How hot is too hot?

[Oliver davies]

Hey, I'm using a cooler master hyper 212x to cool a xeon x5650. under full synthetic load I hit max temps of about 95*C after about half an hour. should i upgrade my cooler. and if I should what too. (budget is about less than £40)

Many thanks!

[Jurrunio]

Take the cooler off and check your previous thermal paste application. Does it cover most of the IHS?

[Oliver davies]

1 minute ago, Jurrunio said:

Take the cooler off and check your previous thermal paste application. Does it cover most of the IHS?

yeah it does last time I reapplied. I know the interface is good but I need to know if the temps are too high.

[LukeSavenije]

Well, anything above 80 isn't going to be good for that cpu. Maybe a artic freezer 33 esports one will help you out?

[Oliver davies]

1 minute ago, LukeSavenije said:

Well, anything above 80 isn't going to be good for that cpu. Maybe a artic freezer 33 esports one will help you out?

seems to be a good idea ill have proper look at reviews

 

[LukeSavenije]

Just now, Oliver davies said:

seems to be a good idea ill have proper look at reviews

 

Well, there is a model of it that can keep treadripper cool, so yeah...

[Oliver davies]

Just now, LukeSavenije said:

Well, there is a model of it that can keep treadripper cool, so yeah...

They have large IHS's tho and large die areas so that may help. does anyone know if you can delid x58 chips?

 

[corrado33]

I tend to say that 75 is the max I'd be happy with for sustained long term use. Although that's a bit conservative. On all my folding rigs I set the max temp to 73 and I'm happy. Then again, I have good coolers for all of them so they can pretty much run full out and never reach that. 

[Oliver davies]

1 minute ago, corrado33 said:

I tend to say that 75 is the max I'd be happy with for sustained long term use. Although that's a bit conservative. On all my folding rigs I set the max temp to 73 and I'm happy. Then again, I have good coolers for all of them so they can pretty much run full out and never reach that. 

Oh ok ill see if i can drop some of the overclock in that case

 

[Jurrunio]

44 minutes ago, Oliver davies said:

yeah it does last time I reapplied. I know the interface is good but I need to know if the temps are too high.

they are high, but it could be just not enough mounting pressure. I got 75C on the NH-D15 on my 2600k at stock at the first time, but 60C the second time when I literally torqued it down.

[qqqqqq]

You can delid X58 chips just as you can any other Intel CPU (mostly), it will yeild great results and may even result in not needing to upgrade your cooler.

[NelizMastr]

17 minutes ago, qqqqqq said:

You can delid X58 chips just as you can any other Intel CPU (mostly), it will yeild great results and may even result in not needing to upgrade your cooler.

Delidding a soldered CPU like this isn't that great of an idea, though. 

 

18 hours ago, Oliver davies said:

Hey, I'm using a cooler master hyper 212x to cool a xeon x5650. under full synthetic load I hit max temps of about 95*C after about half an hour. should i upgrade my cooler. and if I should what too. (budget is about less than £40)

Many thanks!

So is this at stock settings or overclocked? For stock this is very high to a degree where I also suspect poor contact. You did peel off the plastic protector film on the cooler, right?  

[Dedayog]

19 hours ago, LukeSavenije said:

Well, anything above 80 isn't going to be good for that cpu. Maybe a artic freezer 33 esports one will help you out?

Where do you get 80 from?  How much not good is going to happen?

 

This is synthetic load tho, he needs to check real gaming temps.

[LukeSavenije]

4 minutes ago, jstudrawa said:

Where do you get 80 from?  How much not good is going to happen?

 

This is synthetic load tho, he needs to check real gaming temps.

Well, it's a general rule I use. Temps around 90 are too hot imo because on the long term it can be damaging for thd cpu. That's why my general rule is <80

[Dedayog]

8 minutes ago, LukeSavenije said:

Well, it's a general rule I use. Temps around 90 are too hot imo because on the long term it can be damaging for thd cpu. That's why my general rule is <80

I get what you mean, I was curious where the 80c was coming from.  If it's a scientific CPU damaging temp or just your guideline.  Now you're saying 90c is damaging.

 

I'll do some checking as well, I am unsure about that since the CPU has a throttling feature that should kick in before damage, which is 100c or is it 105c?

 

[LukeSavenije]

4 minutes ago, jstudrawa said:

I get what you mean, I was curious where the 80c was coming from.  If it's a scientific CPU damaging temp or just your guideline.  Now you're saying 90c is damaging.

 

I'll do some checking as well, I am unsure about that since the CPU has a throttling feature that should kick in before damage, which is 100c or is it 105c?

 

It's 105 on the core and 73 on the dye if I remeber right. But that's a limit, not a recommended situation. It's kinda my guidelines, it's not a scientific theoratical guideline, just mine

[Mira Yurizaki]

Heat will put wear and tear on the part regardless of how hot the part is. I've heard this thrown around in electrical engineering circles but the rule of thumb is every 10C hotter reduces the expected life of the component by half. The problem is that there's no real way to determine life expectancy. You can infer it's supposed to last at least as long as the warranty period with the stock cooler, but there are 10 year old processors that have been in regular use. There is a "too hot" temperature though, as semiconductors will exhibit thermal runaway once beyond that point.

 

For the purposes of this topic though, you can say a CPU is definitively too hot when:

• Typically if the core temperature exceeds the specified maximum Tjunction. Unfortunately for this processor, Intel only lists a Tcase temperature of 81.3C. Tcase is always lower than Tjunction, and the only thing I could find that suggests how the two are related is at https://malwaretips.com/threads/intel-cpu-temperatures.66242/ (they say maximum Tjunction is about 5C higher than maximum Tcase)
• The CPU cannot maintain base clocks during a nominal workload. If the CPU is not turbo-boosting but maintaining base clocks, then it's technically fine. It's when it drops below base clocks that's a problem, because now it's thermal throttling.

Edited December 3, 2018 by M.Yurizaki

[LukeSavenije]

6 minutes ago, M.Yurizaki said:

Heat will put wear and tear on the part regardless of how hot the part is. I've heard this thrown around in electrical engineering circles but the rule of thumb is every 10C hotter reduces the expected life of the component by half. The problem is that there's no real way to determine life expectancy. You can infer it's supposed to last at least as long as the warranty period with the stock cooler, but there are 10 year old processors that have been in regular use. There is a "too hot" temperature though, as semiconductors will exhibit thermal runaway once beyond that point.

 

For the purposes of this topic though, you can say a CPU is definitively too hot when:

• Typically if the core temperature exceeds the specified maximum Tjunction. Unfortunately for this processor, Intel only lists a Tcase temperature of 81.3C. Tcase is always lower than Tjunction, and the only thing I could find that suggests how the two are related is at https://malwaretips.com/threads/intel-cpu-temperatures.66242/ (they say maximum Tjunction is about 5C higher than maximum Tcase)
• The CPU cannot maintain base clocks during a nominal workload. If the CPU is not turbo-boosting but maintaining base clocks, then it's technically fine. It's when it drops below base clocks that's a problem, because now it's thermal throttling.

well. I wasn't that far off with 80 then... Are you a cooling nerd?

[Mira Yurizaki]

4 minutes ago, LukeSavenije said:

well. I wasn't that far off with 80 then... Are you a cooling nerd?

I just like learning how these things work and what the practical implications are.

[Oliver davies]

2 hours ago, NelizMastr said:

Delidding a soldered CPU like this isn't that great of an idea, though. 

 

So is this at stock settings or overclocked? For stock this is very high to a degree where I also suspect poor contact. You did peel off the plastic protector film on the cooler, right?  

overclocked to 4.5Ghz and i cant drop it because windows gets fucked when i do (something to do with base clock issues i thing)

[Oliver davies]

30 minutes ago, M.Yurizaki said:

Heat will put wear and tear on the part regardless of how hot the part is. I've heard this thrown around in electrical engineering circles but the rule of thumb is every 10C hotter reduces the expected life of the component by half. The problem is that there's no real way to determine life expectancy. You can infer it's supposed to last at least as long as the warranty period with the stock cooler, but there are 10 year old processors that have been in regular use. There is a "too hot" temperature though, as semiconductors will exhibit thermal runaway once beyond that point.

 

For the purposes of this topic though, you can say a CPU is definitively too hot when:

• Typically if the core temperature exceeds the specified maximum Tjunction. Unfortunately for this processor, Intel only lists a Tcase temperature of 81.3C. Tcase is always lower than Tjunction, and the only thing I could find that suggests how the two are related is at https://malwaretips.com/threads/intel-cpu-temperatures.66242/ (they say maximum Tjunction is about 5C higher than maximum Tcase)
• The CPU cannot maintain base clocks during a nominal workload. If the CPU is not turbo-boosting but maintaining base clocks, then it's technically fine. It's when it drops below base clocks that's a problem, because now it's thermal throttling.

thanks and as it doesnt thermal throttle  so ill asumme for only short periods of time when it peaks its ok.

[Oliver davies]

2 hours ago, qqqqqq said:

You can delid X58 chips just as you can any other Intel CPU (mostly), it will yeild great results and may even result in not needing to upgrade your cooler.

looks like ill consider that.. its only a cheap chip

 

[Dedayog]

58 minutes ago, LukeSavenije said:

well. I wasn't that far off with 80 then... Are you a cooling nerd?

Per what I read from this thread, the CPU controls its thermals well, by throttling.  So 100c+ is damage, not 80c.

[LukeSavenije]

Just now, jstudrawa said:

Per what I read from this thread, the CPU controls its thermals well, by throttling.  So 100c+ is damage, not 80c.

yep. that's why 80 is my max. I don't say 80 is bad. I say everything above it is bad (not talking about a degree or two)

[Dedayog]

Just now, LukeSavenije said:

yep. that's why 80 is my max. I don't say 80 is bad. I say everything above it is bad (not talking about a degree or two)

What I'm getting at is being careful stating opinions as fact... 80c and 90c do necessarily adversely damage a CPU under a stress test or gaming.  it is well within their normal operating limits.