Message boards :
Graphics cards (GPUs) :
Performance of 3D Graphic @ PS3GRID
Message board moderation
Previous · 1 · 2 · 3 · 4 · Next
| Author | Message |
|---|---|
|
Send message Joined: 17 Aug 08 Posts: 2705 Credit: 1,311,122,549 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Quattro GVS 290 will take 175 hours :-(( It's actually difficult to find out what this piece of hardware is, but wiki says it uses a NVG86, which is basically the same as the slow 16-shader Geforce 8400. Crunching is not what these are made for. MrS Scanning for our furry friends since Jan 2002 |
|
Send message Joined: 26 Aug 08 Posts: 55 Credit: 1,475,857 RAC: 0 Level ![]() Scientific publications ![]() ![]()
|
Quattro GVS 290 will take 175 hours :-(( I'd like to see how a tesla S1070 GPU would perform here. 960 cores, 16 GB ram, and 4.3 Tflops assumed to be based on a 280 GTX. Less than 2 hours per task? I never was an ATI fan even after they released the 2 Tflops 4870x2, after 2 generations of bad experience with ATI and 4 good experiences with nvidias, I'm sticking with nvidia. That, and its industry wide acceptance from being run on PS3 and Xbox and its logo being slapped into every game! |
|
Send message Joined: 17 Aug 08 Posts: 2705 Credit: 1,311,122,549 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
The Tesla S1070 uses 4 GT200 chips with more memory than on the desktop cards (4GB per chip). I don't know if the software sees them as 4 seperate GPUs or as a single one. In the first case you'd get about 4 WUs every 7h, in the second it should be a WU in less than 2h. Well, if you spend 8000$ you'd better get some serious performance for your money! (its got the shaders clocked at 1.5 GHz instead of 1.3 GHz on the 280GTX) I never was an ATI fan I doesn't matter if your arguments are valid or not, please don't mention this topic here. Experience shows that any sane, technical conversation can be turned into a flame war within seconds, if the other company is mentioned ;) MrS Scanning for our furry friends since Jan 2002 |
[XTBA>XTC] ZeuZSend message Joined: 15 Jul 08 Posts: 60 Credit: 108,384 RAC: 0 Level ![]() Scientific publications
|
Can you say more about this please? what does "soon" mean for you? next week? next month? next year? xD Thanks XTC_ZeuZ |
GDFSend message Joined: 14 Mar 07 Posts: 1958 Credit: 629,356 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() |
It does not depends on me, but on BOINC. We have at least to wait for version 6.3.11. The application and server are ready. I would say one or two weeks. gdf |
[XTBA>XTC] ZeuZSend message Joined: 15 Jul 08 Posts: 60 Credit: 108,384 RAC: 0 Level ![]() Scientific publications
|
Ah ok, thank you :D |
|
Send message Joined: 17 Aug 08 Posts: 2705 Credit: 1,311,122,549 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Guys, I'm confused! 2 Questions. First: how fast is the GT200? The data which Thomas collected, i.e. 25.000 and 28.000 s/WU, seems to be the norm. That means a GT200 shader does considerably more work per clock cycle than a G9x shader. However, I just stoumbled across these 2 machines (1 from Stefan & Nr 2). The first uses a stock GTX260 and Vista and with 6.3.10 and 6.43 it needs 35.000 - 36.000 s/WU. That's what would be expected if its shaders were as fast as G92 per clock and I'd figure in a 20% bonus for the Vista driver. The other machine is a OC'ed GTX280 and produced one 35.200 s result. Maybe it's just the odd man out. The other question: is the theory about the Vista driver being ~20% faster really true? So far there was nothing which forced me to that conclusion, but nothing contradicted it, so it was a reasonable thing to assume. And.. damn! I accidently closed the tab. Well, I found a machine which did not behave as expected but forgot the details. MrS Scanning for our furry friends since Jan 2002 |
|
Send message Joined: 16 Jul 07 Posts: 209 Credit: 5,496,860,456 RAC: 8,998 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
Well, I just finished my first task on a GPU, and am concerned with the result. It took 58437.03 seconds, or 16.23 hours, which is about twice as long as I expected. Am I doing something wrong? http://www.ps3grid.net/result.php?resultid=56210 stderr out : <core_client_version>6.3.10</core_client_version> <![CDATA[ <stderr_txt> # Using CUDA device 0 # Device 0: "GeForce GTX 280" # Clock rate: 1404000 kilohertz MDIO ERROR: cannot open file "restart.coor" called boinc_finish </stderr_txt> ]]> Machine: Q6600 (@2.7ghz), 6gb RAM, running XP64; BFG GTX 280 BFGEGTX2801024OC2E; BOINC 6.3.10; Driver is 177.41_geforce_winxp_64bit_english_whql, freshly downloaded yesterday. I ran this task with no other tasks running. It is a dedcated cruncher, so nothing but BOINC is running on it. Reno, NV Team: SETI.USA |
Stefan LedwinaSend message Joined: 16 Jul 07 Posts: 464 Credit: 298,573,998 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
Over 16 hours is a little bit slow for that card... You could try newer drivers (177.84), from the CUDA download site - Direct link to Win64 177.84 driver -http://www.nvidia.com/object/thankyou.html?url=/compute/cuda/2_0/windows/driver/NVIDIADisplayWin2KAMD64(177_84)Int.exe. pixelicious.at - my little photoblog |
GDFSend message Joined: 14 Mar 07 Posts: 1958 Credit: 629,356 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() |
Well, I just finished my first task on a GPU, and am concerned with the result. It took 58437.03 seconds, or 16.23 hours, which is about twice as long as I expected. Am I doing something wrong? I have looked at it. The new windows application coming up will report the time per step which should tells us what's wrong with your system. gdf |
|
Send message Joined: 16 Jul 07 Posts: 209 Credit: 5,496,860,456 RAC: 8,998 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
I have looked at it. The new windows application coming up will report the time per step which should tells us what's wrong with your system. Thanks. When do you expect it to be released? For what it's worth, I installed the Nvidia tools, and took a look at the GPU settings. It reports everything as expected: GPU Core: 650 MHz GPU mem speed: 1163 MHz GPU shader clock: 1404 MHz These are all factory settings. I have not changed anything. Reno, NV Team: SETI.USA |
|
Send message Joined: 17 Aug 08 Posts: 2705 Credit: 1,311,122,549 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
I'd also say give the new drivers a shot. Either 177.84 or 177.92. MrS Scanning for our furry friends since Jan 2002 |
|
Send message Joined: 24 Aug 08 Posts: 45 Credit: 3,431,862 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]()
|
I'd also say give the new drivers a shot. Either 177.84 or 177.92. The newest is 177.98, Running here under VISTA 64 fine, |
[SETI.USA]Tank_MasterSend message Joined: 8 Jul 07 Posts: 85 Credit: 67,463,387 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
Where did you get the 177.98? I still only see 177.92 when I select beta drivers from nvidia's site for vista x64. |
|
Send message Joined: 17 Aug 08 Posts: 2705 Credit: 1,311,122,549 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Any performance changes? MrS Scanning for our furry friends since Jan 2002 |
|
Send message Joined: 24 Aug 08 Posts: 45 Credit: 3,431,862 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]()
|
Where did you get the 177.98? I still only see 177.92 when I select beta drivers from nvidia's site for vista x64. I took the driver from this side: http://www.laptopvideo2go.com/forum/index.php?showforum=94 |
|
Send message Joined: 24 Aug 08 Posts: 45 Credit: 3,431,862 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]()
|
Any performance changes? At the same time I overclocked my card, so I can't answer your question. |
|
Send message Joined: 16 Jul 07 Posts: 209 Credit: 5,496,860,456 RAC: 8,998 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
I'd also say give the new drivers a shot. Either 177.84 or 177.92. Good call. Looks like I will be down to a bit over 6 hours when my current task finishes. Reno, NV Team: SETI.USA |
|
Send message Joined: 16 Jul 07 Posts: 209 Credit: 5,496,860,456 RAC: 8,998 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
I'd also say give the new drivers a shot. Either 177.84 or 177.92. Well, I am bottoming out around 24,600 seconds, or 6.8 hours, with 177.92. Much better than 16+ hours! http://www.ps3grid.net/result.php?resultid=56209 I understand that Vista is supposed to be faster than XP. But how does the speed of XP64 compare to Linux64? Thanks again for all the help! Reno, NV Team: SETI.USA |
koschiSend message Joined: 14 Aug 08 Posts: 127 Credit: 913,858,161 RAC: 13 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
The speed is almost the same. One member of our team has a 9800GTX+ (XP64) which is only little better clocked than my 8800GTS OC (Linux) and we have nearly the same time at around 44000 seconds. |
©2025 Universitat Pompeu Fabra