• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

BOINC benchmark question

biodoc

Diamond Member
I've been crunching Malaria for a while and I noticed my linux box (ubuntu 64 bit gutsy gibbon) was out performing my Windows XP box. Both are Q6600s @3.2 GHz.

I installed linux on the XP box (dual boot system now) and here are the boinc benchmarks:

Windows XP:

Measured floating point speed 3186.16 million ops/sec
Measured integer speed 7034.12 million ops/sec

64 bit linux:

Measured floating point speed 3214.15 million ops/sec
Measured integer speed 8949.37 million ops/sec

Any thoughts on why the integer benchmark is much higher on the linux box?

My PPD have gone up significantly since I switched both to linux.

 
Before I got rid of XP64, I could run BOINC benchmarks a half-dozen times and get a half-dozen different answers for the Integer Speed.

Under Linux, I get consistent numbers.
 
You're lucky biodoc. Usually, Linux scores lower in the BOINC benchmarks. You can see that phenomenon spread over almost all BOINC projects. Some admins even raised the score of WUs that were computed using Linux to even the board. Of course using the 64bit OS might make the difference here vs WinXP 32bit. But to be sure, you should run the benchmarks at least a couple of time. Just as Fred said, I've seen these benches differ by margins that are just too crazy to imagine! 🙁
 
Originally posted by: BlackMountainCow
You're lucky biodoc. Usually, Linux scores lower in the BOINC benchmarks. You can see that phenomenon spread over almost all BOINC projects. Some admins even raised the score of WUs that were computed using Linux to even the board. Of course using the 64bit OS might make the difference here vs WinXP 32bit. But to be sure, you should run the benchmarks at least a couple of time. Just as Fred said, I've seen these benches differ by margins that are just too crazy to imagine! 🙁

Yeah, it does make sense that the integer benchmark would be higher on a 64 bit OS.

Strebor, I have the same RAM (manufacturer, speed, etc) in both systems since I built both at about the same time.

rabrittain, linux lighter and more efficient that XP? I'm not sure about that but you could be right.

Rudy Toody, now that both boxes are running linux, the numbers are pretty close:

box1:

Measured floating point speed 3214.15 million ops/sec
Measured integer speed 8949.37 million ops/sec

box2:

Measured floating point speed 3217.11 million ops/sec
Measured integer speed 8970.62 million ops/sec


Thanks for the feedback guys!🙂
 
I have the same RAM (manufacturer, speed, etc) in both systems since I built both at about the same time.

Guess that's not it then. I've got two identical P4 630 systems, except that one has memory in dual channel and the other just has one stick. The one with just one stick is always a bit behind. This was true in Red Hat and XP.
 
Here is a comparison of my FX-60s (all using Linux64.)

blueheron1:
2485 floating point MIPS (Whetstone) per CPU
7044 integer MIPS (Dhrystone) per CPU (has 2GB memory)

blueheron2:
2484 floating point MIPS (Whetstone) per CPU
6870 integer MIPS (Dhrystone) per CPU (has 4GB memory ---mySQL and apache2 running in background)

blueheron3:
2479 floating point MIPS (Whetstone) per CPU
6944 integer MIPS (Dhrystone) per CPU (has 2GB memory ---mySQL running in background)
 
Back
Top