AMD Phenom II X4 965BE safe operating temps

[KeyboardDemon]

I recently had some stability issues, I've fixed this with some minor CPU voltage tweaking. But I was wondering if anyone could tell me what the safest max temperature I can my CPU at is?

Right now I'm knocking on 59 degrees with Prime95 running, it's quite possible that after a few hours I might see temps of around 62 to 64 degree, this is with my CPU at 1.40v.

[ehhhhhhh]

Those temps are fine in my opinion.

I had mine @ 68C (load) back then.

[J0s3e]

Usually people aim for threshold of 80 degrees C. For me, mid 70s are OKish, also considering the weather... It was really hot a week ago, and my CPU was maxing at 79 while running prime on a really hot day. Normally it doesn't hit 75. So I guess You could OC a bit higher, if you feel like

[deejayburnout]

I am sure that Phenom chips don't like temps over 65deg before the ramp down to save themselves.

My chip runs around 57 on full load and 32 on idle and I am OCed to 3.8.

Not sure what yours are set like?

Posting on the go

[dek100]

This is the same processor as I have. Before I got the case I am using now, my temps were 65 at load, now they are 55. 65 will not hurt your CPU
Maybe think of better cooling though.

[dek100]

My cpu is also overclocked to 3.8

[KeyboardDemon]

I was running my CPU at 3.91 and everything was quite stable till the warmer weather started and now I'm getting blue screens, in the crash reports it say that WHEA recorded shutdowns due to thermal issues and my temps fully loaded were at 64 degrees according to HWMonitor. AMD state 62 degrees but I'm sure that's factoring in a safety margin, however if the margin is only 3 degree or about 5% it is smaller than I thought it would be. Looks like I might have to run a lighter overclock instead of doing what I really wanted to do and push over the 4.0ghz barrier.

I have changed my heatsink but I think I need to re-seat it as I noticed that the backplate wasn't quite straight while I was putting my case back together. I have swapped my Corsair H50 for an NZXT Havik 140.

[deejayburnout]

if I get over 57 deg on my CPU start to worry.

I do however clean my rig of all dust every month. Take the fans off and clean my CPU cooler. Even with all the filters I have, its surprising how much dust gets in there.



Posting on the go

[HalloweenJack]

my X6 @ 3.6 is at 68 on a passive hsf

[KeyboardDemon]

Well I gave my rig a little checking over today, not reseated the cooler yet, thought I'd check it to see if it's making proper contact which it looks like it is, but I'm getting 58c fully loaded on stock with a room temp of 32c (it's hot in this room!) and internal case temp is 40c.

[watercooled]

According to the AMD spec page, max recommended temp is 62C: http://products.amd.com/pages/Deskto...f10=&f11=&f12=

Are you using the stock heatsink?

[KeyboardDemon]

According to the AMD spec page, max recommended temp is 62C: http://products.amd.com/pages/Deskto...f10=&f11=&f12=

Are you using the stock heatsink?

I guess I should have worded my title better, I was really trying to gauge the difference between the recommended max and the safest max temps, reading the replies above I can see that people have been safe enough to go beyond the recommended max but not by a big margin.

Also I was running a Corsair H50 but I'm in the process of swapping to a NZXT Havik 140 cooler as the all in one water loop didn't seem to be performing as well as I had hoped it would.

[watercooled]

Yeah there's usually a fair gap between Tcase and TJmax/thermtrip, but AMD don't actually state the latter AFAICT. However, going by Intel's specs, it's probably around 80+.

However, Tcase is still the recommended max case temp, TJmax/thermtrip is the temperature where power will be removed from the CPU to stop it destroying itself. This is also why, as I recently posted elsewhere, it's unfair to compare Intel TJmax to AMD's Tcase temperature as quite a few people do.

You would get away with even hitting thermtrip a few times, it shouldn't cause immediate damage (which is why it's there), but running a CPU very hot, especially with frequent thermal cycling, may kill it over time.

[KeyboardDemon]

Now it looks like I'm in need of a little education (maybe a lot of education!) and your summary of TCase and TJmax have helped me to a certain degree, thermtrip is also useful, I'm guessing that if thermtrip is triggered it will pass the message with the Thermal issues warning to the Windows Error log file or System memory dump and in turn this is what gets picked up and reported by WhoCrashed, which is how I know that the latest crashes I was getting were down to thermal issues as that's what I was getting from the WhoCrashed error summaries.

It seems obvious that the only way I can reduce TCase would be to reduce the room temperature or wait for colder weather, till then I really need to make the time to deal with reseating my heatsink, I don't think this is the problem but it is an easy fix so it's worth doing.

[watercooled]

Tcase doesn't refer to the computer case, it is measured by Intel/AMD by drilling a channel in the heatspreader and putting a thermocouple in, there's no exact way to measure it like that in software but the closest will probably be socket/CPU temperature, but that and core temperature should be within a few C anyway so it's not that important. Basically it's just the max temperature the heatspreader should reach.

Occasionally, thermal events are logged erroneously; for example I've seen some motherboards reset and report a thermal event when power to the board is briefly interrupted, I could be wrong but I don't think any software would be able to positively log thermtrip because once the CPU sends the signal to the motherboard, by spec power must be cut immediately which would also effectively destroy anything held in RAM.

However, to further confuse things, there's another signal called prochot, which AFAIK doesn't trip the CPU but is the point where it will start to throttle and presumably Windows would know about it but I'm not certain how it responds; it may just carry on, pop up some warning, or BSOD.

[deejayburnout]

the way i see it with temp is keep it as low as possible.

In real terms your cpu will never be running at full load for any length of time so those peak temps are not a real refection of normal everyday use.

if i were hitting 60deg on 100% load, i would do something to fix it.