Jump to content

i7 versus Xeon


Recommended Posts

At work we need a new number crunching computer. These computers will run Linux CentOS 5, and will run all day every day at high load for the next 6-8 months. Right now, we have a bank of systems using two Xeon E5355 2.66GHz quad core processors. There are physically 2 processors in each machine, yielding 8 cores, and yes our software utilizes them all.

 

How would a single Core i7 stack up against dual quad Xeons? A single i7 965 extreme costs about the same as two of the quad Xeons, so if anyone knows where there exists a direct comparison between the two setups, I'd be extremely grateful, however price really is not an issue. I just want the absolute best performance.

 

Opinions?

Share this post


Link to post
Share on other sites

What software is it? What you are running will determine how well the i7 scales versus the dual Xeons. If you are memory-constrained and you have a lot of core-to-core communication the i7 will shine. If you're primarily number-crunching without a whole lot of memory or core-to-core access the Xeons might very well be faster.

Share this post


Link to post
Share on other sites

Its custom software, AFAIK its limited to the number crunching ability of the processor. Its pretty cool, basically its a custom made AI system. The computer we're replacing runs a genetic algorithm based program which controls 20 other identical systems, all of which run a simulation program which was also custom built specifically for our lab. But ya... its a number cruncher only.

Share this post


Link to post
Share on other sites

I think your best bet would be to find someone with an i7 system and see how it does. In terms of pure number crunching ability I think the Xeons may have an edge.

Share this post


Link to post
Share on other sites

well to compare a normal 2.6ghz i7 is faster than the top quad xeon at 3.2ghz. if i were to compare total computing power of a single 775, 771, i7 and i7 xeon (which isn't out yet) it would go like this

 

775 Core2 Q9650= 100%

771 xeon E5450= 100%

dual 771 xeon E5450= 200%

1366 i7 965 = 200%

dual ???? xeon 965= 400%

 

edit: xeon sometimes have an edge because it has the newest SSE support. an example would be back when Q6600 was new they only had SSE 3.1 support while xeon already had 4.1/2 (which cuts video rendering times in half using after-effects, etc.) now they seem to all be equal but i bet you anything xeons will always have cutting edge tech at least 6months to a year before desktop levels.

 

 

to break it down more unlike desktop level xeons can handle triple amount of memory (if you have the cash). it really helps when rendering out large projects, i know my 4k videos won't even move on 8gbs, i have to go to the lab and use this massive 64gb rig.

 

Max amount of ram

775= 8gb (90% of the time)

771= 24gb

1336= 12gb

i7 xeon=24/48?

Edited by hornybluecow

Share this post


Link to post
Share on other sites

in this case i don't think so, 771 and 775 are the same really, just different SSE support and memory. i say that lightly but the computing power of a core2 and single 771 xeon should be the same given closely related hardware.

Share this post


Link to post
Share on other sites

in this case i don't think so, 771 and 775 are the same really, just different SSE support and memory. i say that lightly but the computing power of a core2 and single 771 xeon should be the same given closely related hardware.

I didn't say the 771/775 were that different...I said you can't make a blanket statement like you did about the i7 versus the Xeon/Core 2. The winner totally depends upon what you are doing.

Share this post


Link to post
Share on other sites

oh man, i need to sell everything i got a build a xeon workstation now...MHmmm 16-cores 32 threads...*drools* maybe 4k images will render now...

 

 

edit:oh wait i guess they are still on 4cores untill the summer...

Edited by hornybluecow

Share this post


Link to post
Share on other sites

well if you are using custom parallel processing why not run th threads on a multi gpu system nvidia tesla for example couple of teraflops power per tesla system or custom ati based system on the same concept ati 4870x2 2.4tera flops a piece with single digit precision floating integers, one issue is you are limited to, 2 cards per board currently cause of hardware constraints from ati, nvidia does not have this problem but their number crunching power is less per card about 800gflops feasibly, but you can run more than two cards, and there are better docs on them. I say this because a Xeon I7 965 produces ~150gflops per chip btw new xeons so so much faster than old xeons with old old xeon 3.2ghz 771 112gflops per chip, great job intel :closedeyes:

Edited by spectrascope

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...