda |
![]() ![]() Send message Joined: 2 Apr 09 Posts: 16 Credit: 1,517,499 RAC: 0 ![]() ![]() |
It was built with Microsoft Visual Studio 2003. This version was done for testing in the first instance. SSE is only useful for single precision, so it doesn't provide any benefit for MW (using double precision calculations) and one sees the differences between the different compilers/switches used for that and the stock app. In fact SSE2 and up do provide packed double precision operation. (SIMD) so that might provide a good speedup. See: table B-25. formats and encodings of SSE2 floating point instructions. Intel document 253667.pdf Regards, Hydro. Team Bletchley Park. |
Send message Joined: 10 Apr 09 Posts: 4 Credit: 23,190 RAC: 0 ![]() ![]() |
Finally I found a working combination on XP32. I have a HD 4870 512MB. I used Catalyst 9.7, 0.19g with SSE2 and calculation is going on without BSOD! I have BOINC 6.10.3.0 Previously I tried: Catalyst 9.9 with 0.19f sse2, Catalyst 9.9 with 0.20 and Catalyst 9.7 with 0.20. Version 0.20 downloads only 6 packets and will not begin to work. |
©2023 Astroinformatics Group