Question Do AM5 chips run hot or what ?

[krzyimprt]

Am I crazy to think my CPU hit 95c? idling 37c, gaming 40-65c, rendering 70-80c. prime95 96c.
Peerless assassin 120.

https://imgur.com/2KjSYSK

 

[drivinfast247]

Remember, engineering is about designing a product or solution to meet specifications. If those specifications call for a disposable product designed to last 7-10 years, that is what you get. If they were designing a chip to last 25-50 years they would do things very differently.

Wouldn't matter in the gaming PC market as CPUs are only viable for less than 10 years. Just look at game requirements in the last 5-6 years. I would bet the majority of consumers buying the high end and flagship CPUs are going to replace in a generation or two anyways.

 

[Eximo]

Wouldn't matter in the gaming PC market as CPUs are only viable for less than 10 years. Just look at game requirements in the last 5-6 years. I would bet the majority of consumers buying the high end and flagship CPUs are going to replace in a generation or two anyways.

Yes, that is exactly my point. AMD can afford to have their CPUs run hot knowing that PC hardware rarely gets used very long. As long as they have the efficiency advantage over Intel they have nothing to lose but having to warranty a percentage of chips, assuming people bother.

AMD has increased risk since there are more points of failure in the design. Chiplets on a substrate means you also have to worry about the interconnects between the chips and thermal expansion rates being different between chiplets as they are used differently and the soldered TIM to the heatspreader. But these are things they have been dealing with for a while now.

Common AMD chip failures I am aware of is the I/O die losing memory channels, that seems to be prevalent amongst the Ryzen 5000 series. Complete CPU failure is more common with 2000 and 3000 series chips, which at this point are roughly half way through their useful life.

Intel will be in the same boat next gen with multiple chips on a substrate. But it already looks like their chips are not good performers, so I suspect the desktop chips will be overclocked to ridiculous power levels again.

 

[TerryLaze]

Wouldn't matter in the gaming PC market as CPUs are only viable for less than 10 years. Just look at game requirements in the last 5-6 years. I would bet the majority of consumers buying the high end and flagship CPUs are going to replace in a generation or two anyways.

Yeah, that doesn't mean that those people just throw their "old" systems out the window, they either sell them or keep them for secondary PCs NAS or whatever.
Either way having them last longer is important.

But also:

Higher temperatures mean greater thermal expansion for one. Two, you have increased internal resistance, the higher the temperature the more power you have to push through, so less efficient the warmer it is. Three, chemical and electrochemical reactions work faster at higher temperatures. So anything going on inside the chip at the microscopic level is accelerated. That would be things like electromigration or crystallization effects.

Yeah, we are not talking about hundreds of degrees of difference here, between room temperature and about 100 degrees there will be so little effect of what you say that you can completely disregard it.

Yes, that is exactly my point. AMD can afford to have their CPUs run hot knowing that PC hardware rarely gets used very long. As long as they have the efficiency advantage over Intel they have nothing to lose but having to warranty a percentage of chips, assuming people bother.

Reputation is extremely important and being known as the one with CPUs that don't last very long will result in people avoiding them.
95 is safe for long term even if it's not optimal, it will not kill or degrade CPUs.
You have seen how fast AMD reacts to actually bad settings that can really kill CPUs, they released one agesa after the after to try and contain the exploding CPU debacle.

 

[Eximo]

I wouldn't say it can be disregarded, it is just a design consideration. You have to account for such stresses. Cold start to stress test on one CCD I'm sure is a test they performed to see how the solder and heatspreader reacted. And they would have cycled that hundreds of times to see how it would do long term. Once an acceptable level was determined, they go with that.

They will have a certain expectation of customer returns. Their goal is to keep it to an acceptable level.

PR disasters like the burning CPUs is well outside the norm.

 

[TerryLaze]

I wouldn't say it can be disregarded,

I didn't say that it can be disregarded, I said it can be disregarded by you, by all of us that are just simple users.

 

[krzyimprt]

Late reply! Got extremely busy with life. I replaced all my case fans with thermalright PWM fans. I settled on PBO 80c lvl 4. Depending on the situation all cores boost to 5.4ghz. Temps are looking MUCH better. I also went through a few bios updates which may have helped the situation. Running P95 small FFT temps were around 60-70c. Room temp is 20c. https://imgur.com/a/Z7IcPAP