Whats it mean when

[Gerard]

I run the sp2004 test and one core seems to be behind in doing the calculations compared to the other core?

Example:

As of right now the core on the right is onto the second test in that catagory and other one on the left is still doing test 6 of the 14k catagory.

 

[FreeStream]

Normal with Conroe, probably normal with AMD too.

One core is slightly quicker than the other, gets more mem b/w or has the most choice in operations or something.

 

[FrostedNipple]

yes its all down to windows choosing which core to do what on, and as iteration is not one thing but multipole calculations each calc will go to a core, and it wont be 100% 50-50 as you can see there

as long as it reports both cores to be the right speed and its stable then you have nothing to wory about

 

[Gerard]

ty

Always wondered about that, sometimes when i was stress testing it seemed the core that was behind was always the one to fail so i assumed this was a sign of a core not being able to handle the clockspeeds.

 

[Final8y]

I get that when im doing things on my comp while priming.

 

[Corasik]

yes its all down to windows choosing which core to do what on, and as iteration is not one thing but multipole calculations each calc will go to a core, and it wont be 100% 50-50 as you can see there

as long as it reports both cores to be the right speed and its stable then you have nothing to wory about

Actually thats wrong, That test specifically says what core its running on, so it very likely automatically sets affinity correctly for each core, so it doesnt keep jumping from core to core.

However, dont forget that windows is running too, and that uses cpu power, so in addition to the tests, you probably have at least 20 other 'windows' background and housekeeping tasks running, which will cause the cores to have a slightly different performance, depending on how much background work is going on.

 

[El Jimben]

Thanks for clearing that up, I'd always wondered that too!

 

[w3bbo]

However, dont forget that windows is running too, and that uses cpu power, so in addition to the tests, you probably have at least 20 other 'windows' background and housekeeping tasks running, which will cause the cores to have a slightly different performance, depending on how much background work is going on.

Q F T