While I was doing stablility testing with Prime9.5 with all 8 threads maxing out. Core Temperatures are closing in 100C. After I got back from a bathroom break, The machine was shut-down. Subsequently attempts to reboot it, it failed to post, the fans aren't spinning and no lights on motherboard were brinking. What's my problem?
I assume that the cpu was overheating and auto shut-down or crashed or whatever. Why didn't that core i7 cpu throttling down feature kick in when overheatting?
This is 2nd core i7 I built and oc'ed. Never seen this problem before even when temp was over 100c during stress test. Usually when if it crashed, I was able to reboot again.
No, I haven't clear CMOS because I thought it has dual BIOS, why should I? It seems to have a sympton of over-heating circuit of psu. I wonder if any of you have run into this before. I am just hoping that nothing got fried, but it needs time to cool down first.
You didn't manually set the voltage? I bet the reason for the high temps were because motherboards tend to be generous (to put it mildly) with the voltage when left on auto and overclocked. Also, those temps are way high. Although they shouldn't fry the CPU, they're way out of spec, and I would keep the CPU temps to 80C at most (full stress test) whenever possible. Oh, and you said you've exceeded 100C before? That's surprising - the i7 should throttle at 100C (don't ask me how I know), and cool down almost instantly when it throttles.
This is the follow up. So the something can be noted.
GA-EX58-UD3R mobo was dead, everything (PSU, core i7, mem) else was fine because I was able to swap parts to verify that. Now I got exact the same mobo replacement and is running fine. Yes, overheating caused the failure of the mobo.
The cpu that fried the mobo was core i7 D0 stepping. Apparently, this core isn't 'good' low temp core compared with my other core i7 C0 stepping.
my D0 49C 100C
my C0 40C 90C
This D0 stepping core can often reach 100C during stress test(Prime95) while the C0 stepping rarely reach over 90C. Temps are measured by Core Temp.
The throttle isn't kicked in when it's over 100C. I even have BIOS warning set 90C and never heard or seen anything happend when Core Temp reports over 90C. The problem is that I really don't know the real temp the cores are, just rough guess range from reading the temps from Core Temp or Real Temp or SpeedFan.
ARCTIC COOLING Freezer XTREME is my cooler. I wish it had higher rpm(max=1500)
At default speed @2.66 GHZ, core temp is already at 72C during load. Don't tell me that I didn't seat cpu correctly.
The voltage is bios default. I didn't change it. cpu is oc'd @3.8GHZ. vcore=1.36v during load reported by cpu-z.
100C of core temp scares me too. That's why I am looking for reasons. I don't know whom I should trust: Core Temp, Real Temp, or SpeedFan. I somehow suspect that this particular core i7 D stepping I have messed up the temp readings of Core/Real Temp/SpeedFan. During Prime95 stress test with all 8 threads @ 100% utilization:
In my BIOS, I also have lower CPU Warning Setting to 80C and it never kicks in and never warn me. So it appears BIOS doesn't think CPU temp is over 80C yet. I've tried this cpu with two mobos (same kind) and the same result. So I know it has nothing to do with mobos.
Oh, right. You could've also suffered an incorrect overclock.
Getting to 3.8 GHz without altering some of the BIOS settings which are normally at "Auto" can fry your system.
Have you checked the DRAM voltage, the Vcore, etc? Also, there should only be a .5 v spread between the uncore voltage and the DRAM voltage. Anything more means that your system is likely to get VERY hot and die.
Coool122, you are absolutely right about that I suffered an incorrect overclock which I wish I could know this sooner. That's the exact problem with my high core temps.
By default, BIOS has VCORE at auto mode. If you don't overclock at all, the VCORE would never go higher than 1.22v at auto mode.
However, if you ever overclock with vcore at auto mode, VCORE would automatically jump to 1.36V which is very high voltage for any air cooling even a good aftermarket cooler. The temp can reach up to 100C with 8 threads max out.
Core i7 920 has overheating protection circuitry, but this motherboard would die before CPU ever reaches its throttling down temperature level. That was the case for me. I've never ever seen or experienced CPU throttling down. However, some motherboards can withstand the heat while others overheat and die.
So don't use Auto setting for vcore because it's not necessary for mild OCing. CPU can easily be OC'ed up to 3.8 GHZ w/o ever raising its "Normal" voltage which is at 1.22V.
After setting my VCORE=Normal mode, my temperature is much lower than before while still enjoying 40% OC @3.8 GHZ. Idle=40C, Load=80C.
Why did motherboard manufacturers set VCORE=Auto by default? I don't see any benefit of doing this, but RMA'ing dead motherboards.