mersenneforum.org  

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

Reply
 
Thread Tools
Old 2003-12-07, 23:53   #78
PageFault
 
PageFault's Avatar
 
Aug 2002
Dawn of the Dead

5×47 Posts
Default

Similar machines, 2.4C's on P4P800, 250 fsb on both, ddr frequency 200. First box has 2 x 256 MB Geil gold dragoon and second box has 256 MB Kingston Hyperx. Use of dual channel memory has no significant effect.

Code:
Intel(R) Pentium(R) 4 CPU 2.40GHz
CPU speed: 3000.29 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX, SSE, SSE2
L1 cache size: 8 KB
L2 cache size: 512 KB
L1 cache line size: 64 bytes
L2 cache line size: 128 bytes
TLBS: 64
Prime95 version 23.7, RdtscTiming=1
Best time for 384K FFT length: 12.053 ms.
Best time for 448K FFT length: 14.326 ms.
Best time for 512K FFT length: 16.325 ms.
Best time for 640K FFT length: 19.481 ms.
Best time for 768K FFT length: 23.779 ms.
Best time for 896K FFT length: 28.097 ms.
Best time for 1024K FFT length: 31.503 ms.
Best time for 1280K FFT length: 41.395 ms.
Best time for 1536K FFT length: 50.760 ms.
Best time for 1792K FFT length: 60.324 ms.
Best time for 2048K FFT length: 68.320 ms.
Code:
Intel(R) Pentium(R) 4 CPU 2.40GHz
CPU speed: 3000.16 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX, SSE, SSE2
L1 cache size: 8 KB
L2 cache size: 512 KB
L1 cache line size: 64 bytes
L2 cache line size: 128 bytes
TLBS: 64
Prime95 version 23.7, RdtscTiming=1
Best time for 384K FFT length: 12.198 ms.
Best time for 448K FFT length: 14.529 ms.
Best time for 512K FFT length: 16.594 ms.
Best time for 640K FFT length: 19.801 ms.
Best time for 768K FFT length: 24.143 ms.
Best time for 896K FFT length: 28.453 ms.
Best time for 1024K FFT length: 32.044 ms.
Best time for 1280K FFT length: 42.284 ms.
Best time for 1536K FFT length: 51.669 ms.
Best time for 1792K FFT length: 61.328 ms.
Best time for 2048K FFT length: 69.770 ms.
PageFault is offline   Reply With Quote
Old 2003-12-08, 03:07   #79
E_tron
 
E_tron's Avatar
 
Sep 2002
Austin, TX

56110 Posts
Default

Here are the Results from a VIA C3 "E" series processor.

VIA Samuel 2
CPU speed: 799.96 MHz
CPU features: RDTSC, MMX
L1 cache size: 64 KB
L2 cache size: 64 KB
L1 cache line size: 32 bytes
L2 cache line size: 32 bytes
TLBS: 128
Prime95 version 23.7, RdtscTiming=1
Best time for 384K FFT length: 369.462 ms.
Best time for 448K FFT length: 446.807 ms.
Best time for 512K FFT length: 490.236 ms.
Best time for 640K FFT length: 648.078 ms.
Best time for 768K FFT length: 770.683 ms.
Best time for 896K FFT length: 930.605 ms.
Best time for 1024K FFT length: 1030.509 ms.
Best time for 1280K FFT length: 1360.306 ms.
Best time for 1536K FFT length: 1668.055 ms.
Best time for 1792K FFT length: 2050.080 ms.
Best time for 2048K FFT length: 2305.938 ms.

Sure it is... different, but hey it doesn't need a heat sink. VIA has a different view on processors.

Prime95 is the only program yet i benched with that totally shut out the processor. Prime95's optimizations obviously don't go well with this processor . With every thing else i tested with so far, it actually performs like a 600mhz celeron.
E_tron is offline   Reply With Quote
Old 2003-12-08, 07:35   #80
Dresdenboy
 
Dresdenboy's Avatar
 
Apr 2003
Berlin, Germany

192 Posts
Default

Anandtech should have included VIA CPUs in their value CPU roundup. And BTW Intel should think over their Celeron pricing.

Does someone have a Via Nehemia available for testing? I think Prime95 will run much better on it than on Samuel and other predecessors. The FPU has been overhauled and offered about twice the speed (AFAIR).
Dresdenboy is offline   Reply With Quote
Old 2003-12-08, 09:21   #81
xtreme2k
 
xtreme2k's Avatar
 
Aug 2002

2·3·29 Posts
Default

Current generation of Celeron has been crippled way too much.

Cache too small, FSB too slow...
xtreme2k is offline   Reply With Quote
Old 2003-12-08, 11:34   #82
QuintLeo
 
QuintLeo's Avatar
 
Oct 2002
Lost in the hills of Iowa

26×7 Posts
Default

For comparison, my K6-400 iterates at 1792K at just over 3000 ms. Or a hair better per Mhz than the C3 800.

The K6 was well known for POOR FP performance in it's day - it ran the FPU at half CPU clock speed being the main reason why.

The Nemiah will have to do a lot more than 2x the performance to even match the P-III based Celerons on FP per Mhz (my BP6 dual-Celeron 600s run 1000-1250 ms, depending on what the other CPU is doing), much less anything current.
QuintLeo is offline   Reply With Quote
Old 2003-12-08, 11:54   #83
Dresdenboy
 
Dresdenboy's Avatar
 
Apr 2003
Berlin, Germany

16916 Posts
Default

Quote:
Originally posted by QuintLeo
The K6 was well known for POOR FP performance in it's day - it ran the FPU at half CPU clock speed being the main reason why.

The Nemiah will have to do a lot more than 2x the performance to even match the P-III based Celerons on FP per Mhz (my BP6 dual-Celeron 600s run 1000-1250 ms, depending on what the other CPU is doing), much less anything current.
AFAIK the K6 didn't have a FPU at half CPU clock speed, but it was not pipelined. The latency was smaller than on a P5/P6 but without pipelining throughput=latency, thus higher than on a pipelined FPU. BTW the 3DNow! units were pipelined with a throughput of up to 2 FP ops/cycle. The K6 Optimization manual didn't give numbers ( I have one at home which has them) but http://www.sysopt.com/k6-fpu.html says: On the AMD K6, it takes the FPU 2 cycles to complete the important instructions - FADD, FSUB, FMUL - the Pentiums, despite the higher latency of 3 - can effectively do them in one cycle if the code is well optimized.
And I think the often used FXCH was not free on K6.

The Cyrix and Centaur CPUs and most successors had a half speed partially pipelined FPU - which has been turned into a full speed fully pipelined FPU with Nehemiah.
Dresdenboy is offline   Reply With Quote
Old 2003-12-08, 16:23   #84
Quacky
 
Sep 2002

2×5 Posts
Default

Ive noticed on a fresh boot of winXp prime95 will be slighly slower...after loading up a game for a few seconds and then exiting it seems to clear out whatever was holding it up. It then drops back down to normal timings. Shrug
Quacky is offline   Reply With Quote
Old 2003-12-08, 18:40   #85
Xyzzy
 
Xyzzy's Avatar
 
"Mike"
Aug 2002

816810 Posts
Default

http://www.sandpile.org/impl/k6.htm
Xyzzy is offline   Reply With Quote
Old 2003-12-09, 18:46   #86
E_tron
 
E_tron's Avatar
 
Sep 2002
Austin, TX

3×11×17 Posts
Default

Quote:
Originally posted by E_tron
Here are the Results from a VIA C3 "E" series processor.

VIA Samuel 2
CPU speed: 799.96 MHz
CPU features: RDTSC, MMX
L1 cache size: 64 KB
L2 cache size: 64 KB
L1 cache line size: 32 bytes
L2 cache line size: 32 bytes
TLBS: 128
Prime95 version 23.7, RdtscTiming=1
Best time for 384K FFT length: 369.462 ms.
Best time for 448K FFT length: 446.807 ms.
Best time for 512K FFT length: 490.236 ms.
Best time for 640K FFT length: 648.078 ms.
Best time for 768K FFT length: 770.683 ms.
Best time for 896K FFT length: 930.605 ms.
Best time for 1024K FFT length: 1030.509 ms.
Best time for 1280K FFT length: 1360.306 ms.
Best time for 1536K FFT length: 1668.055 ms.
Best time for 1792K FFT length: 2050.080 ms.
Best time for 2048K FFT length: 2305.938 ms.

Sure it is... different, but hey it doesn't need a heat sink. VIA has a different view on processors.

Prime95 is the only program yet i benched with that totally shut out the processor. Prime95's optimizations obviously don't go well with this processor . With every thing else i tested with so far, it actually performs like a 600mhz celeron.
Ok, this machine is going to factor .
E_tron is offline   Reply With Quote
Old 2003-12-09, 20:56   #87
GP2
 
GP2's Avatar
 
Sep 2003

2×5×7×37 Posts
Default

Quote:
Originally posted by Dresdenboy
Does someone have a Via Nehemia available for testing? I think Prime95 will run much better on it than on Samuel and other predecessors. The FPU has been overhauled and offered about twice the speed (AFAIR).
There's a review of the Nehemiah CPU and motherboard at http://www.mini-itx.com/. It says it has a full-speed FPU.

I've occasionally looked at the stuff at mini-itx.com and wondered if those things could be used as farm building blocks. But the floating point performance was always unsuitable. Maybe the Nehemiah is different.

The new Nano-ITX standard will apparently let you use a power daughterboard that can take 12V DC input in addition to the standard ATX power connector. No more power supply...
GP2 is offline   Reply With Quote
Old 2003-12-10, 23:35   #88
geoff
 
geoff's Avatar
 
Mar 2003
New Zealand

13·89 Posts
Default

This system has a Gigabyte GA-8VD667K mainboard (default BIOS settings but with onboard graphics disabled), 256Mb Kingmax DDR333 CL2.5 memory (at 266MHz), Linux kernel 2.4.22.

The benchmark times fluctuate quite a bit, the following are the best results of 10 runs. The 640K result is accurate (I am getting consistent 0.041 sec/iteration times on a doublecheck in the 10.6M range), but I haven't checked the others yet.

Intel(R) Celeron(R) CPU 2.40GHz
CPU speed: 2423.27 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX, SSE, SSE2
L1 cache size: 8 KB
L2 cache size: 128 KB
L1 cache line size: 64 bytes
L2 cache line size: 128 bytes
TLBS: 64
Prime95 version 23.5, RdtscTiming=1
Best time for 384K FFT length: 22.674 ms.
Best time for 448K FFT length: 25.871 ms.
Best time for 512K FFT length: 39.819 ms.
Best time for 640K FFT length: 40.419 ms.
Best time for 768K FFT length: 50.382 ms.
Best time for 896K FFT length: 58.563 ms.
Best time for 1024K FFT length: 64.733 ms.
Best time for 1280K FFT length: 96.191 ms.
Best time for 1536K FFT length: 119.185 ms.
Best time for 1792K FFT length: 155.347 ms.
Best time for 2048K FFT length: 187.486 ms.
geoff is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Perpetual "interesting video" thread... Xyzzy Lounge 39 2021-03-12 14:19
LLR benchmark thread Oddball Riesel Prime Search 5 2010-08-02 00:11
Perpetual I'm pi**ed off thread rogue Soap Box 19 2009-10-28 19:17
Perpetual autostereogram thread... Xyzzy Lounge 10 2006-09-28 00:36
Perpetual ECM factoring challenge thread... Xyzzy Factoring 65 2005-09-05 08:16

All times are UTC. The time now is 01:31.

Tue May 18 01:31:41 UTC 2021 up 39 days, 20:12, 0 users, load averages: 1.85, 2.07, 2.03

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.