Message boards :
Number crunching :
AP r2083
Message board moderation
Author | Message |
---|---|
cov_route Send message Joined: 13 Sep 12 Posts: 342 Credit: 10,270,618 RAC: 0 |
I noticed a new version of AP for ATI on Mikes site, r2083. I benched it with Clean_01LC.wu against r1843. Run times are nearly identical at 45 sec but the CPU usage has fallen dramatically from 15 sec to 3 sec. Intriguing. I'll bench some more wu's. I wonder if there is a reason that might happen. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
I noticed a new version of AP for ATI on Mikes site, r2083. I benched it with Clean_01LC.wu against r1843. Run times are nearly identical at 45 sec but the CPU usage has fallen dramatically from 15 sec to 3 sec. I think Raistmer learned some new facts about GPU synching from Oliver Bock at Einstein. http://einstein.phys.uwm.edu/forum_thread.php?id=10215&nowrap=true#127648 |
cov_route Send message Joined: 13 Sep 12 Posts: 342 Credit: 10,270,618 RAC: 0 |
Some more wu's tested, the CPU usage is down by a factors between 3 and 8. Quick timetable WU : ap_18se08aa_B6_P1_00046_1LC25.wu AP6_win_x86_SSE2_OpenCL_ATI_r1843.exe -unroll 10 -ffa_block 2048 -ffa_block_fetch 1024 : Elapsed 46.817 secs CPU 16.969 secs AP6_win_x86_SSE2_OpenCL_ATI_r2083.exe -unroll 10 -ffa_block 2048 -ffa_block_fetch 1024 : Elapsed 47.014 secs, speedup: -0.42% ratio: 1.00x CPU 4.563 secs, speedup: 73.11% ratio: 3.72x WU : Clean_01LC.wu AP6_win_x86_SSE2_OpenCL_ATI_r1843.exe -unroll 10 -ffa_block 2048 -ffa_block_fetch 1024 : Elapsed 45.815 secs CPU 15.563 secs AP6_win_x86_SSE2_OpenCL_ATI_r2083.exe -unroll 10 -ffa_block 2048 -ffa_block_fetch 1024 : Elapsed 45.086 secs, speedup: 1.59% ratio: 1.02x CPU 3.016 secs, speedup: 80.62% ratio: 5.16x WU : Clean_20LC.wu AP6_win_x86_SSE2_OpenCL_ATI_r1843.exe -unroll 10 -ffa_block 2048 -ffa_block_fetch 1024 : Elapsed 843.815 secs CPU 282.313 secs AP6_win_x86_SSE2_OpenCL_ATI_r2083.exe -unroll 10 -ffa_block 2048 -ffa_block_fetch 1024 : Elapsed 847.676 secs, speedup: -0.46% ratio: 1.00x CPU 33.953 secs, speedup: 87.97% ratio: 8.31x WU : short_ap_21oc08ab_B2_P0_00081_20081130_08605.wu AP6_win_x86_SSE2_OpenCL_ATI_r1843.exe -unroll 10 -ffa_block 2048 -ffa_block_fetch 1024 : Elapsed 37.058 secs CPU 7.109 secs AP6_win_x86_SSE2_OpenCL_ATI_r2083.exe -unroll 10 -ffa_block 2048 -ffa_block_fetch 1024 : Elapsed 37.033 secs, speedup: 0.07% ratio: 1.00x CPU 2.406 secs, speedup: 66.16% ratio: 2.95x |
cov_route Send message Joined: 13 Sep 12 Posts: 342 Credit: 10,270,618 RAC: 0 |
Apparently this is my idea of a good time on Saturday nights now... |
betreger Send message Joined: 29 Jun 99 Posts: 11361 Credit: 29,581,041 RAC: 66 |
Apparently this is my idea of a good time on Saturday nights now... Sad, I'm going over to neighbor's house for a cocktail. |
cov_route Send message Joined: 13 Sep 12 Posts: 342 Credit: 10,270,618 RAC: 0 |
Sad, I'm going over to neighbor's house for a cocktail. I had dinner at a sports bar with my 9 year old son. A wild and crazy time was had by...nobody. |
anniet Send message Joined: 2 Feb 14 Posts: 7105 Credit: 1,577,368 RAC: 75 |
:) |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
I noticed a new version of AP for ATI on Mikes site, r2083. I benched it with Clean_01LC.wu against r1843. Run times are nearly identical at 45 sec but the CPU usage has fallen dramatically from 15 sec to 3 sec. And I think that TWIN_FFA mod does its work. Conversation about synching led to quite paradoxal result in TOTAL synching for Intel GPU build, ATi build untouched. SETI apps news We're not gonna fight them. We're gonna transcend them. |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
@Richard From your link (I started to worry if I forgot something indeed?) : Looks like AMD SDK vs Intel SDK doesn't matter for CPU consumption. But what really matters is the synching style. When each runtime call followed by clFinish CPU usage drops considerably. So, synching on blocking read not the same as synching on clFinish for intel (and i suspect for NV too) GPU. AMD GPUs don't affected. Perhaps, you are too optimistic in what that conversation (though very useful indeed) could do :D SETI apps news We're not gonna fight them. We're gonna transcend them. |
Ulrich Metzner Send message Joined: 3 Jul 02 Posts: 1256 Credit: 13,565,513 RAC: 13 |
Did i miss something? Is there anywhere an AP6 r2083 (or later) version for Nvidia available? I only got AP6 r2058 running. Aloha, Uli |
Mike Send message Joined: 17 Feb 01 Posts: 34258 Credit: 79,922,639 RAC: 80 |
Did i miss something? Not yet. We are working on new builds. After additinal testing i will make them available. With each crime and every kindness we birth our future. |
arkayn Send message Joined: 14 May 99 Posts: 4438 Credit: 55,006,323 RAC: 0 |
|
©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.