Take the 2-minute tour ×
Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.

I have two similar systems, but one is running a bit slower, and I can't figure out why.

System 1

time convert rose: -resize 5000x5000 -bench 5 /tmp/rose.png
Performance: 5i 0.221729ips 26.050u 0:22.549

real 0m22.554s
user 0m25.795s
sys 0m0.262s

System 2

time convert rose: -resize 5000x5000 -bench 5 /tmp/rose.png
Performance: 5i 0.220167ips 28.710u 0:22.710

real 0m22.731s
user 0m25.553s
sys 0m3.172s

While the overall time is about the same.. The sys time (time spent in kernel) is over 10 times as long.

What would make this happen? What can I look into to figure this out?

Memory is the same and /tmp is not full.

Since these are Core i7 2600's I'm thinking it might have to do with the chipset recall and slower drive performance. Or maybe they were set up differently.

share|improve this question

migrated from stackoverflow.com Feb 6 '11 at 20:42

This question came from our site for professional and enthusiast programmers.

    
How different are the two /tmp/ partitions on both systems? –  eumiro Feb 6 '11 at 18:27

1 Answer 1

If system 2 is rather tight on RAM, that could explain it: the kernel is spending time swapping out other apps. If my theory is right, the results should not be very repeatable because system 2 will have freed up enough RAM for a second run.

Otherwise, eumino's comment about differences in /tmp could possibly explain it (/tmp is almost full and fragmented?) but that's a less likely scenario in my opinion.

share|improve this answer
    
same ram: system 1: free total used free shared buffers cached Mem: 4035356 1777364 2257992 0 30452 434468 system 2: free total used free shared buffers cached Mem: 4000168 272900 3727268 0 16464 160880 -/+ buffers/cache: 95556 3904612 Swap: 2096472 0 2096472 -/+ buffers/cache: 1312444 2722912 Swap: 2096472 647572 1448900 –  Dognose Feb 6 '11 at 18:49

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.