AMD Brisbane 6000+ slower then Brisbane 5000+ ?

Forum > Windows

(1/3) > >>

Fomes:
Hello everyone!

I am running boinc + SETI on following system:
- Windows XP Professional 64 (no change)
- cpu: now AMD Brisbane 6000+, before AMD Brisbane 5000+
- memory: 4 GB (no change)

the question:
With the "old" AMD Brisbane 5000+ I reached about 1000 credits. Why do I reach only about 700 credits with the new 6000+ CPU?? I really don't understand that. Tried everything that came in my mind - checked latest CPU/motherboard drivers, changed optimized Seti@home App from Windows 64 bit AK v8.0 SSE3 (AMD/INTEL) to 2.4V Windows x64 SSE2, skipped from BOINC 6.4.5 back to BOINC 6.2.19 - nothing :-\

Help!

Gecko_R7:

--- Quote from: Fomes on 19 Feb 2009, 07:14:39 pm ---Hello everyone!

I am running boinc + SETI on following system:
- Windows XP Professional 64 (no change)
- cpu: now AMD Brisbane 6000+, before AMD Brisbane 5000+
- memory: 4 GB (no change)

the question:
With the "old" AMD Brisbane 5000+ I reached about 1000 credits. Why do I reach only about 700 credits with the new 6000+ CPU?? I really don't understand that. Tried everything that came in my mind - checked latest CPU/motherboard drivers, changed optimized Seti@home App from Windows 64 bit AK v8.0 SSE3 (AMD/INTEL) to 2.4V Windows x64 SSE2, skipped from BOINC 6.4.5 back to BOINC 6.2.19 - nothing :-\

Help!

--- End quote ---

Is the 700 credit result running 2.4V or AKv8? 2.4V of course is much slower than AKv8.

Fomes:

--- Quote from: Gecko_R7 on 19 Feb 2009, 07:18:41 pm ---Is the 700 credit result running 2.4V or AKv8? 2.4V of course is much slower than AKv8.
--- End quote ---

Tried both app - about 700 credits both - no remarkable differences.

btw - i remember a thread in here, where the 2.4V SSE2 is recommended for AMD cpu because of the bad implementation of SSE3 on those cpu. Am I wrong?

Raistmer:

--- Quote from: Fomes on 19 Feb 2009, 07:43:36 pm ---
--- Quote from: Gecko_R7 on 19 Feb 2009, 07:18:41 pm ---Is the 700 credit result running 2.4V or AKv8? 2.4V of course is much slower than AKv8.
--- End quote ---

Tried both app - about 700 credits both - no remarkable differences.

btw - i remember a thread in here, where the 2.4V SSE2 is recommended for AMD cpu because of the bad implementation of SSE3 on those cpu. Am I wrong?

--- End quote ---
Probably you are wrong.
Better to use x86 SSE2 AK_v8 on x64 windows for AMD.
Oops, bad memory... You are right
KWSN_2.4V_SSE2_MB.exe - on my own Venice as production app for x64 WinServer 2003. (was few years ago...)
(will do test if v8b AMD-specific build faster or not and post here. AMD CPUs are strange beasts... ;) )

ADDON: Hm.... but on my another AMD x64 host, Winchester core, x86 AK_v8 SSE2 used as production app... maybe just forgot to renew 2.4V ?... will bench again...

ADDON: 2.4V was in x86 BOINC directory for WinXP not used few years already on that host... x64 BOINC (active one) using AK_v8_win_x64_SSE3_AMD_IPP_m7_PGO_XEON_r33.exe build (see bench results) but now probably will be upgraded to another build.

Raistmer:
Ok, first part of fresh data arrived:

Core - Venice.
Quick timetable

WU : PG0009.wu
KWSN_2.4V_SSE2_MB.exe : 1271.969 secs CPU
AK_v8_win_x64_SSE3_AMD_IPP_m7_PGO_XEON_r33.exe : 1052.422 secs CPU
Speedup : 17.26%
Ratio : 1.21 x
AK_v8b_win_SSE3_AMD.exe : 1085.469 secs CPU
Speedup : 14.66%
Ratio : 1.17 x
AK_v8b_win_x64_SSE3.exe : 1027.438 secs CPU
Speedup : 19.22%
Ratio : 1.24 x
KWSN_2.4V_SSE2_MB.exe : 1256.781 secs CPU
Speedup : 1.19%
Ratio : 1.01 x

WU : PG0395.wu
KWSN_2.4V_SSE2_MB.exe : 1441.188 secs CPU
AK_v8_win_x64_SSE3_AMD_IPP_m7_PGO_XEON_r33.exe : 986.469 secs CPU
Speedup : 31.55%
Ratio : 1.46 x
AK_v8b_win_SSE3_AMD.exe : 998.797 secs CPU
Speedup : 30.70%
Ratio : 1.44 x
AK_v8b_win_x64_SSE3.exe : 993.453 secs CPU
Speedup : 31.07%
Ratio : 1.45 x
KWSN_2.4V_SSE2_MB.exe : 1441.078 secs CPU
Speedup : 0.01%
Ratio : 1.00 x

WU : PG0444.wu
KWSN_2.4V_SSE2_MB.exe : 1251.031 secs CPU
AK_v8_win_x64_SSE3_AMD_IPP_m7_PGO_XEON_r33.exe : 860.500 secs CPU
Speedup : 31.22%
Ratio : 1.45 x
AK_v8b_win_SSE3_AMD.exe : 859.563 secs CPU
Speedup : 31.29%
Ratio : 1.46 x
AK_v8b_win_x64_SSE3.exe : 870.453 secs CPU
Speedup : 30.42%
Ratio : 1.44 x
KWSN_2.4V_SSE2_MB.exe : 1251.641 secs CPU
Speedup : -0.05%
Ratio : 1.00 x

WU : PG1327.wu
KWSN_2.4V_SSE2_MB.exe : 1069.797 secs CPU
AK_v8_win_x64_SSE3_AMD_IPP_m7_PGO_XEON_r33.exe : 959.594 secs CPU
Speedup : 10.30%
Ratio : 1.11 x
AK_v8b_win_SSE3_AMD.exe : 883.734 secs CPU
Speedup : 17.39%
Ratio : 1.21 x
AK_v8b_win_x64_SSE3.exe : 979.391 secs CPU
Speedup : 8.45%
Ratio : 1.09 x
KWSN_2.4V_SSE2_MB.exe : 1071.000 secs CPU
Speedup : -0.11%
Ratio : 1.00 x

That is, for AMD with SSE3 support only 2 builds worth to try on x64 system (form tested) SSE3_AMD and x64 SSE3.
2.4V not an option.
Next run will compare SSE2 and SSE3 AK_v8 on this host.

(and I understood from where KWSN SSE2 2.4V came on my host, see edit to earlier post)

Navigation

[0] Message Index

[#] Next page

Go to full version