• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

AMD Ryzen 9 7950X Cooling Requirements & Thermal Throttling

AMD target timeframe is the warranty period, which is 24 months in the worst case (EU).
They are sure the CPU won’t break in 24 months running at those max temperatures, and that’s enough. And here we are speaking about silicon degradation, not a complete failure, which is even more difficult to assess. A CPU requiring higher voltage to keep the same performance is technically still in working conditions, so warranty wise is still ok.
Customers target timeframe is another story. Here above we have people looking for 10+ years. Even users like me, upgrading every 2/3 years , are expecting to sell/give the CPU in working conditions to someone else.


I would not accept such high temperatures in your case. I would set lower power limits, or stick to ECO Mode.
ECO mode is too low of a power limit and hurts performance more than 10% for MT apps since it goes from 230W to 145W. One can just use the board temp limit option that forces less voltage and a lower frequency when needed.
 
ECO mode is too low of a power limit and hurts performance more than 10% for MT apps since it goes from 230W to 145W. One can just use the board temp limit option that forces less voltage and a lower frequency when needed.
Not really, it depends on the application. But mostly you should be around 10~15% again generally as per the workload.
Cinebench R23 nT power consumption 7950X review


That's a better way for sure, you could argue however that ECO mode is for users who don't like to tinker too much or aren't used to AMD platform's (too many?) options for lowering power/thermals :toast:
 
Last edited:
It's a lot longer than that, 3 years in most parts of the world. Unless you think they're specifically shipping only lower quality stuff to Europe & maybe even Oz & NZ :wtf:
It's a lot longer than that, 3 years in most parts of the world. Unless you think they're specifically shipping only lower quality stuff to Europe & maybe even Oz & NZ :wtf:
No, no your are right. It’s 3 years for AMD PIB processors

ECO mode is too low of a power limit and hurts performance more than 10% for MT apps since it goes from 230W to 145W. One can just use the board temp limit option that forces less voltage and a lower frequency when needed.
It depends on the application you are using, but you can still find a solution in between ECO mode and default limit, or as you said just use a different temp limit.
 
As we get smaller and smaller chips with greater and greater density thermal throttling is going to be a fact of life

there comes a point in the density of ANY IC where you simply can not get the heat out of the DIE fast enough, there just isn't enough surface area to conduct enough heat quickly enough, the size of the cooler just stops mattering

we are getting pretty close to that point now

also as we get smaller as a general rule the critical operating temp increases

AMD says ZEN 4 is good to 115C so thats still 20C before you need to be concerned

now i think thats pretty warm but again density and smaller newer process changes the rules a bit
 
In fact, as a guy who has promoted the use of Older /500 watt PC's for heater's here why use 500/1000 watts just to heat when you can get computation effectively at the same time.

As someone who has actually ran hardware for consumers at its limits day in day out continuously from both teams as an unbiased engineer I have experience that "this is fine".

You haven't used said equation to prove your right, just mentioning it instead, like no one else is aware, we are.
So are AMD, they're scientific based testing did too, so again put up some evidence, do the Math.

And your pulling the fanboi argument out your pocket reaks of desperation, I suppose you too could just keep repeating the same shit while trying to undermine others with bias tags but know this, it reflects badly on you're own bias's IMHO.

Oh and as someone who has degraded CPUs with they're activity, I am actually ok with CPU designed to provide Max performance within a useful lifespan, Moore's law dictated shit before but got vague in the quad core erra, but given modern performance increases we should be back to about three years useful use, with conditional restricted use beyond that relative to the latest tech IE shit should be surpassed in every metric within 3 years anyway.

Usually migration took 3 years to begin to develop though obviously with effort you can do wonderful things, not.
To be fair, I'm not OK with pushing a CPU to its limits for 3 years only. I'm not OK for degradation due to unnecessary power consumption or heat or any other limits of the silicon being breached. But I'm OK with going into the BIOS and clicking "Enable" on Eco mode, just as I do with XMP.

And that's my ultimate point in the matter: if you're OK with 95 °C, fine. If you're not, just enable the damn Eco mode.

Temps won't go up, because of the design, but the clock speeds might go down. By how much? Personally my unfounded speculation is that after say three years or so (handy given the warranty!) maybe you get a few hundred MHz less, which would be fine to me. BTW, is there a guaranteed clock speed and power draw on these things? I assume there is not, because that would be nuts.
As far as I'm concerned, the guaranteed clock speed for each chip is the base clock, around 4.5 GHz for Zen 4.

ECO mode is too low of a power limit and hurts performance more than 10% for MT apps since it goes from 230W to 145W. One can just use the board temp limit option that forces less voltage and a lower frequency when needed.
A 10% performance loss in heavily threaded workloads and a much lower loss in light work and games for 30-40% less power and a lot less heat on such powerful CPUs is a good trade, imo.

Record setting in Cinebench is for kids.
 
It will be very intresting to see how zen4 will compere to RL on that 'default 95 degree' matter because AMD approch is now quite different from intel.

I think that the whole debate - pros vs cons of AMD decision - will be much more clear especially when you have equality with threads and wattage (to the point of +/- 20W).

A wonderful week ahead!
 
For 5+ years period, I would be more worried about finding motherboard VRMs and capacitors that can adapt to 1ms changes at that point.
 
please check your facts: it is literally what AMD said in response to the initial concerns about high temperatures.


It is not safe to run 95° on the long run, no matter what AMD marketing is saying.
And NO, using a 12700K with power limit unlocked on a daily basis I don’t go north of 70° with a regular 360mm AIO, so that’s not how “modern Intel CPUs behave”.
By the way this “Intel too” narrative is childish: we are speaking about AMD HERE, where a low end 6 core CPU is expected, according to their wording, to reach 95°. Which is totally ridiculous.
I don’t care if a 12900K is reaching the same temperature. I criticized Intel for the same reason
So when it comes to AMD engineers and you, clearly you're the expert.

And of course - the temperature setup in YOUR system is cold therefore all 12th gen CPU's are cold - but if AMD says the CPU's are fine at 95C, that means they all run at 95C all the time.



Heres something for you: practise this without making any assumptions. Only use facts repeated from multiple sources - not multiple people quoting or misquoting the same source, but multiple sources.
Because removing your assumptions and guesses, you really don't say anything

A few years ago 65C would kill a CPU
A GPU at 60C would be dying

Don't make stupid claims about a customised setup as if it's the out of the box experience, and compare it to another brands out of the box experience - especially when they're not competing products.
 
So when it comes to AMD MARKETING and you, clearly you're the expert.
FIXED for you

On a side note, as a Moderator you should not use wording like “stupid claims”.
 
Last edited:
FIXED for you
Why would AMD's marketing say anything without backing it up by data measured by their engineers? Do you think AMD would risk a class action lawsuit based on an entire generation of prematurely failing CPUs? Or do you think "AMD Marketing" and "AMD Engineering" are two separate companies?

I still don't understand why turning on the damned Eco mode is so hard for you.
 
Why would AMD's marketing say anything without backing it up by data measured by their engineers? Do you think AMD would risk a class action lawsuit based on an entire generation of prematurely failing CPUs? Or do you think "AMD Marketing" and "AMD Engineering" are two separate companies?
As I explained several time above, AMD doesn’t risk anything since they are just thinking about warranty timeframe (3 years) and since silicon degradation is really not something you can guarantee over the time. The only parameter they are assuring is base clock.

I still don't understand why turning on the damned Eco mode is so hard for you.
It is not.

By the way we already have been through “AMD engineering” BS in the past:

 
As I explained several time above, AMD doesn’t risk anything since they are just thinking about warranty timeframe (3 years) and since silicon degradation is really not something you can guarantee over the time. The only parameter they are assuring is base clock.
Yes they do. This could risk law suit from EU, under planned obsolescence, which is illegal.
Apple and Samsung have got hit by fines regarding down-tuning their phones (for alleged battery performance), AMD could risk the same by having "planned degradation"

and no, saying boost up till, and failing that is also a 9law suit risk under false marketing/advertisment
 
The only parameter they are assuring is base clock.
Which has been the case on both AMD and Intel since boost clocks were invented.

By the way we already have been through “AMD engineering” BS in the past:

This is one of GN's pointless "hot air complaint yadda yadda" videos, imo. Steve is basically talking about AMD's claim that 110 °C junction temp is OK on the 5700 XT even though 99% of the cards never reach it.

This video explains the situation much better:
 
As I explained several time above, AMD doesn’t risk anything since they are just thinking about warranty timeframe (3 years) and since silicon degradation is really not something you can guarantee over the time. The only parameter they are assuring is base clock.


It is not.

By the way we already have been through “AMD engineering” BS in the past:

Everyone gets it, you have a far superior thought process to TSMC, AMD, and need to let us the community know about it so we can be as fanatical as you.


They are obviously trying to pull a fast one by releasing a poorly designed HALO product that requires too much thought for the average peon to use as only you correctly know how. They must be stopped by repeating the same mantra to others until they chant it back!!!


No one here knows how dangerous TSMC and AMD are to hide this, assassins are being dispatched currently to silence the opposition.
 
Everyone gets it, you have a far superior thought process to TSMC, AMD, and need to let us the community know about it so we can be as fanatical as you.


They are obviously trying to pull a fast one by releasing a poorly designed HALO product that requires too much thought for the average peon to use as only you correctly know how. They must be stopped by repeating the same mantra to others until they chant it back!!!


No one here knows how dangerous TSMC and AMD are to hide this, assassins are being dispatched currently to silence the opposition.
Do you really think you’re funny ?

Which has been the case on both AMD and Intel since boost clocks were invented.


This is one of GN's pointless "hot air complaint yadda yadda" videos, imo. Steve is basically talking about AMD's claim that 110 °C junction temp is OK on the 5700 XT even though 99% of the cards never reach it.

This video explains the situation much better:
Are you really taking a Jayz’s video to explain something Steve said ?
Seriously….
One is competent, the other is a nice guy… guess who is the competent one.
 
Are you really taking a Jayz’s video to explain something Steve said ?
Seriously….
I watched both videos. Did you?

Steve's video is just a rant about an allegedly OK temperature range that no 5700 XT ever reached under normal conditions while Jay actually heats up a card with a heat gun pointed at it to show how effective modern safety measures are.
 
FIXED for you

On a side note, as a Moderator you should not use wording like “stupid claims”.
When backed by zero proof as your claims are, it's warranted.

As are AMD'S specified running conditions for zen 4, backed by Tests.

Whereas we have your opinion backed by little more than a der8haur video about not running it stock in any way shape or form.
 
I watched both videos. Did you?

Steve's video is just a rant about an allegedly OK temperature range that no 5700 XT ever reached under normal conditions while Jay actually heats up a card with a heat gun pointed at it to show how effective modern safety measures are.
I should stop speaking with people considering an heat gun similar to the heat generated by CURRENT.
It is quite clear to me that very few of you really understand what electromigration is and how it affects components.
 
I should stop speaking with people considering an heat gun similar to the heat generated by CURRENT.
It is quite clear to me that very few of you really understand what electromigration is and how it affects components.
You make it sound like black magic it's a simple equation, one you still haven't used to prove your right.
And now you think others aren't aware.
Well that's nice and all but it's well over 70 years old and if you can't show AMD is wrong with it you shouldn't be spouting about knowing it proves AMD wrong.

Science is NEVER beaten by opinion, only real science and math can do that and your not capable clearly.
 
I should stop speaking with people considering an heat gun similar to the heat generated by CURRENT.
It is quite clear to me that very few of you really understand what electromigration is and how it affects components.
It is quite clear to me that you don't understand the (importance of) built-in safety measures that prevent you from destroying your CPU or GPU due to any kind of heat.
 
It is quite clear to me that you don't understand the (importance of) built-in safety measures that prevent you from destroying your CPU or GPU due to any kind of heat.
I NEVER spoke about “destruction”. Those measures are in place to avoid that.
I’m speaking about degradation, which is the reason why I won’t run a CPU at 95° for an extended period of time. And degradation is caused by heat AND current combined, not something you can show with a silly test with an heat gun (that is causing a totally different damage to the board but is very cool as a clickbait).
 
Look, it's clear Max has made up their mind here. Regardless of who's right, can we push the "agree to disagree" button and move on?
 
I NEVER spoke about “destruction”. Those measures are in place to avoid that.
I’m speaking about degradation, which is the reason why I won’t run a CPU at 95° for an extended period of time. And degradation is caused by heat AND current combined, not something you can show with a silly test with an heat gun (that is causing a totally different damage to the board but is very cool as a clickbait).
An overclocked 2080 Super running Heaven Benchmark has absolutely no current flowing through it. I get ya. :pimp:

Who's asking you to run your CPU at 95 °C?
 
An overclocked 2080 Super running Heaven Benchmark has absolutely no current flowing through it. I get ya. :pimp:
The heat is added FROM OUTSIDE with an heat gun, not generated by the current.
You are comparing apples to pears here…

Who's asking you to run your CPU at 95 °C?
AMD
 
The heat is added FROM OUTSIDE with an heat gun, not generated by the current.
You are comparing apples to pears here…
Are you seriously suggesting that Tjmax doesn't apply when part of the heat is coming from an outside source while the component is under full load?

"95 °C is normal and expected behaviour" was not the same as "you have to run your CPU at 95 °C at all costs" the last time I checked.
 
Back
Top