Dual Xeon Workstation - Page 2

delta_t · 2003-12-16, 03:32

I just built a dual Xenon using the Asus PC-DL Deluxe board and it runs quite well. I have two 2.4G processors with 533MHz bus, and 1G of DDR333 memory set up as dual channel. Nice and stable and I run two LL instances. See my benchmarks here:http://www.mersenneforum.org/showthr...5&pagenumber=4 .
I should note that this benchmark is with the second processor running a 33M LL test, so the benchmark is quite nice. Actually the benchmark is about the same if i only run one instance of Prime95.
I went dual Xenon's because as I priced out a dual Athlon 2800+ MP system it would have cost me about the same. Since I already had a dual MP system running my server I thought I'd like to try a dual Xenon.

lpmurray · 2003-12-16, 13:05

Quote:

Originally posted by delta_t
I have two 2.4G processors with 533MHz bus, and 1G of DDR333 memory set up as dual channel.

Actually the benchmark is about the same if i only run one instance of Prime95.

Hi Delta_t I was wondering if the the timings of running 1 LL verse 2 LL's is consistant. I'm wondering if its your duel channel memory thats allowing you to have the same timings where as mine vary 10-15% running 1 verses 2 LL's. My board doesn't have the duel channel... I think it might be time to rip out the dell boards. Also you said you did a small oc does the board you used allow you to change the fsb to increase the speed. it looks like if I change the boards so the i can bump the speed only about 5% and with the duel channel memory i might be able to squeeze a 20% increase out of my boxes. based on your timing of .070 for 1792fft over my .086

RickC · 2003-12-16, 17:21

Thanks for all the replies.

Dell has $400 off a Precision 450 and $500 off a Precision 650.

I really like the idea of having two processors in one box rather than an extra machine to deal with. I also like having ECC memory.

The thing that's bugging me is the 533Mhz bus. If I buy one the 800Mhz might come out. I guess there's no way to know when.

I'm buying a machine to be my main personal machine but I like to keep Prime95 in consideration when buying. I've never been an overclocker so I don't care if the dell can do that.

Rick

delta_t · 2003-12-16, 19:11

Hello lpmurray,

Yes actually, the timings are consistent. Using the Affinity setting I set one to CPU0 the other to CPU1. I'll run one LL test on one CPU and the benchmark on the other and the values are the same regardless of which CPU is running the LL test.

It's a very interesting board because it uses the Intel 875P chipset. The ability to overclock is limited to the FSB. However I need to note that there is no PCI/AGP lock so you could only overclock so far. I think the processors could handle it, but in my case it was my PCI SCSI card that couldn't handle the higher speeds on the PCI bus. I have mine at a FSB of 140 instead of the usual 133. Not much of a OC, but since this is my main workstation I need this to be ultra-stable so I haven't tried pushing it much further. Another note, if you use serial ATA hard drives, all overclocking is automatically disabled.

Here may be an interesting benchmark. Because these are HT processors I wanted to see what slow down running something on the virtual processor would do to the main LL tests. Here's the scenario. One LL-test (33M range) on CPU1 (real). One trial factoring (10M range) on CPU2 (virtual). Then the benchmark on CPU0 (other real). Results are below.

Intel(R) Xeon(TM) CPU 2.40GHz
CPU speed: 2525.71 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: 19.738 ms.
Best time for 448K FFT length: 23.529 ms.
Best time for 512K FFT length: 26.581 ms.
Best time for 640K FFT length: 31.619 ms.
Best time for 768K FFT length: 38.587 ms.
Best time for 896K FFT length: 46.073 ms.
Best time for 1024K FFT length: 51.387 ms.
Best time for 1280K FFT length: 67.707 ms.
Best time for 1536K FFT length: 82.960 ms.
Best time for 1792K FFT length: 99.243 ms.
Best time for 2048K FFT length: 111.275 ms.

RickC · 2003-12-17, 01:35

I would also think that a dual CPU machine would cost less to run (electricity cost and time performing software updates) than two single CPU machines.

Does anyone happen to know this kind of stuff?

Thanks

2003-12-17, 01:35	#16
RickC Mar 2003 45₁₀ Posts	cost of ownership I would also think that a dual CPU machine would cost less to run (electricity cost and time performing software updates) than two single CPU machines. Does anyone happen to know this kind of stuff? Thanks

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Dual Xeon 5355	bgbeuning	Information & Answers	5	2015-11-17 17:53
2x Xeon x5450 QuadCore Workstation for factorization?	beans	Hardware	15	2012-05-21 14:09
benchmarks on dual i7-xeon	fivemack	Msieve	1	2009-12-14 12:51
Advice wanted - HP workstation with XEON processors	AntonVrba	Hardware	4	2006-06-04 11:01
Dual Xeon Help	euphrus	Software	12	2005-07-21 14:47

2003-12-16, 03:32	#12
delta_t Nov 2002 Anchorage, AK 101100101₂ Posts	I just built a dual Xenon using the Asus PC-DL Deluxe board and it runs quite well. I have two 2.4G processors with 533MHz bus, and 1G of DDR333 memory set up as dual channel. Nice and stable and I run two LL instances. See my benchmarks here:http://www.mersenneforum.org/showthr...5&pagenumber=4 . I should note that this benchmark is with the second processor running a 33M LL test, so the benchmark is quite nice. Actually the benchmark is about the same if i only run one instance of Prime95. I went dual Xenon's because as I priced out a dual Athlon 2800+ MP system it would have cost me about the same. Since I already had a dual MP system running my server I thought I'd like to try a dual Xenon.

2003-12-16, 17:21	#14
RickC Mar 2003 3²·5 Posts	Thanks for all the replies. Dell has $400 off a Precision 450 and $500 off a Precision 650. I really like the idea of having two processors in one box rather than an extra machine to deal with. I also like having ECC memory. The thing that's bugging me is the 533Mhz bus. If I buy one the 800Mhz might come out. I guess there's no way to know when. I'm buying a machine to be my main personal machine but I like to keep Prime95 in consideration when buying. I've never been an overclocker so I don't care if the dell can do that. Rick

2003-12-16, 19:11	#15
delta_t Nov 2002 Anchorage, AK 3·7·17 Posts	Hello lpmurray, Yes actually, the timings are consistent. Using the Affinity setting I set one to CPU0 the other to CPU1. I'll run one LL test on one CPU and the benchmark on the other and the values are the same regardless of which CPU is running the LL test. It's a very interesting board because it uses the Intel 875P chipset. The ability to overclock is limited to the FSB. However I need to note that there is no PCI/AGP lock so you could only overclock so far. I think the processors could handle it, but in my case it was my PCI SCSI card that couldn't handle the higher speeds on the PCI bus. I have mine at a FSB of 140 instead of the usual 133. Not much of a OC, but since this is my main workstation I need this to be ultra-stable so I haven't tried pushing it much further. Another note, if you use serial ATA hard drives, all overclocking is automatically disabled. Here may be an interesting benchmark. Because these are HT processors I wanted to see what slow down running something on the virtual processor would do to the main LL tests. Here's the scenario. One LL-test (33M range) on CPU1 (real). One trial factoring (10M range) on CPU2 (virtual). Then the benchmark on CPU0 (other real). Results are below. Intel(R) Xeon(TM) CPU 2.40GHz CPU speed: 2525.71 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: 19.738 ms. Best time for 448K FFT length: 23.529 ms. Best time for 512K FFT length: 26.581 ms. Best time for 640K FFT length: 31.619 ms. Best time for 768K FFT length: 38.587 ms. Best time for 896K FFT length: 46.073 ms. Best time for 1024K FFT length: 51.387 ms. Best time for 1280K FFT length: 67.707 ms. Best time for 1536K FFT length: 82.960 ms. Best time for 1792K FFT length: 99.243 ms. Best time for 2048K FFT length: 111.275 ms.