読者です 読者をやめる 読者になる 読者になる

だ・か・らっ、Dia“l”yだってばさ!

これは『戯れ言』です。また、“Diary”ではなく“Dialy”です。つまり、日記に似て非なるものです。 所謂『日記』ではありません。お間違えの無いようお願いします。(^^;A

9500 GTでも試してみた

dnetc CUDA

LSTな8.04にCUDA入れ直すの苦労したけど、まなんとかなった。
#つか、そのまんま動けや、ゴルァ


dnetc511beta7-linux-amd64-cuda22 on CUDA 2.2

$ ./dnetc -bench rc5-72

distributed.net client for CUDA 2.2 on Linux Copyright 1997-2009, distributed.net
Please visit http://www.distributed.net/ for up-to-date contest information.
Start the client with '-help' for a list of valid command line options.


dnetc v2.9105-511-CTL-09070517-*dev* for CUDA 2.2 on Linux (Linux 2.6.24-22-generic).
Please provide the *entire* version descriptor when submitting bug reports.
The distributed.net bug report pages are at http://bugs.distributed.net/
Using email address (distributed.net ID) 'sr@hyper.cx'

[Jul 08 02:42:38 UTC] RC5-72: using core #0 (CUDA 1-pipe 64-thd).
[Jul 08 02:42:56 UTC] RC5-72: Benchmark for core #0 (CUDA 1-pipe 64-thd)
                      0.00:00:16.13 [10,957,753 keys/sec]
[Jul 08 02:42:56 UTC] RC5-72: using core #1 (CUDA 1-pipe 128-thd).
[Jul 08 02:43:15 UTC] RC5-72: Benchmark for core #1 (CUDA 1-pipe 128-thd)
                      0.00:00:16.57 [15,924,182 keys/sec]
[Jul 08 02:43:15 UTC] RC5-72: using core #2 (CUDA 1-pipe 256-thd).
[Jul 08 02:43:34 UTC] RC5-72: Benchmark for core #2 (CUDA 1-pipe 256-thd)
                      0.00:00:16.64 [23,040,115 keys/sec]
[Jul 08 02:43:34 UTC] RC5-72: using core #3 (CUDA 2-pipe 64-thd).
[Jul 08 02:43:53 UTC] RC5-72: Benchmark for core #3 (CUDA 2-pipe 64-thd)
                      0.00:00:16.60 [14,971,520 keys/sec]
[Jul 08 02:43:53 UTC] RC5-72: using core #4 (CUDA 2-pipe 128-thd).
[Jul 08 02:44:12 UTC] RC5-72: Benchmark for core #4 (CUDA 2-pipe 128-thd)
                      0.00:00:16.38 [21,490,274 keys/sec]
[Jul 08 02:44:12 UTC] RC5-72: using core #6 (CUDA 4-pipe 64-thd).
[Jul 08 02:44:31 UTC] RC5-72: Benchmark for core #6 (CUDA 4-pipe 64-thd)
                      0.00:00:16.48 [22,393,084 keys/sec]
[Jul 08 02:44:31 UTC] RC5-72: using core #7 (CUDA 4-pipe 128-thd).
[Jul 08 02:44:50 UTC] RC5-72: Benchmark for core #7 (CUDA 4-pipe 128-thd)
                      0.00:00:17.32 [25,359,169 keys/sec]
[Jul 08 02:44:50 UTC] RC5-72: using core #9 (CUDA 1-pipe 64-thd busy wait).
[Jul 08 02:45:09 UTC] RC5-72: Benchmark for core #9 (CUDA 1-pipe 64-thd busy wait)
                      0.00:00:16.29 [13,985,342 keys/sec]
[Jul 08 02:45:09 UTC] RC5-72: using core #10 (CUDA 1-pipe 64-thd sleep 100us).
[Jul 08 02:45:28 UTC] RC5-72: Benchmark for core #10 (CUDA 1-pipe 64-thd sleep 100us)
                      0.00:00:16.26 [11,942,234 keys/sec]
[Jul 08 02:45:28 UTC] RC5-72: using core #11 (CUDA 1-pipe 64-thd sleep dynamic).
[Jul 08 02:45:48 UTC] RC5-72: Benchmark for core #11 (CUDA 1-pipe 64-thd sleep dynamic)
                      0.00:00:17.49 [10,726,932 keys/sec]
[Jul 08 02:45:48 UTC] RC5-72 benchmark summary :
                      Default core : #0 (CUDA 1-pipe 64-thd)
                      Fastest core : #7 (CUDA 4-pipe 128-thd)
[Jul 08 02:45:48 UTC] Core #7 is significantly faster than the default core.
                      The CUDA core selection has been made as a tradeoff between cor ...
                      and responsiveness of the graphical desktop.
                      Please file a bug report along with the output of -cpuinfo
                      only if the the faster core selection does not degrade graphics ...

遅い・・・
遅過ぎる。


CUDAのVer. を2.1落として、dnetc511beta7-linux-amd64-cuda21

$ ./dnetc -bench rc5-72

distributed.net client for CUDA 2.1 on Linux Copyright 1997-2009, distributed.net
Please visit http://www.distributed.net/ for up-to-date contest information.
Start the client with '-help' for a list of valid command line options.


dnetc v2.9105-511-CTL-09070517-*dev* for CUDA 2.1 on Linux (Linux 2.6.24-22-generic).
Please provide the *entire* version descriptor when submitting bug reports.
The distributed.net bug report pages are at http://bugs.distributed.net/
Using email address (distributed.net ID) 'sr@hyper.cx'

[Jul 08 11:21:53 UTC] RC5-72: using core #0 (CUDA 1-pipe 64-thd).
[Jul 08 11:22:12 UTC] RC5-72: Benchmark for core #0 (CUDA 1-pipe 64-thd)
                      0.00:00:16.63 [13,245,156 keys/sec]
[Jul 08 11:22:12 UTC] RC5-72: using core #1 (CUDA 1-pipe 128-thd).
[Jul 08 11:22:31 UTC] RC5-72: Benchmark for core #1 (CUDA 1-pipe 128-thd)
                      0.00:00:17.13 [20,167,043 keys/sec]
[Jul 08 11:22:31 UTC] RC5-72: using core #2 (CUDA 1-pipe 256-thd).
[Jul 08 11:22:50 UTC] RC5-72: Benchmark for core #2 (CUDA 1-pipe 256-thd)
                      0.00:00:16.65 [31,524,549 keys/sec]
[Jul 08 11:22:50 UTC] RC5-72: using core #3 (CUDA 2-pipe 64-thd).
[Jul 08 11:23:08 UTC] RC5-72: Benchmark for core #3 (CUDA 2-pipe 64-thd)
                      0.00:00:16.06 [21,081,654 keys/sec]
[Jul 08 11:23:08 UTC] RC5-72: using core #4 (CUDA 2-pipe 128-thd).
[Jul 08 11:23:28 UTC] RC5-72: Benchmark for core #4 (CUDA 2-pipe 128-thd)
                      0.00:00:16.66 [25,095,716 keys/sec]
[Jul 08 11:23:28 UTC] RC5-72: using core #6 (CUDA 4-pipe 64-thd).
[Jul 08 11:23:46 UTC] RC5-72: Benchmark for core #6 (CUDA 4-pipe 64-thd)
                      0.00:00:16.08 [33,765,960 keys/sec]
[Jul 08 11:23:46 UTC] RC5-72: using core #7 (CUDA 4-pipe 128-thd).
[Jul 08 11:24:05 UTC] RC5-72: Benchmark for core #7 (CUDA 4-pipe 128-thd)
                      0.00:00:16.11 [39,352,143 keys/sec]
[Jul 08 11:24:05 UTC] RC5-72: using core #9 (CUDA 1-pipe 64-thd busy wait).
[Jul 08 11:24:24 UTC] RC5-72: Benchmark for core #9 (CUDA 1-pipe 64-thd busy wait)
                      0.00:00:17.20 [21,855,901 keys/sec]
[Jul 08 11:24:24 UTC] RC5-72: using core #10 (CUDA 1-pipe 64-thd sleep 100us).
[Jul 08 11:24:44 UTC] RC5-72: Benchmark for core #10 (CUDA 1-pipe 64-thd sleep 100us)
                      0.00:00:16.97 [17,871,031 keys/sec]
[Jul 08 11:24:44 UTC] RC5-72: using core #11 (CUDA 1-pipe 64-thd sleep dynamic).
[Jul 08 11:25:02 UTC] RC5-72: Benchmark for core #11 (CUDA 1-pipe 64-thd sleep dynamic)
                      0.00:00:16.28 [7,554,771 keys/sec]
[Jul 08 11:25:02 UTC] RC5-72 benchmark summary :
                      Default core : #0 (CUDA 1-pipe 64-thd)
                      Fastest core : #7 (CUDA 4-pipe 128-thd)
[Jul 08 11:25:02 UTC] Core #7 is significantly faster than the default core.
                      The CUDA core selection has been made as a tradeoff between cor ...
                      and responsiveness of the graphical desktop.
                      Please file a bug report along with the output of -cpuinfo
                      only if the the faster core selection does not degrade graphics ...

かなりマシ。


Win版にはCUDA 2.0用clientがあるみたいですねぇ。
きっと速いんでしょうねぇ。
羨ましいですねぇ。