IBT question at 4.5 GHz

Jman13

Senior member
Apr 9, 2001
811
0
76
So curiosity got the better of me today, and while I am not going to run it at this speed now, I just wanted to see what kind of temps I get and if my machine will easily handle 4.5 GHz.

I set it to 4.5 and had voltage at 1.28V at first, and it loaded Windows fine, but Prime95 errored out within about 3-4 minutes. I upped the voltage to 1.308V, and Prime 95 seemed happy for the short little time, but since this was more an experiment rather than a true stress test for stability, I stopped that and opened up IBT for a few runs at Very High settings.

It started running, and never gave an error or BSODed, but the first run was fine...the second run was fine, and had decent speeds. The third and fourth runs, however, were very odd.

They were slower (about the same speed as I get with the machine at 4.1 GHz), and the fans in my system kept speeding up and slowing down, as the load was not pegged at 100% as it usually is during an IBT test. It would load at 100, then drop to 95, load to 100, drop to 90...it was kind of strange.

IBT load temps on the second run peaked at 84C, which is hotter than I'd like, but not unexpected given my cooler. Prime95 was running in the low-mid 70s (and I'm sure actual use would be in the mid-high 60s). On the third and fourth runs, IBT peaked at 79, but due to the varying load would jump from mid 50s to high 70s.

What was happening with the run? While I've reset to 4.1 and plan to keep it there for the time being (don't need the speed...will use 4.5 as a nice bump in a year or so.), is this some odd instability thing with IBT, or was it possible just other CPU tasks were taking focus off of IBT?
 

TakeNoPrisoners

Platinum Member
Jun 3, 2011
2,600
1
81
I think it has something to do with the voltage not being high enough and the SB error correction is stepping in and slowing down the processor.

Try bumping the voltage up a bit.
 

VirtualLarry

No Lifer
Aug 25, 2001
56,352
10,050
126
I think it has something to do with the voltage not being high enough and the SB error correction is stepping in and slowing down the processor.

Try bumping the voltage up a bit.

That wouldn't explain the fans, would it?

It might have been throttling.

Also, what did you set your long and short power limits to? If you don't increase them enough, it won't stay in turbo mode multiplier for long.
 

Jman13

Senior member
Apr 9, 2001
811
0
76
Thing is, the frequency never budged. Anyway, it ran fine (but too hot for my tastes) at 1.34V. But I went for a small bump up and am comfortably running at 4.2 and 1.23V now.
 

Smoblikat

Diamond Member
Nov 19, 2011
5,184
107
106
Thing is, the frequency never budged. Anyway, it ran fine (but too hot for my tastes) at 1.34V. But I went for a small bump up and am comfortably running at 4.2 and 1.23V now.

Im glad it worked out. I would assume it was due to error correction and throttling that caused the weird run through.
 

ehume

Golden Member
Nov 6, 2009
1,511
73
91
I have done hundreds of runs testing fans using OCCT 3.1/Linpack, which maximally stresses my i7 860. I have left this rig for many many runs using the same multiplier, BCLK, Vcore, Vdimm and other settings. They remain the same because I am comparing the cooling power of different fans.

I have learned to start the Linpack run as soon as possible. And then I must watch the first minute or so of the hot part of each run. Then I can leave the rig alone for the rest of the run.

Why do I do these things? Sometimes Linpack just doesn't seem to catch on properly. The temp trace waggles up and down. It is more likely to do this if I let other apps load before Linpack starts. And even then it occasionally doesn't catch.

This sometimes occurred when I had my Vcore set on notch higher.

I raise this issue because OCCT/Linpack and IBT both use Linpack.

Bottom line: don't give up on a set of OC settings on the basis of a single Linpack run. Just shut it down and try it again.