Got my 2500K up and running

[Smartazz]

Hello all, I just got my new 2500K build up and running BOINC, specifically World Community Grid. I replaced my 3GHz Q6600, any ideas what kind of increase in performance I should get?

 

[somethingsketchy]

Besides the new architecture (two generations), a lower TDP and a higher base clock speed (3.3GHz stock), you should have quite the jump in performance.

While I'm not certain if it will be 2x performance (I don't own a Sandy Bridge processor to verify), it should be quite a jump.

Taking Rosetta@home for example (a CPU only project), a 2500K (1.) is #41 in top computers (~4600 credits a day), while a Q6600 (2.) is #367 in top computers (~2600 credits a day). Granted there is overclocking involved (my Q6600 on stock never left ~1200 credits a day), a stock 2500K should be a nice little jump.



1. http://boinc.bakerlab.org/top_hosts.php?sort_by=expavg_credit&offset=40 - first 2500K on the list

2. http://boinc.bakerlab.org/top_hosts.php?sort_by=expavg_credit&offset=360 - first Q6600 on the list

 

[Smartazz]

Thanks! This was what I was looking for.

 

[blckgrffn]

In the real world, how is the performance increase scoring for you?

 

[Smartazz]

I unfortunately haven't been running it 24/7 yet, waiting on my H80 and a modest overclock. I'll update once I got it installed.

 

[mk]

This may or may not be helpful, but in Correlizer my 2600K@4.3GHz has a RAC of ~6000 while my Q6600@3.0GHz does ~2600 (based on runtimes and awarded credit/WU).

 

[somethingsketchy]

That's actually very close to what I had mentioned five days ago with Rosetta@home. In this case a 2500K/2600K has double the output of a Q6600, not bad for two generations of innovations.

I'm wondering if Ivy Bridge will push that to 3x performance of a Q6600?

 

[Smartazz]

All at lower power consumption to boot. By the way, in BOINC if you click "Run CPU Benchmarks", what results do you get? I get 3687 floating point MIPS per CPU which seems really low considering my Q6600 got about 2900. Does this score matter at all?

 

[mk]

All at lower power consumption to boot. By the way, in BOINC if you click "Run CPU Benchmarks", what results do you get? I get 3687 floating point MIPS per CPU which seems really low considering my Q6600 got about 2900. Does this score matter at all?

I'm getting a bit over 4k FP MIPS. AFAIK the benchmark results are used by the project scheduler to estimate the amount of work needed by the computer in question.

 

[somethingsketchy]

I'm getting a bit over 4k FP MIPS. AFAIK the benchmark results are used by the project scheduler to estimate the amount of work needed by the computer in question.

+1, what mk said.

 

[Smartazz]

Okay, thanks. Unfortunately since this is a new install of Windows, it periodically restarts my computer and interrupts BOINC. Anyone else have problems with that?