Message boards :
Number crunching :
Update for GPU AP (both NV and ATi) to rev560
Message board moderation
Previous · 1 · 2 · 3 · 4 · Next
Author | Message |
---|---|
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
I also saw that the sbs can be set in <cmdline>-instances_per_device 2 -unroll 9 -ffa_block 8192 -ffa_block_fetch 4096 -sbs 256 -hp</cmdline> It was no effect on build included into installer. In r560 and later it will check amount of needed memory (based on unroll value) vs limits imposed with this -sbs switch and issue warning if requested memory exceeds imposed limit. Memory will be allocated anyway for now, but user at least will know that app usus more memory for single buffer than he/she thought. |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
You should increase unroll factor to 10 at least to get better speed on your card. Please look into app_info's <cmdline> tag. It contains mentioned params with some default values. Increase them as Mike suggested. For more info please look into app's release notes. |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
I use 16 for my HD6950 for few months already w/o issues. Perhaps bigger possible too but still to test. |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
Thanks for correction. As non-native english speaker I tend to do mistakes :D |
TRuEQ & TuVaLu Send message Joined: 4 Oct 99 Posts: 505 Credit: 69,523,653 RAC: 10 |
I've read somewhere that DATA_CHUNK_UNROLL should be setted to: half of the Max compute number units: Yes, it is supposed to be "set" and not "setted". I wonder why I didn't see that. It is often(offen) after I've written something and then click ok to post it I'v(e) need to click on the "edit" button to fix(correct) something I wrote(I've written) incorrectly or just just missed to write something(did I wrote that?)... I like (to) write stuff in English and then see the errors I make. It's kinda fun. It took me several seconds to understand "smth" was actually something. But I only read that. :) Thank you for all the (i)nfo(rmation) here. It makes sence. //TRuEQ //TRuEQ TRuEQ & TuVaLu |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
It's in there among cast, cost, cut, hit, hurt, let, set, shut and slit. Possibly some others too. Just check the list. And here ended my grammar lessons. :) |
perryjay Send message Joined: 20 Aug 02 Posts: 3377 Credit: 20,676,751 RAC: 0 |
Well, I finished one.http://setiathome.berkeley.edu/result.php?resultid=2388692084 Waiting for my wingman to see how I did. Looks okay but I'm a little worried about a message I got. Did I mess up putting everything in place? INFO: can't open binary kernel file, continue with recompile... PROUD MEMBER OF Team Starfire World BOINC |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 |
Well, I finished one.http://setiathome.berkeley.edu/result.php?resultid=2388692084 Waiting for my wingman to see how I did. Looks okay but I'm a little worried about a message I got. Did I mess up putting everything in place? The latest app's compile a couple of files the first time they run, you should have similar files to these (But with your GPU or CPU name instead): AstroPulse_Kernels_r560.cl_GeForce GTX 460.bin_V6 oclFFTplan_GeForce GTX 460_32768_r560.bin r560_IntelRCoreTMi72600KCPU340GHz.wisdom Claggy |
X-Files 27 Send message Joined: 17 May 99 Posts: 104 Credit: 111,191,433 RAC: 0 |
This system doesn't suffer from excessive CPU usage: Whats bugling me is that I can't replicate this low cpu usage on my other rig - always settled at 85%. The good thing is that there's no more sleep bug in this NV driver version. |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
Hehe, one more typo noticed, thanks :) |
Fred E. Send message Joined: 22 Jul 99 Posts: 768 Credit: 24,140,697 RAC: 0 |
My first one validated this morning, so it is looking good. Thought it interesting that it validated against an ATI 2600 with only 256MByte memory. Workunit is http://setiathome.berkeley.edu/workunit.php?wuid=966607530 -- won't be there long. Glad to see something go right after a rough week having to replace the power supply & Motherboard. Looks like another winner, Raistmer. Hope the Nvidia driver issues get resolved someday soon. If there are typos here, notify me at autodiscard@trash.com. :=) Another Fred Support SETI@home when you search the Web with GoodSearch or shop online with GoodShop. |
betreger Send message Joined: 29 Jun 99 Posts: 11361 Credit: 29,581,041 RAC: 66 |
If that typo is the worst thing you do this week you will have an excellent week! |
Mike Send message Joined: 17 Feb 01 Posts: 34258 Credit: 79,922,639 RAC: 80 |
My first one validated this morning, so it is looking good. Thought it interesting that it validated against an ATI 2600 with only 256MByte memory. She is running Brook+ version. With each crime and every kindness we birth our future. |
petri33 Send message Joined: 6 Jun 02 Posts: 1668 Credit: 623,086,772 RAC: 156 |
Hi, If there were an APV6-executable for NV560ti448 on 64bit linux what kind of performance could I expect from it? My rig uses CPU for APV6(AVX) and setiathomeEnhanced(ssse3) and GPU for setiathomeEnhanced. The APV6 with AVX gives this kind of performance (the host 5643864): http://setiathome.berkeley.edu/workunit.php?wuid=966928760 petri33 To overcome Heisenbergs: "You can't always get what you want / but if you try sometimes you just might find / you get what you need." -- Rolling Stones |
Josef W. Segur Send message Joined: 30 Oct 99 Posts: 4504 Credit: 1,414,761 RAC: 0 |
Hi, The GPU performance should not be affected much by what OS the CPU is running, perhaps these valid AP v6 tasks on a 560ti under Windows give a good enough approximation. As you know, the 560ti448 should outperform that regular 560ti somewhat. Joe |
Horacio Send message Joined: 14 Jan 00 Posts: 536 Credit: 75,967,266 RAC: 0 |
I have a question about the NV 560, Process Lasso show this app as having a default affinity to cores 0,1 while all the other tasks running from boinc have the default afinity set to all cores... Is this on pourpouse? I mean, can I extend the affinity to all cores so it dosnt get restricted? Can this improove (or break) the performance? |
BilBg Send message Joined: 27 May 07 Posts: 3720 Credit: 9,385,827 RAC: 0 |
<cmdline>-no_cpu_lock -instances_per_device 1 ......</cmdline> Â - ALF - "Find out what you don't do well ..... then don't do it!" :) Â |
JohnDK Send message Joined: 28 May 00 Posts: 1222 Credit: 451,243,443 RAC: 1,127 |
So what's the advantage/disadvantage using that option or not using it? |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
So what's the advantage/disadvantage using that option or not using it? If your GPU app is limited to CPUs 0 & 1 which are also running tasks then when you are loading work into the GPU it has to share CPU time with the other process while loading. With very fast GPUs this could add up to a significant amount of time in the long run. Even though it may only be a few seconds per task. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
Mike Send message Joined: 17 Feb 01 Posts: 34258 Credit: 79,922,639 RAC: 80 |
Correct in principle. It helps very much on high blanked units. Since blanking is calculated by CPU its switching between cores while computing. Blanking sometimes is fragmented so it uses different cores. With each crime and every kindness we birth our future. |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.