mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Software

Reply
 
Thread Tools
Old 2003-07-03, 02:49   #12
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

11101011011012 Posts
Default

Quote:
Originally Posted by E_tron
:arrow: I'm just wondering , but does this new version use SSE :D ?
Should I also post benchmarks from a P2@266, K6@450, P1@233, and an Intel i486@100 :? ? I havn't noticed any gains on any of them :( .
The only change is in the memory prefetching. P2, K6, P1, 486 will not see any benefit - nor harm.

This version should be safe. I ran the torture test for two minutes on all FFT sizes from 40K to 4096K.

I'm surprised the Duron's tiny 64K L2 cache didn't cause more problems than it did. I guess the fact that AMD caches are additive, the Duron (64K L1 + 64K L2) acts more like a Celeron with a 128K L2 cache. I'm inclined to ignore the Duron's slowdown over 896K FFT as I imagine most Duron's are doing double-checking.

Thanks for the benchmarks. I don't own a PIII or Duron so I really needed those benchies.
Prime95 is online now   Reply With Quote
Old 2003-07-03, 02:50   #13
outlnder
 
outlnder's Avatar
 
Aug 2002

2·3·53 Posts
Default

So, George, you got that XP I sent you working??
outlnder is offline   Reply With Quote
Old 2003-07-03, 14:02   #14
QuintLeo
 
QuintLeo's Avatar
 
Oct 2002
Lost in the hills of Iowa

1C016 Posts
Default

Old (22.9)

AMD Athlon(TM) XP1800+ (Palomino)
CPU speed 1575.65 MHz (137 Mhz x 11.5)
CPU features RDTSC, CMOV, PREFETCH, MMX, SSE
L1 cache size 64 KB
L2 cache size 256 KB
L1 cache line size 64 bytes
L2 cache line size 64 bytes
L1 TLBS 32
L2 TLBS 256
Prime95 version 22.9, RdtscTiming=1
Best time for 256K FFT length 26.836 ms.
Best time for 320K FFT length 34.444 ms.
Best time for 384K FFT length 44.015 ms.
Best time for 448K FFT length 49.066 ms.
Best time for 512K FFT length 54.502 ms.
Best time for 640K FFT length 64.912 ms.
Best time for 768K FFT length 85.228 ms.
Best time for 896K FFT length 99.598 ms.
Best time for 1024K FFT length 113.575 ms.
Best time for 1280K FFT length 151.338 ms.
Best time for 1536K FFT length 181.286 ms.
Best time for 1792K FFT length 218.545 ms.

Real world testing M33xxxxxx gives around 226 ms/iteration IIRC


New (23.5)

AMD Athlon(TM) XP1800+
CPU speed 1575.07 MHz (137 Mhz x 11.5)
CPU features RDTSC, CMOV, PREFETCH, MMX, SSE
L1 cache size 64 KB
L2 cache size 256 KB
L1 cache line size 64 bytes
L2 cache line size 64 bytes
L1 TLBS 32
L2 TLBS 256
Prime95 version 23.5, RdtscTiming=1
Best time for 384K FFT length 40.596 ms.
Best time for 448K FFT length 46.434 ms.
Best time for 512K FFT length 49.916 ms.
Best time for 640K FFT length 65.815 ms.
Best time for 768K FFT length 79.155 ms.
Best time for 896K FFT length 94.170 ms.
Best time for 1024K FFT length 105.352 ms.
Best time for 1280K FFT length 140.656 ms.
Best time for 1536K FFT length 167.981 ms.
Best time for 1792K FFT length 209.371 ms.

2048k line deleted, as there's nothing to compare it to.

I dunno why the CPU clock picks up differently - I didn't change anything between the tests. It's LOWER on the new version, though, and very minor - shouldn't affect the benches significantly.

ASUS A7V266, slightly overclocked due to CPU cooling limits (even Alpha 8045/Deltas have
some problem keeping a Palomino cool when the ambient is significantly above 80 F!) - I
normally OC this machine to 144 Mhz, but have had to down-clock a little since the summer
heat hit.

I'll be curious to see what my Thoroughbreds and the P-IV manage, as those are my fastest
machines - can't check the K5 and K6 and Celerons yet, as those are all Linux boxen.
QuintLeo is offline   Reply With Quote
Old 2003-07-03, 14:14   #15
Pi Rho
 
Pi Rho's Avatar
 
Sep 2002

22×11 Posts
Default

Nice bump for DC's and is helping a wee bit with the number it's crunching on now. Hope this helps.


Intel(R) Celeron(R) processor
CPU speed: 601.57 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX, SSE
L1 cache size: 16 KB
L2 cache size: 128 KB
L1 cache line size: 32 bytes
L2 cache line size: 32 bytes
TLBS: 64
Prime95 version 23.4, RdtscTiming=1
Best time for 384K FFT length: 134.047 ms.
Best time for 448K FFT length: 160.199 ms.
Best time for 512K FFT length: 180.421 ms.
Best time for 640K FFT length: 235.474 ms.
Best time for 768K FFT length: 287.380 ms.
Best time for 896K FFT length: 338.050 ms.
Best time for 1024K FFT length: 384.591 ms.
Best time for 1280K FFT length: 501.187 ms.
Best time for 1536K FFT length: 609.477 ms.
Best time for 1792K FFT length: 729.859 ms.
Best time for 2048K FFT length: 833.709 ms.

Intel(R) Celeron(R) processor
CPU speed: 601.00 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX, SSE
L1 cache size: 16 KB
L2 cache size: 128 KB
L1 cache line size: 32 bytes
L2 cache line size: 32 bytes
TLBS: 64
Prime95 version 23.5, RdtscTiming=1
Best time for 384K FFT length: 128.460 ms.
Best time for 448K FFT length: 152.925 ms.
Best time for 512K FFT length: 171.262 ms.
Best time for 640K FFT length: 229.175 ms.
Best time for 768K FFT length: 274.399 ms.
Best time for 896K FFT length: 332.778 ms.
Best time for 1024K FFT length: 380.162 ms.
Best time for 1280K FFT length: 505.080 ms.
Best time for 1536K FFT length: 609.864 ms.
Best time for 1792K FFT length: 743.162 ms.
Best time for 2048K FFT length: 840.896 ms.
Pi Rho is offline   Reply With Quote
Old 2003-07-03, 14:26   #16
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

35·31 Posts
Default

Quote:
Originally Posted by outlnder
So, George, you got that XP I sent you working??
Yes, it is now operating at full speed. Thanks!
Prime95 is online now   Reply With Quote
Old 2003-07-03, 14:32   #17
QuintLeo
 
QuintLeo's Avatar
 
Oct 2002
Lost in the hills of Iowa

26×7 Posts
Default

Old (23.2)

AMD Athlon(TM) XP 1800+ (Thoroughbred)
CPU speed 1827.50 MHz (174 FSB x10.5x Multiplier)
DDR PC2700 CAS2.5 @ 174 Mhz
CPU features RDTSC, CMOV, PREFETCH, MMX, SSE
L1 cache size 64 KB
L2 cache size 256 KB
L1 cache line size 64 bytes
L2 cache line size 64 bytes
L1 TLBS 32
L2 TLBS 256
Prime95 version 23.2, RdtscTiming=1
Best time for 384K FFT length 37.431 ms.
Best time for 448K FFT length 40.995 ms.
Best time for 512K FFT length 45.425 ms.
Best time for 640K FFT length 57.962 ms.
Best time for 768K FFT length 71.206 ms.
Best time for 896K FFT length 83.125 ms.
Best time for 1024K FFT length 95.108 ms.
Best time for 1280K FFT length 126.465 ms.
Best time for 1536K FFT length 151.975 ms.
Best time for 1792K FFT length 182.876 ms.
Best time for 2048K FFT length 204.786 ms.

Actual long-term usage generates a mix of 0.184 and 0.185 timings with these settings testing an exponent in the 33xxxxxx range

New (23.5)

AMD Athlon(TM) XP 1800+
CPU speed 1827.92 MHz
CPU features RDTSC, CMOV, PREFETCH, MMX, SSE
L1 cache size 64 KB
L2 cache size 256 KB
L1 cache line size 64 bytes
L2 cache line size 64 bytes
L1 TLBS 32
L2 TLBS 256
Prime95 version 23.5, RdtscTiming=1
Best time for 384K FFT length 34.337 ms.
Best time for 448K FFT length 38.498 ms.
Best time for 512K FFT length 41.535 ms.
Best time for 640K FFT length 55.057 ms.
Best time for 768K FFT length 65.726 ms.
Best time for 896K FFT length 78.710 ms.
Best time for 1024K FFT length 87.958 ms.
Best time for 1280K FFT length 116.590 ms.
Best time for 1536K FFT length 140.003 ms.
Best time for 1792K FFT length 172.403 ms.
Best time for 2048K FFT length 198.344 ms.


All of these tests done on a ASUS A7V333-X motherboard (VIA KT333 chipset) with BIOS v1.002



Turns out I can't check the P-IV yet - I'd forgotten that was a LINUX boxen. (*doh*)
QuintLeo is offline   Reply With Quote
Old 2003-07-03, 14:52   #18
Pi Rho
 
Pi Rho's Avatar
 
Sep 2002

22×11 Posts
Default

One more benchmark, with a PIII this time. Right around 3-4% improvement for all FFT sizes.

Intel(R) Pentium(R) III processor
CPU speed: 730.89 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX, SSE
L1 cache size: 16 KB
L2 cache size: 256 KB
L1 cache line size: 32 bytes
L2 cache line size: 32 bytes
TLBS: 64
Prime95 version 23.4, RdtscTiming=1
Best time for 384K FFT length: 103.950 ms.
Best time for 448K FFT length: 122.894 ms.
Best time for 512K FFT length: 138.382 ms.
Best time for 640K FFT length: 180.617 ms.
Best time for 768K FFT length: 218.330 ms.
Best time for 896K FFT length: 259.535 ms.
Best time for 1024K FFT length: 292.879 ms.
Best time for 1280K FFT length: 379.096 ms.
Best time for 1536K FFT length: 458.864 ms.
Best time for 1792K FFT length: 549.086 ms.
Best time for 2048K FFT length: 622.693 ms.

Intel(R) Pentium(R) III processor
CPU speed: 730.90 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX, SSE
L1 cache size: 16 KB
L2 cache size: 256 KB
L1 cache line size: 32 bytes
L2 cache line size: 32 bytes
TLBS: 64
Prime95 version 23.5, RdtscTiming=1
Best time for 384K FFT length: 99.956 ms.
Best time for 448K FFT length: 117.510 ms.
Best time for 512K FFT length: 131.901 ms.
Best time for 640K FFT length: 175.793 ms.
Best time for 768K FFT length: 207.839 ms.
Best time for 896K FFT length: 248.791 ms.
Best time for 1024K FFT length: 279.385 ms.
Best time for 1280K FFT length: 372.035 ms.
Best time for 1536K FFT length: 442.121 ms.
Best time for 1792K FFT length: 548.278 ms.
Best time for 2048K FFT length: 603.855 ms.
Pi Rho is offline   Reply With Quote
Old 2003-07-03, 17:31   #19
mephisto
 
mephisto's Avatar
 
Feb 2003
Norway

3816 Posts
Default

Duron 800, 100MHz SDRAM, Win ME: Slower above 1024 FFT, quicker below.

* before *
AMD Duron(tm) Processor
CPU speed: 801.39 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX
L1 cache size: 64 KB
L2 cache size: 64 KB
L1 cache line size: 64 bytes
L2 cache line size: 64 bytes
L1 TLBS: 24
L2 TLBS: 256
Prime95 version 22.12, RdtscTiming=1
Best time for 384K FFT length: 90.692 ms.
Best time for 448K FFT length: 100.453 ms.
Best time for 512K FFT length: 110.996 ms.
Best time for 640K FFT length: 146.442 ms.
Best time for 768K FFT length: 181.653 ms.
Best time for 896K FFT length: 217.375 ms.
Best time for 1024K FFT length: 253.832 ms.
Best time for 1280K FFT length: 346.008 ms.
Best time for 1536K FFT length: 426.853 ms.
Best time for 1792K FFT length: 521.110 ms.

* after *
AMD Duron(tm) Processor
CPU speed: 801.27 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX
L1 cache size: 64 KB
L2 cache size: 64 KB
L1 cache line size: 64 bytes
L2 cache line size: 64 bytes
L1 TLBS: 24
L2 TLBS: 256
Prime95 version 23.5, RdtscTiming=1
Best time for 384K FFT length: 85.092 ms.
Best time for 448K FFT length: 97.163 ms.
Best time for 512K FFT length: 104.962 ms.
Best time for 640K FFT length: 141.596 ms.
Best time for 768K FFT length: 175.018 ms.
Best time for 896K FFT length: 216.444 ms.
Best time for 1024K FFT length: 250.944 ms.
Best time for 1280K FFT length: 346.110 ms.
Best time for 1536K FFT length: 442.589 ms.
Best time for 1792K FFT length: 526.401 ms.
Best time for 2048K FFT length: 606.990 ms.
mephisto is offline   Reply With Quote
Old 2003-07-04, 00:21   #20
PageFault
 
PageFault's Avatar
 
Aug 2002
Dawn of the Dead

5·47 Posts
Default

ok, I'll be an early adopter. Just to play safe, I'll run a few hours of torture. It hasn't caused any funny stuff, i.e., cpu detected as normal.

btw with respect to durons, I do run first time tests, even the big 19M ones. That is, until the orphanage found a bonanza in the 12 - 13M range.

Quote:
Originally Posted by Prime95
This version should be safe. I ran the torture test for two minutes on all FFT sizes from 40K to 4096K.
PageFault is offline   Reply With Quote
Old 2003-07-04, 07:05   #21
Xyzzy
 
Xyzzy's Avatar
 
"Mike"
Aug 2002

2·23·179 Posts
Default

IBM T23... linux-2.4.20-gentoo-r5...

Mobile Intel(R) Pentium(R) III CPU - M 1200MHz
CPU speed: 1198.79 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX, SSE
L1 cache size: 16 KB
L2 cache size: 512 KB
L1 cache line size: 32 bytes
L2 cache line size: 32 bytes
TLBS: 64
Prime95 version 23.4, RdtscTiming=1
Best time for 384K FFT length: 78.827 ms.
Best time for 448K FFT length: 93.056 ms.
Best time for 512K FFT length: 106.546 ms.
Best time for 640K FFT length: 139.678 ms.
Best time for 768K FFT length: 175.679 ms.
Best time for 896K FFT length: 198.735 ms.
Best time for 1024K FFT length: 235.868 ms.
Best time for 1280K FFT length: 299.419 ms.
Best time for 1536K FFT length: 360.386 ms.
Best time for 1792K FFT length: 428.657 ms.
Best time for 2048K FFT length: 481.800 ms.

Mobile Intel(R) Pentium(R) III CPU - M 1200MHz
CPU speed: 1199.01 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX, SSE
L1 cache size: 16 KB
L2 cache size: 512 KB
L1 cache line size: 32 bytes
L2 cache line size: 32 bytes
TLBS: 64
Prime95 version 23.5, RdtscTiming=1
Best time for 384K FFT length: 72.605 ms.
Best time for 448K FFT length: 88.911 ms.
Best time for 512K FFT length: 97.770 ms.
Best time for 640K FFT length: 124.243 ms.
Best time for 768K FFT length: 154.981 ms.
Best time for 896K FFT length: 178.834 ms.
Best time for 1024K FFT length: 214.707 ms.
Best time for 1280K FFT length: 292.923 ms.
Best time for 1536K FFT length: 337.967 ms.
Best time for 1792K FFT length: 424.030 ms.
Best time for 2048K FFT length: 459.100 ms.

384K FFT length: 108.57%
448K FFT length: 104.66%
512K FFT length: 108.98%
640K FFT length: 112.42%
768K FFT length: 113.36%
896K FFT length: 111.13%
1024K FFT length: 109.86%
1280K FFT length: 102.44%
1536K FFT length: 106.63%
1792K FFT length: 101.09%
2048K FFT length: 104.94%
Xyzzy is offline   Reply With Quote
Old 2003-07-04, 15:46   #22
gbvalor
 
gbvalor's Avatar
 
Aug 2002

3×37 Posts
Default

Hi,

Nice work George. I get about 10% improvement.

Athlon XP "Barton", 512KB L2 cache, FSB 333, 512MB DDR-333

OLD 2.34 RESULTS:

AMD Athlon(tm) XP 2500+
CPU speed: 1830.05 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX, SSE
L1 cache size: 64 KB
L2 cache size: 512 KB
L1 cache line size: 64 bytes
L2 cache line size: 64 bytes
L1 TLBS: 32
L2 TLBS: 256
Prime95 version 23.4, RdtscTiming=1
Best time for 384K FFT length: 36.546 ms.
Best time for 448K FFT length: 41.683 ms.
Best time for 512K FFT length: 46.324 ms.
Best time for 640K FFT length: 59.474 ms.
Best time for 768K FFT length: 72.669 ms.
Best time for 896K FFT length: 85.455 ms.
Best time for 1024K FFT length: 98.204 ms.
Best time for 1280K FFT length: 130.217 ms.
Best time for 1536K FFT length: 158.145 ms.
Best time for 1792K FFT length: 191.176 ms.
Best time for 2048K FFT length: 212.930 ms.

NEW 23.5:

AMD Athlon(tm) XP 2500+
CPU speed: 1829.77 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX, SSE
L1 cache size: 64 KB
L2 cache size: 512 KB
L1 cache line size: 64 bytes
L2 cache line size: 64 bytes
L1 TLBS: 32
L2 TLBS: 256
Prime95 version 23.5, RdtscTiming=1
Best time for 384K FFT length: 33.311 ms.
Best time for 448K FFT length: 38.799 ms.
Best time for 512K FFT length: 42.042 ms.
Best time for 640K FFT length: 55.087 ms.
Best time for 768K FFT length: 66.773 ms.
Best time for 896K FFT length: 79.887 ms.
Best time for 1024K FFT length: 89.449 ms.
Best time for 1280K FFT length: 119.380 ms.
Best time for 1536K FFT length: 140.621 ms.
Best time for 1792K FFT length: 174.586 ms.
Best time for 2048K FFT length: 191.278 ms.

Regards.

Guillermo.
gbvalor is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Benchmarks davieddy Information & Answers 21 2013-04-28 11:16
AMD benchmarks NO LONGER needed for v26 Prime95 Software 11 2012-01-13 15:06
Celeron - special benchmarks needed Prime95 Software 9 2011-04-07 02:19
V24.12 special benchmarks needed Prime95 Software 29 2005-07-04 09:59
Celeron D Benchmarks needed E_tron Hardware 4 2004-08-10 11:28

All times are UTC. The time now is 16:39.


Sun Aug 1 16:39:24 UTC 2021 up 9 days, 11:08, 0 users, load averages: 1.25, 1.33, 1.49

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.