WinXP, Win7 and IBT temps

[strumpusplunket]

I made the change from XP to 7 at Christmas. Yesterday I made the decision to try to push my overclock a bit further on my Q6600. It's been stable at 3.2GHz for ages, but seeing some people here put much higher voltages through the chip than I have made me want to try again for 3.5GHz or so.

I upped the vcore two steps (I knew one step wouldn't do it) and set my clock to 3.4GHz, then tried IBT for 8 passes. Didn't pass IBT at that vcore or the one up from that. I was about to start pushing NB and SB volts too, but I noticed that in my last IBT test my temps had gone way higher than I was comfortable with, so I clocked back to my old vcore and 3.2GHz.

Now, here's the oddity:
I ran IBT again to make sure all was still well. It passed okay, but my temps were far higher than the last run under XP. Same vcore, same clocks, exact same system, case, system temp, everything, but:

IBT under XP:
Max temps on cores: 67-67-65-63
Max PWM temp: 60C
System temp: 30C

IBT under 7 (64-bit):
Max temps on cores: 75-75-73-73
Max PWM temp: 79C
System temp: 29C

I used compressed air to blow some dust out of my case, but it didn't make any difference (I keep it fairly well-dusted anyway, so I didn't really believe that could account for such a rise).

This is a huge leap in temps. The only variable that's changed in between the tests is the OS. Is this 8C difference really just down to the 64-bit OS allowing IBT to test with more memory? Any other explanation?

 

[Geckovich]

You made the change from XP to 27 at Christmas you say? I assume then that your last IBT run was during the cold crappy weather we were having, whereas your current runs will be during this fairly warm weather.

I can't see it making too much difference but it will have definately contributed.

 

[strumpusplunket]

To be honest, I think I can rule that out.

Although I made the switch from XP to 7 at Christmas, that wasn't when I last ran IBT under XP. My overclock had been stable for some time before that.

The last IBT run I did under XP was in the spring and not at Christmas, and the house was probably slightly warmer then than it is now. In fact, my records show a system temp 1C higher in my last run under XP than the one yesterday.

I can understand a slightly higher temp if Win7 accesses more memory for a more exacting test, but 8C on the cores and 19C on the PWM just seems crazy.

 

[Geckovich]

Perhaps your CPU cooler has come slightly loose?

I'm clutching at straws now if I'm honest, I just can't see an OS change making that much difference on it's own.

 

[strumpusplunket]

Aye. I suppose that's possible. It's a big brute of a thing (a Scythe Zipang - only mediocre cooling, but whisper-quiet). It still seems to be fairly stable, though. And wouldn't I expect to see a variation in my idle temps as well if it was loose? (Idle temps remain at about 35C, same as always).

Hmmm... I guess the next logical step would be to check temps under P95 as well. A shame that, as I switched to IBT for my test of choice a while ago, I only have records of my previous temps under P95 at a slightly lower clock and a 0.06 lower vcore...

Anyone else noticed a significant difference in load temps between 32- and 64-bit runs of IBT?

 

[SirCanealot]

Doesn't IBT test run "better" when running in 64 bit mode? Perhaps it's stressing your CPU MORE now it's running in 64 bit mode...

I'm sure I read that somewhere anyway; I recently started using IBT a little

 

[strumpusplunket]

Yeah, this is the only explanation I can think of that fits the facts. I was expecting a sterner test given more RAM is brought into the equation under a 64-bit OS with 4GB available, but I was really taken aback by the scale of it. 8C on the cores and 19C on the PWM is really a huge jump in temps!

As I've not been able to find anyone else reporting this, though, I thought I'd discuss it here to see if anyone else has experienced it, or if there's something I might be missing.