looks like i hit the lotto

Nagasama

disgruntled ex-moderator
Messages
6,956
Location
probably on the lake
The haswell lotto, that is.
Just playing around real quick with my chip it will hit 4.4 on stock volts everything on auto.
4.6 at 1.2 is as high as I've messed with it, temps under 80 during ibt.
Betting when I fine tune and my tim cures I can get 4.8+ at 1.25v or less with acceptable load temps.

Yay me.

This thing smokes everything in it's path at a measley 4ghz. Really even at stock.
Bf3 full 64 server barely breaks like 45% Cpu.
And my 780 is pegged at 98-99% usage. Yes.
 
From what I have read p95 no bueno for haswell.
Supposedly the best for haswell is aida64 or Intel xtu.
I only use ibt as an on they fly test to discover glaring instability in any event.
I've had p95 test out for ten hours before failing on settings that fail ibt in ten minutes or less.
And really ibt is not suggested as it has the possibility of overvolting a haswell.
 
IBT runs into instability issues because it gives an unnatural amount of load to the CPU which can cause issues just like Furmark fries GPUs for the same reason. I used to use IBT myself but have switched to strictly Prime95 with no issues. Prime95 is perfectly acceptable for Haswell.

IBT can't change voltages, it's only a stress test. So the last part is false. Part of the VRM circuitry is on the chip itself, but nothing else has changed really.
 
From what I have read aida64, p95 and ibt can all cause overvoltage if voltage parameters are not set up properly. Meaning they don't actually change the voltage BUT if you have something wrong in bios adios mr Cpu. Which is the case for any kind of stress test if you're rookie enough not to have them set up improperly, I suppose.
Intel XTU supposedly will not allow the Cpu to become overvolted under any circumstances.
It is also the only one out of the four that Intel updates (obviously), and it and aida64 are validated for haswell instructions.
Just what I read. I have always used p95.in the past and I know ibt is a brute dude and it puts out some serious heat.
Now im curious what my temps will be with ixtu and p95. Much less I bet than ibt.
Gonna give it a go this evening.
 
The CPU can only pull the set amount of volts designated by the bios. The only time anything like that can happen is if you're running software to overclock in which the CPU can become unstable which can glitch the software and shoot more or less volts than intended. Even still, you have to be running a high enough amount of over provision. But, if over temp and overvolt protection are enabled in the bios they will automatically kick in saving your CPU. That's why I tell people if you're running decent cooling and aren't a total noob it's impossible to fry a CPU unless you're just running **** hardware. All the precautionary measures put into place prevent certain things from happening but programs like IBT put unnatural loads on other elements not designed for such loads much like Furmark does to GPUs. That is why on Bjorn3D all reviewers must not use IBT or Furmark for stress testing. Had a big argument with a fellow reviewer about all this about a year ago so not all the specific details are in my mind but most are.

In any case, I wish I had known way back in the day. In my S939 days I used to use IBT to test my X2 4800+ and it would always fail despite running stable 24/7. Could probably have run a higher OC if I used Prime for testing instead but hey live and learn.
 
Hahaha
Yeah I set it to large fft and let er rip while I went to the gym.
About to head back now, so I'll see what's up.
Temps were barely over 50c when I left.
 
Last edited:
Sweet. Cpuz shows 1.152vcore (stock voltage for this chip), running 4.4 since 5:30pm no problem.
Large fft
All cores within 5c of each other, hottest was 61c, all others mid 50s.
Yuuuup.
I think I hit the haswell lotto.
Gonna run blend all night and see what happens.
 
Last edited:
Back
Top Bottom