Prime95 asymmetric completion - Overclock.net

Quantum Reality · 4 Weeks Ago

What I mean by the subject title is that my prime95 test of my Athlon II X4 620 at stock speeds seems to produce results where one core "falls behind" and from then on the cores consistently tend to finish the worker threads about the same time apart instead of in complete unison.

I've noticed this effect with my Q9550 but always put it down to Windows loading core 0.

That said, I did NOT install the Dual Core Optimizer, but I DID install the AMD CPU driver in XP.

Can anyone shed light on whether this could be a major issue?

Quantum Reality · 4 Weeks Ago

Incidentally this happens whether I run the blend or small FFTs.

Nenkitsune · 4 Weeks Ago

that's fine. it's because one core has to process the OS things as well, so naturally it's going to lag behind since it's doing other work.

Quantum Reality · 4 Weeks Ago

But this badly? Look at the attached picture.

Also, for some odd reason I'm now getting I/O errors with my USB stick and it seems to be taking a long time (way more than usual) to copy files.

Nenkitsune · 4 Weeks Ago

That's not bad. that's really normal.

Quantum Reality · 4 Weeks Ago

Normal for the prime95 to go out of sync by over 2 minutes between cores AND my computer to start throwing weird errors with USB devices after a P95 run? I'd be more inclined to suspect a dodgy CPU.

Edit: Actually it may be that my case's front USB ports don't work very well. I tested using the motherboard's back USB ports and they seem fine.

Dopamin3 · 4 Weeks Ago

Quote:

Originally Posted by Quantum Reality

Normal for the prime95 to go out of sync by over 2 minutes between cores AND my computer to start throwing weird errors with USB devices after a P95 run? I'd be more inclined to suspect a dodgy CPU.

Edit: Actually it may be that my case's front USB ports don't work very well. I tested using the motherboard's back USB ports and they seem fine.

No, it is a memory issue. Due to the IMC.

Quantum Reality · 4 Weeks Ago

Yes, but that test in my screenshot was small FFTs, which almost completely cuts the memory out of the picture.

jimibgood · 4 Weeks Ago

Quote:

Originally Posted by Quantum Reality

What I mean by the subject title is that my prime95 test of my Athlon II X4 620 at stock speeds seems to produce results where one core "falls behind" and from then on the cores consistently tend to finish the worker threads about the same time apart instead of in complete unison.

I've noticed this effect with my Q9550 but always put it down to Windows loading core 0.

That said, I did NOT install the Dual Core Optimizer, but I DID install the AMD CPU driver in XP.

Can anyone shed light on whether this could be a major issue?

Quote:

Originally Posted by Dopamin3

No, it is a memory issue. Due to the IMC.

There are many theories to this from volatge to the cores running at 100% and processes running in the background leading to the cores getting out of sync. Intel quad cores go through the same issue. I don't think it has to do with memory.....or IMC.

Long/short....Don't worry about it. As long as you pass and get no errors or crashes your golden.

Dopamin3 · 4 Weeks Ago

Quote:

Originally Posted by Quantum Reality

Yes, but that test in my screenshot was small FFTs, which almost completely cuts the memory out of the picture.

I was saying that because your USB ports or whatever were failing, which could've been memory issues even if you were running small fft. But I guess your case headers are just bad.

		Overclock.net - Overclocking.net > AMD > AMD CPUs
Prime95 asymmetric completion

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)