1) Questions and Answers : GPU applications : nvidia ion (Message 961222) Posted 6 Jan 2010 by enusbaum Post: I was able to get the CUDA version of SETI@Home running on my Zotac ION bassed system by setting the video RAM in the BIOS to 512 (the default is 256). You might want to poke around as I'm sure you probably have the same option. http://setiathome.berkeley.edu/show_host_detail.php?hostid=5252538

2) Message boards : Number crunching : Astropulse finished... I can retire now. (Message 904382) Posted 6 Jun 2009 by enusbaum Post: Also, lest we forget that a lot of AP units expire and have to be re-sent. So although the current queue is empty, data will be re-sent because it expired on the 1st pass. Keep the faith! :) I have a Quad Xeon box dedicated to only AP

3) Message boards : Number crunching : TW's new bandwith cap (Message 883557) Posted 8 Apr 2009 by enusbaum Post: Yeah, if it weren't for the Price/Speed offered by Time Warner for Road Runner, I'd dump the service so fast! If they put a cap limit on it, the price per gigabyte is too high! I honestly have no idea what they're thinking...

4) Message boards : Number crunching : TW's new bandwith cap (Message 883378) Posted 8 Apr 2009 by enusbaum Post: I broke down the cost of the new Time Warner plans in my blog: http://www.enusbaum.com/blog/2009/04/07/time-warner-cable-have-they-gone-mental/ Basically, you'd get way more monthly transfer if you went dialup ;) Also, depending on how you look at the numbers, getting a dedicated DS-3 would be cheaper than using Road Runner! LOL! :D

5) Message boards : Number crunching : AP WUs - when will they ever be available again? (Message 861990) Posted 4 Feb 2009 by enusbaum Post: My BOINC client just got a (single) AP unit! Guess the splitter is back online! :)

6) Questions and Answers : GPU applications : JACT (Jet Another Cuda Thread): GPU undeperforms when all CPU cores are processing WUs (Message 847441) Posted 31 Dec 2008 by enusbaum Post: Another thing to keep in mind is the number of Stream Processors available to the CUDA interface depending on your video card. - 8300GS only has 8 Stream Processors with a Shader Clock of 900Mhz - 8600GT only has 32 Stream Processors with a Shader Clock of 1.19Ghz - 8800GTX has 128 Stream Processors with a Shader Clock of 1.35Ghz So you can see that just using the 8xxx series of cards for an example, there's a drastic difference in processing power depending on the card you're using. Just on raw numbers, the 8800GTX is over 20x faster than an 8300GS. For this reason I believe, nVidia left the 8300GS off it's list of "CUDA Enabled Devices", which is why you're only able to run CUDA tasks on the 8600GT you have installed (since running them on the 8300GS would be PAINFULLY slow) :)

7) Questions and Answers : GPU applications : JACT (Jet Another Cuda Thread): GPU undeperforms when all CPU cores are processing WUs (Message 847436) Posted 31 Dec 2008 by enusbaum Post: From what I've clocked so far, it seems that my 8800GTX is 3.4x faster than a single Xeon X5355 core clocked at 2.66Ghz with the AK v8 optimized (SSSE3) version. So from that, you can say that a single GeForce 8800GTX (768MB edition) is almost as fast as a Quad Core Xeon X5355 @ 2.66ghz. Not too shabby! (This is all based on wall clock comparisons for comparable SETI@Home classic work units)

8) Message boards : Number crunching : if you think cuda is fast (Message 846105) Posted 28 Dec 2008 by enusbaum Post: I think it'll all boil down to several things. Mostly the Drivers and the actual Shader Processors ability to process. If the 1600 Shader Processors have a limited instruction set, it might take 10 Shader Processors to compute the same amount of data as a single nVidia Stream Processor. I'm not saying I have any basis for these estimates, but I've yet to see good benchmarks comparing the two.

9) Message boards : Number crunching : if you think cuda is fast (Message 846090) Posted 28 Dec 2008 by enusbaum Post: It's not a 1:1 ratio Shader Processors != Stream Processors It'll still be interesting to see how they stack up in a science app side by side.

10) Message boards : Number crunching : Modified SETI MB CUDA + opt AP package for full GPU utilization (Message 845548) Posted 27 Dec 2008 by enusbaum Post: I also think the issue is partly driver related. I'm running an 8800GTX on the 180.84 beta drivers and started getting nothing but -9 overflows until I rebooted and that seemed to also fix the problem. So this all might be just a combination of immature drivers and also the API perhaps not clearing our resources after a crash?? I'm not too familiar with the CUDA API.

11) Questions and Answers : GPU applications : Is your CUDA client crashing/giving computation errors? (Message 844484) Posted 24 Dec 2008 by enusbaum Post: I had the same problem. I tried three different drivers (178.74, 180.48, 180.84) and also updated to the LATEST BOINC client (6.5.0) with no help to the issue. What fixed my issue? Cooling! I installed an application called SpeedFan which told me while a SETI CUDA client was running that my GPU temp was getting as high as 90c (194f)! Holy smokes! It seems that the nVidia drivers for whatever reason don't realize that CUDA work is spiking the GPU (because it's not DirectX or OpenGL) so it keeps the fan speed low as to not cause too much noise. The problem? The heat then causes system instability! I even had my Vista system reboot on me with a BSOD! The solution? An application RivaTuner. It allowed me to set my GPU fan at 100% permanently which has solved my issue (for now at least)! My GPU temp now hovers around 65c (149f) which is hot, but within safe limits :) My only request to the SETI CUDA developers would be to hook into the nvidia API and set the fan speed while crunching GPU units. Please give this solution a go and let me know if it works for anyone else! I think GPU computing is the future for SETI and if we can stabilize the situation quickly it'll be much better for the project! Cheers! SpeedFan: http://www.almico.com/sfdownload.php Riva Tuner: http://downloads.guru3d.com/downloadget.php?id=163&file=7&evp=2d213e528fb556c354432a3976bff55a

12) Questions and Answers : GPU applications : nvidia drivers not responding (Message 844467) Posted 24 Dec 2008 by enusbaum Post: I'm having the exact same driver issue with Vista 32 bit. I'm running the latest 180.84 drivers on an 8800GTX.

13) Questions and Answers : GPU applications : The new CUDA device and Boinc. (Message 844466) Posted 24 Dec 2008 by enusbaum Post: I've had the same issue where the CUDA client ran fine for the first day or so then I started to get 'Computation Errors' and then Vista would report that the video driver is having problems. I came in just now and Vista reported that it just rebooted and recovered from a BSOD :( I'm running the latest nvidia drivers (180.48) on a 8800GTX.... soooo.. what gives??

14) Message boards : Number crunching : Modified SETI MB CUDA + opt AP package for full GPU utilization (Message 844388) Posted 24 Dec 2008 by enusbaum Post: it seemed to be working fine for a day or two but now it seems i'm getting a slew of computational errors: http://setiathome.berkeley.edu/result.php?resultid=1098999346 Cuda error 'find_pulse_kernel2<5, true>' in file 'd:/BTR/seticuda/seti_boinc/client/cuda/cudaAcc_pulsefind.cu' in line 1178 : unknown error. this only started to happen recently (in the last day). perhaps something different with the work units that are being sent? i have an 8800GTX stock. i updated the nvidia drivers to latest (180.48) and also installed the latest BOINC client (6.5.0) but i'm still getting the error. any thoughts?? :) thanks again for your hard work and port of this application! :)

15) Message boards : Number crunching : Intel P4 Hyperthreading (Message 2711) Posted 29 Jun 2004 by enusbaum Post: The point I think people are missing is that a HyperThreaded CPU is not TWO physical CPUs. Your maximum processing speed is the speed of the processor. There are a lot of cache control optimizations on the chip itself, so thats why you see a slight performance increase when you enable HT, but, for the most part, you only have ONE CPU. So when you run two WUs on a 3Ghz HT CPU, each WU isnt getting 3Ghz. Each WU is (roughly) getting 1.5Ghz (with cache control, it might be 1.6, but this is just for example). Don't think that when you enable HT your computer gives each WU 3Ghz, thus becoming a 6Ghz processor.. because that just isnt how it works. This is the reason why your benchmarks are for the most part HALFED and why work units almost take twice as long on an HT CPU. It's not code optimizations or anything of that nature which is causing BOINC to run 'slow' when HT is enabled, you just have two threads running that are both using 50% of your ONE CPU (each 50% being a 'Virtual CPU').

 
©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.