How long before my CPU is toast? (Haswell @ 100C)

U

USMC2Hard4U

Supreme [H]ardness
Joined
Apr 4, 2003
Messages
6,157
So I have a question and would like to make sure I didn't kill my CPU. Rig is 4790K @ 44x100 / 1.255v on H80i push/pull - Mem @ 1866 8-8-8-24-1T @ 1.55v

I have been stressing for about 48 hours with different tools because I want to keep my temps at a certain point.

I did LinX for about 8 hours, CPU max was about 91C
OCCT - Small/Med/Large for about 4 hours, Max was about 89C
AIDA64 Stress Test CPU/FPU/Cache/Mem (Did them individually and together) Max was about 90C
x264 v2 Bench for about 4 hours - max 72C


So I figured I knew about what my max temp was... I ran Prime95 Blend and for a period of about 20 minutes while I stepped away, I was pushing 100C+ and was throttling.

When I came back and noticed that my CPU graph was Bouncing between 95C and 101C for like 20 minutes and my CPU went from 4.4GHz down to whatever it throttled to, I turned off Prime95.

I have experienced no lockups, reboots, bsods... but just want to make sure I didn't totally own my CPU keeping it at 100+C for 20 min.

Thanks all!
 
I think Intel has said that heat won't kill their cpus, but voltage can. Your voltages seem reasonable so you should be OK in terms of lifespan. Of course the temps would worry me a bit and I would either back off the overclock a bit or get better cooling, but maybe that's just me.
 
My idle temps are in the low 30s and load temps for anything not a stress test are in the 50s and 60s.

I'm ok with my temps, I just wanted to make sure that going over 100c for 20 minutes wasn't like, game over. It throttled appropriately which I'm thankful for, but still scary.
 
I'm sure just like anything, temps will slowly kill the CPU over time. My fear is not if I reduced the average life span of my CPU my a few months on the backend, it's more of a concern of did I just give my CPU a short term, very near death sentence. Chances are, as soon as broadwell comes out, this chip is going to get replaced. 6 months to a year from now.

Everything I read online is "don't hit 100c no matter what" and I did for about 20 minutes. So far, I see no adverse affect.... Should I have?
 
USMC2Hard4U said:
I'm sure just like anything, temps will slowly kill the CPU over time. My fear is not if I reduced the average life span of my CPU my a few months on the backend, it's more of a concern of did I just give my CPU a short term, very near death sentence. Chances are, as soon as broadwell comes out, this chip is going to get replaced. 6 months to a year from now.

Everything I read online is "don't hit 100c no matter what" and I did for about 20 minutes. So far, I see no adverse affect.... Should I have?
Click to expand...

No I don't think so. I throttled the crap out of a couple 4770Ks, the most recent one for longer than you did and I've been using it for about 5 months and it's never changed its voltage needs or anything. Same chip as before, but delidding it fixed the throttling issues.

Also I wouldn't worry about prime95 blend eventually going berserk on you. The effect is after about 30 minutes of blend or so, it starts running very Small FFTs and causes higher temperatures than any other setting or any other benchmark. My delidded 4790k is near the throttling point at 4.8Ghz with Small FFTs on prime95, using a H55 :p So because of that I can't really tell how stable it is, (my POV is that a minute of Small FFTs is probably good enough to call it mostly stable if it doesn't crash) or if it could be stable on prime at any higher than 4.8.

I never really use blend, I set up custom ranges for prime so it doesn't surprise me like that. If I use small FFTs it's for a short term test. Basically it will tell you if your RAM settings are going to trip the CPU up when it gets full loaded with heavy computational math. I tend to run HIGH FFTs for a while, then Low FFTs for a very little while, and if all is good I'm done. These simple tests create a scenario that if your settings pass... you are extremely unlikely to ever encounter any instability during regular performing applications.
 

Thank you for the response :) maybe I should attempt a delid.

Edit: I was using blend because I wanted to test my ram as I took my timings from 9-9-9-27-2T to 8-8-8-24-1T and wanted to give it a workout.
 
Last edited:
USMC2Hard4U said:
So I have a question and would like to make sure I didn't kill my CPU. Rig is 4790K @ 44x100 / 1.255v on H80i push/pull - Mem @ 1866 8-8-8-24-1T @ 1.55v

I have been stressing for about 48 hours with different tools because I want to keep my temps at a certain point.

I did LinX for about 8 hours, CPU max was about 91C
OCCT - Small/Med/Large for about 4 hours, Max was about 89C
AIDA64 Stress Test CPU/FPU/Cache/Mem (Did them individually and together) Max was about 90C
x264 v2 Bench for about 4 hours - max 72C


So I figured I knew about what my max temp was... I ran Prime95 Blend and for a period of about 20 minutes while I stepped away, I was pushing 100C+ and was throttling.

When I came back and noticed that my CPU graph was Bouncing between 95C and 101C for like 20 minutes and my CPU went from 4.4GHz down to whatever it throttled to, I turned off Prime95.

I have experienced no lockups, reboots, bsods... but just want to make sure I didn't totally own my CPU keeping it at 100+C for 20 min.

Thanks all!
Click to expand...

Yeah can definitely dmg your cpu... but it will have to be much hotter than 100C and much much longer periods of time. High voltage is a bigger problem. Do not go above 1.4v and you should be fine. Even running above 1.4v will be fine for short periods of time.
 
You must log in or register to reply here.
Back
Top