Linux app benchmarks, just for curiosity
Message boards :
Number crunching :
Linux app benchmarks, just for curiosity
Message board moderation
| Author | Message |
|---|---|
Gene  Send message Joined: 26 Apr 99 Posts: 150 Credit: 48,393,279 RAC: 118 
|
With stock Linux+Nvidia now available I was curious to see how they would compare with other Linux applications. So, I ran two different work units on four Linux apps, using the KWSN benchmark script, and the results are in the table below. Yeah, I know, its a pitiful low-end GTX650 and the rest of the system details can be found by looking at my account page(s). Abbreviations are used in the table, so here are the full descriptions: Work Units: 13my (short name) was 13my10ae.6234.266441.8.35.119 AR=.4438 blc4 (short name) was blc4_2bit_guppi_57451_26010_HIP69732_OFF_0022.\ 9982.831.18.27.191.vlar AR=.00624 Applications: CPU 8.00 = setiathome_8.00_x86_64-pc-linux-gnu OpenCL 8.10 = setiathome_8.10_x86_64-pc-linux-gnu__opencl_nvidia_sah SoG 8.10 = setiathome_8.10_x86_64-pc-linux-gnu__opencl_nvidia_SoG CUDA x41zi = setiathome_x41zi_x86_64-pc-linux-gnu_cuda60 All runs done in an otherwise idle system. No command line parameters were used for any application. Elapsed times and Flop counts listed as reported in stderr output. All results were evaluated by the benchmark compare routine as "Strongly similar" with Q > 99.19% . Some of the values are not what I expected and I don't have any insights to share with respect to the relative application performance. I just did the experiment and here's the data! Don't jump to any wild conclusions. This is just a statistical sample of ONE!
work unit CPU OpenCL SoG CUDA
8.00 8.10 8.10 x41zi
blc4 6965 sec 14907 sec 6851 sec 3547 sec
37610 Gflp 12443 Gflp 3818 Gflp 37627 Gflp
13my 11030 sec 2364 sec 1704 sec 1479 sec
37076 Gflp 18711 Gflp 6866 Gflp 38171 Gflp
ID: 1790826 ·  |
 jason_gee  Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 
|
Looks cool. I don't recall making a zi build yet, so someone else made that shit up, (which I'm fine with). Following the state of the art as maybe you are trying to do (?) is a noble prospect, and if you like that prepare for some interesting times in 2016 :D "Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions. ID: 1790830 · |
|
Gene Send message Joined: 26 Apr 99 Posts: 150 Credit: 48,393,279 RAC: 118
|
The x41zi Linux app file was downloaded from jgopt.org on or about 19 February. Full download file name: setiathome_x41zi_x86_64-pc-linux-gnu_cuda60.7z Unzipped, installed, running as anonymous platform, no problems so far and obviously happy with performance on the old GTX650. ID: 1791186 · |
|
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0
|
The x41zi Linux app file was downloaded from jgopt.org on or about 19 February. *scratching head*, My Bad! Naturally I run the same build :S Guess I need to pull the versions into line so I don't lose track of which builds are which :) Glad it works for you, seems to go pretty well on the 680 + Ubuntu here as well. "Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions. ID: 1791195 · |
©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.