Phenom Phun

[sdbardwick]

Out of curiosity, I picked up an AMD Phenom 9500 (2200 MHz) and ECS A770M-A motherboard combo for USD 199 at Fry's this week. I threw in whatever other parts were in the spare parts bin*, installed 32 bit WinXP SP2, and noted a few (very few) timings.

My initial impression is that the Phenom scales better than the C2Q (as expected with the better memory architecture), but with somewhat interesting results:

                 CORE
+---------------------------------------+
|____0____|____1____|____2____|____3____|
|   .039  |   .039  |   .039  |   .039  |
|    x    |   .037  |   .037  |   .037  |
|    x    |    x    |   .036  |   .036  |
|    x    |    x    |    x    |   .038  |
+----------------------------------------+

All were testing the same exponent (20501623) using a 1280K FFT size. The timings were identical across all combinations of cores; it didn't matter which cores were idle.

If I ever get some more free time I'll test with each core running a different exponent and FFT size. Right now each core is running a different exponent (double checking) but the same 1280K FFT size, and the timings are the same as in the chart.

* For the curious or detail oriented:
RAM: OCZ Platinum series DDR2-800
Video: ATI 9200 PCI (Not PCI-E) @ 1680x1050x32
HD: Some Maxtor 10GB circa 2000 - might be 5400 r.p.m.
AMD stock HSF
All stock BIOS settings with the exception of turning CPU virtualization off.
Memory timings were 5-5-5-15 (set by SPD, although the RAM is rated for 4-4-4-12)
WinXP was run in normal (not safe) mode, AVG Free Edition was running.

[db597]

That is fantastic news for the Phenom! With the C2Q, you more or less waste 1 core as the memory bottleneck comes into play (and that's with 1066MHz DDR memory).

On the flip side, your timings per core are slower per MHz than a C2Q. When running a single core, I'd place your 2.2GHz Phenom around the speed of a 1.86GHz Core 2 (looking up http://www.mersenne.org/bench.htm for 1280K FFT).

When this and the use of all cores are factored in, your 2.2GHz Phenom is approx a 2.5GHz C2Q [2.2GHz x 1.86x4/(2.2x3) = 2.5].

[TheJudger]

Hi,

Originally Posted by db597

When this and the use of all cores are factored in, your 2.2GHz Phenom is approx a 2.5GHz C2Q [2.2GHz x 1.86x4/(2.2x3) = 2.5].

My Q6600 does 4 doublechecks @1280k FFT somewhere between .036-0.37s per iteration time. (default 2.4GHz, DDR2-800, G33 chipset)
So I'll need aprox 2250-2300 MHz for .039s times.

[ET_]

Originally Posted by TheJudger

Hi,



My Q6600 does 4 doublechecks @1280k FFT somewhere between .036-0.37s per iteration time. (default 2.4GHz, DDR2-800, G33 chipset)
So I'll need aprox 2250-2300 MHz for .039s times.

So it seems that Phenom (and AMD) is not that bad, after all...

Luigi

[jasong]

Originally Posted by ET_

So it seems that Phenom (and AMD) is not that bad, after all...

Luigi

Heh, finally the whole memory situation gets some proper info. I've had someone tell me the same thing, but I didn't feel it would be a good idea to quote them(no, I'm not telling why), so I just sat on the information.

[paulunderwood]

http://www.anandtech.com/cpuchipsets...oc.aspx?i=3260

[cypherplex]

I'm very much sattisfied with my new Phenom 9850 B3 @4x2750MHz!

My quadcore-mashine does full 4x LL (Lucas-Lehmer) primality-tests at >38M length digits:

Below is a report on the work you have queued and any expected completion
dates.
[Worker thread #1] M41335781, Lucas-Lehmer test, Fri Oct 31 11:44 2008
[Worker thread #2] M38618729, Lucas-Lehmer test, Thu Oct 9 04:01 2008
[Worker thread #3] M41336749, Lucas-Lehmer test, Fri Oct 31 17:03 2008
[Worker thread #4] M41337059, Lucas-Lehmer test, Mon Oct 27 08:53 2008

Powerconsuption is: Ubuntu-Desktop, idle => 120W (primary)
Ubuntu-Desktop, 4xLL-tests 190W (primary)
MoBo is: ASUS M3A78 PRO (vga: AMD 780G chipset), 4GiB DDR2-1067 DRAM, be-quiet 430W PSU

######### best result #######################
[Mon Aug 18 02:13:21 2008]
Compare your results to other computers at http://www.mersenne.org/bench.htm
AMD Phenom(tm) 9850 Quad-Core Processor
CPU speed: 2766.08 MHz, 4 cores, DRAM= DDR2-1094 (547MHz)
CPU features: RDTSC, CMOV, Prefetch, 3DNow!, MMX, SSE, SSE2
L1 cache size: 64 KB
L2 cache size: 512 KB
L1 cache line size: 64 bytes
L2 cache line size: 64 bytes
L1 TLBS: 48
L2 TLBS: 512
Prime95 64-bit version 25.6, RdtscTiming=1
Best time for 768K FFT length: 16.606 ms.
Best time for 896K FFT length: 19.688 ms.
Best time for 1024K FFT length: 22.330 ms.
Best time for 1280K FFT length: 27.543 ms.
Best time for 1536K FFT length: 34.139 ms.
Best time for 1792K FFT length: 40.729 ms.
Best time for 2048K FFT length: 46.399 ms.
Best time for 2560K FFT length: 60.527 ms.
Best time for 3072K FFT length: 74.651 ms.
Best time for 3584K FFT length: 88.488 ms.
Best time for 4096K FFT length: 100.376 ms.
Best time for 5120K FFT length: 133.177 ms.
Best time for 6144K FFT length: 163.670 ms.
Best time for 7168K FFT length: 197.966 ms.
Best time for 8192K FFT length: 226.643 ms.
Timing FFTs using 2 threads.
Best time for 768K FFT length: 11.513 ms.
Best time for 896K FFT length: 15.334 ms.
Best time for 1024K FFT length: 18.181 ms.
Best time for 1280K FFT length: 21.109 ms.
Best time for 1536K FFT length: 26.675 ms.
Best time for 1792K FFT length: 30.417 ms.
Best time for 2048K FFT length: 34.781 ms.
Best time for 2560K FFT length: 50.113 ms.
Best time for 3072K FFT length: 57.157 ms.
Best time for 3584K FFT length: 65.265 ms.
Best time for 4096K FFT length: 74.762 ms.
Best time for 5120K FFT length: 92.568 ms.
Best time for 6144K FFT length: 115.150 ms.
Best time for 7168K FFT length: 138.661 ms.
Best time for 8192K FFT length: 156.950 ms.
Timing FFTs using 3 threads.
Best time for 768K FFT length: 10.005 ms.
Best time for 896K FFT length: 17.326 ms.
Best time for 1024K FFT length: 17.202 ms.
Best time for 1280K FFT length: 20.579 ms.
Best time for 1536K FFT length: 25.851 ms.
Best time for 1792K FFT length: 28.921 ms.
Best time for 2048K FFT length: 32.326 ms.
Best time for 2560K FFT length: 46.072 ms.
Best time for 3072K FFT length: 56.295 ms.
Best time for 3584K FFT length: 61.947 ms.
Best time for 4096K FFT length: 67.962 ms.
Best time for 5120K FFT length: 74.275 ms.
Best time for 6144K FFT length: 89.358 ms.
Best time for 7168K FFT length: 103.625 ms.
Best time for 8192K FFT length: 115.110 ms.
Timing FFTs using 4 threads.
Best time for 768K FFT length: 9.740 ms.
Best time for 896K FFT length: 14.956 ms.
Best time for 1024K FFT length: 17.550 ms.
Best time for 1280K FFT length: 20.333 ms.
Best time for 1536K FFT length: 24.185 ms.
Best time for 1792K FFT length: 27.247 ms.
Best time for 2048K FFT length: 30.856 ms.
Best time for 2560K FFT length: 42.195 ms.
Best time for 3072K FFT length: 50.638 ms.
Best time for 3584K FFT length: 56.663 ms.
Best time for 4096K FFT length: 63.414 ms.
Best time for 5120K FFT length: 69.017 ms.
Best time for 6144K FFT length: 83.081 ms.
Best time for 7168K FFT length: 95.880 ms.
Best time for 8192K FFT length: 106.667 ms.
Best time for 58 bit trial factors: 2.888 ms.
Best time for 59 bit trial factors: 2.888 ms.
Best time for 60 bit trial factors: 3.110 ms.
Best time for 61 bit trial factors: 3.062 ms.
Best time for 62 bit trial factors: 3.629 ms.
Best time for 63 bit trial factors: 4.183 ms.
Best time for 64 bit trial factors: 5.058 ms.
Best time for 65 bit trial factors: 5.898 ms.
Best time for 66 bit trial factors: 5.865 ms.
Best time for 67 bit trial factors: 5.850 ms.

Look this:

#############################################
[Mon Aug 18 01:26:16 2008]
Compare your results to other computers at http://www.mersenne.org/bench.htm
AMD Phenom(tm) 9850 Quad-Core Processor
CPU speed: 2755.74 MHz, 4 cores, DRAM= DDR2-800(400MHz)
CPU features: RDTSC, CMOV, Prefetch, 3DNow!, MMX, SSE, SSE2
L1 cache size: 64 KB
L2 cache size: 512 KB
L1 cache line size: 64 bytes
L2 cache line size: 64 bytes
L1 TLBS: 48
L2 TLBS: 512
Prime95 64-bit version 25.6, RdtscTiming=1
Best time for 768K FFT length: 18.294 ms.
Best time for 896K FFT length: 21.511 ms.
Best time for 1024K FFT length: 24.520 ms.
Best time for 1280K FFT length: 30.074 ms.
Best time for 1536K FFT length: 37.075 ms.
Best time for 1792K FFT length: 44.271 ms.
Best time for 2048K FFT length: 50.093 ms.
Best time for 2560K FFT length: 65.227 ms.
Best time for 3072K FFT length: 80.047 ms.
Best time for 3584K FFT length: 95.818 ms.
Best time for 4096K FFT length: 107.810 ms.
Best time for 5120K FFT length: 143.230 ms.
Best time for 6144K FFT length: 175.515 ms.
Best time for 7168K FFT length: 211.858 ms.
Best time for 8192K FFT length: 242.869 ms.
Timing FFTs using 2 threads.
Best time for 768K FFT length: 13.707 ms.
Best time for 896K FFT length: 19.178 ms.
Best time for 1024K FFT length: 20.617 ms.
Best time for 1280K FFT length: 25.377 ms.
Best time for 1536K FFT length: 30.551 ms.
Best time for 1792K FFT length: 35.298 ms.
Best time for 2048K FFT length: 41.433 ms.
Best time for 2560K FFT length: 54.375 ms.
Best time for 3072K FFT length: 64.995 ms.
Best time for 3584K FFT length: 75.848 ms.
Best time for 4096K FFT length: 87.797 ms.
Best time for 5120K FFT length: 106.926 ms.
Best time for 6144K FFT length: 132.055 ms.
Best time for 7168K FFT length: 157.204 ms.
Best time for 8192K FFT length: 179.277 ms.
Timing FFTs using 3 threads.
Best time for 768K FFT length: 12.252 ms.
Best time for 896K FFT length: 20.244 ms.
Best time for 1024K FFT length: 22.192 ms.
Best time for 1280K FFT length: 26.219 ms.
Best time for 1536K FFT length: 30.332 ms.
Best time for 1792K FFT length: 34.026 ms.
Best time for 2048K FFT length: 39.443 ms.
Best time for 2560K FFT length: 55.539 ms.
Best time for 3072K FFT length: 62.783 ms.
Best time for 3584K FFT length: 70.225 ms.
Best time for 4096K FFT length: 80.793 ms.
Best time for 5120K FFT length: 91.447 ms.
Best time for 6144K FFT length: 110.527 ms.
Best time for 7168K FFT length: 129.815 ms.
Best time for 8192K FFT length: 144.761 ms.
Timing FFTs using 4 threads.
Best time for 768K FFT length: 12.156 ms.
Best time for 896K FFT length: 19.685 ms.
Best time for 1024K FFT length: 21.331 ms.
Best time for 1280K FFT length: 25.173 ms.
Best time for 1536K FFT length: 29.116 ms.
Best time for 1792K FFT length: 33.023 ms.
Best time for 2048K FFT length: 37.932 ms.
Best time for 2560K FFT length: 52.377 ms.
Best time for 3072K FFT length: 59.674 ms.
Best time for 3584K FFT length: 68.039 ms.
Best time for 4096K FFT length: 77.422 ms.
Best time for 5120K FFT length: 86.889 ms.
Best time for 6144K FFT length: 105.518 ms.
Best time for 7168K FFT length: 123.274 ms.
Best time for 8192K FFT length: 138.372 ms.
Best time for 58 bit trial factors: 2.903 ms.
Best time for 59 bit trial factors: 2.900 ms.
Best time for 60 bit trial factors: 3.127 ms.
Best time for 61 bit trial factors: 3.081 ms.
Best time for 62 bit trial factors: 3.645 ms.
Best time for 63 bit trial factors: 4.204 ms.
Best time for 64 bit trial factors: 5.084 ms.
Best time for 65 bit trial factors: 5.934 ms.
Best time for 66 bit trial factors: 5.899 ms.
Best time for 67 bit trial factors: 5.879 ms.
#############################################

Interesting?