AMD Ryzen 9 7950X Cooling Requirements & Thermal Throttling

[ARF]

No. AMD boosts up until you reach any limit (power, thermal or max boost).

Setting your own max boost achieves nothing on modern AMD CPUs. You're basically restricting yourself to one performance limiter when you could let the chip run faster to hit another limit that it's designed to hit. You could set it up to boost to X MHz with 1 core, or Y MHz all-core, but why would you if it can boost higher in lighter workloads by default?

The max boost should have no limit. I guess it is tied to the max voltage. But I guess it should have an algorithm for automatic determination of the most optimal ratio for the particular die and its quality

Socket AM5 must throw away any compatibility with am4 size for prevent this situation, with more bigger socket them can prevent this temperature situation

However said somethink like 95 degrees are perfectly safe and normal that dont sound good, maybe seems a cheap trick to try sell a product with temperature problems

Personally seems something like that: you need buy new expensive cpu, new expensive mainboard, new expensive memories aka higher frecuencies memories but you dont need buy new cpu cooler

If buy new expensive components, buy new cooler dont be a problem because users buy am5 must be have enough money and new cooler dont change so much in money

Again personally i think ryzen 5 7600x dont must be exist (same than size dimensions of socket am5) because give a strange message in prices side

Because with all things showed this platform are high end and dont be avalaible for averaje joe users, for them amd still have am4

I wouldn't say that AMD is stupid but that its engineers didn't make the best decision.

They needed new coolers for AM5 designed exclusively for high-TDP CPUs, because obviously this generation is the first from AMD to behave so badly.
And second - the IHS is indeed horrible - it introduced an additional thick layer of thermal resistance and naked dies is the solution to go - for such a premium and super expensive platform it would not be a problem - after all the people who assemble these configurations know what they do and have mastered this practice

 

[big_glasses]

Your making an assumption that after the warranty period the CPU will degrade. No evidence of that whatsoever.

But since you can't predict how these products will be used in each & very system out there, alongside cooling, & of course if the user OC it himself ~ so planning to make it obsolete by AMD themselves outside warranty would be akin to putting your two feet & both arms on a chainsaw! Why would they even do that, they got burned with FX "8 core" lawsuit do you really they'd try an even more brain dead move?

me? I was initially responding to someone earlier in the thread saying that running at 95c is bad, AMD is wrong in saying it's planned, and AMD doesn't care about degradation (implied majorly outside the norm) after warranty period.
My whole point has been that planning for failure isn't legal (in EU), AND that warranty isn't some protection against widespread failure (or massive degradation. And that it would be a a risky, braindead move to "over-run" as they are still at risk even after warranty

Like this statement from earlier in the thread:

It is not safe to run 95° on the long run, no matter what AMD marketing is saying.

AMD just care about warranty period. If the cpu degrade after that, they really don’t care.

 

[R0H1T]

Ok, I haven't followed all 14 pages closely but the bottom line is ~ 95C is (well) within tolerance, you can run it safely 24x7 without fear of melting your precious! Personally I won't run it at those temps but AMD has designed it in such a way that it may be unavoidable for the less informed.

 

[HenrySomeone]

Socket AM5 must throw away any compatibility with am4 size for prevent this situation, with more bigger socket them can prevent this temperature situation

However said somethink like 95 degrees are perfectly safe and normal that dont sound good, maybe seems a cheap trick to try sell a product with temperature problems

Personally seems something like that: you need buy new expensive cpu, new expensive mainboard, new expensive memories aka higher frecuencies memories but you dont need buy new cpu cooler

If buy new expensive components, buy new cooler dont be a problem because users buy am5 must be have enough money and new cooler dont change so much in money

Again personally i think ryzen 5 7600x dont must be exist (same than size dimensions of socket am5) because give a strange message in prices side

Because with all things showed this platform are high end and dont be avalaible for averaje joe users, for them amd still have am4

I have to agree; with how much AM5 costs as a whole, it was absolutely stupid to try to retain cooler compatibility and sacrifice well over 10 degrees (as DerBauer's testing clearly shows) due to the ridiculously thick IHS. Just another typical AMD blunder I guess...

 

[A Computer Guy]

Ok, I haven't followed all 14 pages closely but the bottom line is ~ 95C is (well) within tolerance, you can run it safely 24x7 without fear of melting your precious! Personally I won't run it at those temps but AMD has designed it in such a way that it may be unavoidable for the less informed.

That's a nice summary.

 

[Mussels]

During the early 3000-series, was many who claimed their CPU's couldn't reach the advertised boost.

But that's got nothing to do with what's being discussed at all?
The fact that single core low threaded boost wasn't as high as people wanted, has NOTHING to do with CPU degradation.

 

[A Computer Guy]

But that's got nothing to do with what's being discussed at all?
The fact that single core low threaded boost wasn't as high as people wanted, has NOTHING to do with CPU degradation.

Also if memory serves correctly that had a lot to do platform maturity (much needed BIOS/UEFI and driver updates) but also certain applications caused difficulties reaching boost clocks.

me? I was initially responding to someone earlier in the thread saying that running at 95c is bad, AMD is wrong in saying it's planned, and AMD doesn't care about degradation (implied majorly outside the norm) after warranty period.
My whole point has been that planning for failure isn't legal (in EU), AND that warranty isn't some protection against widespread failure (or massive degradation. And that it would be a a risky, braindead move to "over-run" as they are still at risk even after warranty...

Just a question...How does EU cope with SSD's and what makes that different than CPU's if high temps are considered normal wear and tear?

 

[cadaveca]

All these worries about temperature, yet laptops with the same type of chips run these temps for most of their life and last years and years. I still have a laptop from 2008 that works just as it did the day I bought it. that's 14+ years of running these sorts of temperature, with far less advanced technology, and barely any tech for the CPU protecting itself from damage.

 

[80-watt Hamster]

All these worries about temperature, yet laptops with the same type of chips run these temps for most of their life and last years and years. I still have a laptop from 2008 that works just as it did the day I bought it. that's 14+ years of running these sorts of temperature, with far less advanced technology, and barely any tech for the CPU protecting itself from damage.

One of the thoughts is that the denser nodes may be less robust than the old 14nm+. Z4 is brand new, so there isn't (and can't be) any user-level data on years-long service lives under any conditions. All we can do is wait, speculate and/or complain. Personally, I choose and encourage others to forgo the latter until there's something proven to complain about.

 

[Count von Schwalbe]

here isn't (and can't be) any user-level data on years-long service lives under any conditions.

Apple's M1 is on the same node, but the power envelope wasn't pushed so hard.

 

[HenrySomeone]

All these worries about temperature, yet laptops with the same type of chips run these temps for most of their life and last years and years. I still have a laptop from 2008 that works just as it did the day I bought it. that's 14+ years of running these sorts of temperature, with far less advanced technology, and barely any tech for the CPU protecting itself from damage.

Funny that you should mention a laptop from 2008, because I also happen to have one from that exact year that also still works (HP with a 2Ghz Core 2 Duo), however the temps never exceed 70 degrees!

Apple's M1 is on the same node, but the power envelope wasn't pushed so hard.

This can make all the difference in the world as far as longevity is concerned.

 

[big_glasses]

But that's got nothing to do with what's being discussed at all?
The fact that single core low threaded boost wasn't as high as people wanted, has NOTHING to do with CPU degradation.

False advertisement on advertised boost. I believe I was stating my surprise no one had tried to take it to EU court, in a side note. And thought your "up to" statement(/point 1) was on that small side note.

Just a question...How does EU cope with SSD's and what makes that different than CPU's if high temps are considered normal wear and tear?

No clue, my whole point has been to state that planned obsolescence(and false advertisement), can (keyword) make you suable in/by EU.
Are you referring to the normal wear and tear of SSDs? SSD's usually have rw amount as part of their specs, ye?
I'm no law expert, I've just googled cases, and know some general guidelines (planned obsolescence and false advertisement).

 

[TheoneandonlyMrK]

Funny that you should mention a laptop from 2008, because I also happen to have one from that exact year that also still works (HP with a 2Ghz Core 2 Duo), however the temps never exceed 70 degrees!


This can make all the difference in the world as far as longevity is concerned.

Your comparing a core 2 2 core at 70°c to a 16/32 thread CPU doing 95, the vitreal I have for this statement is beyond words, I can't put it into words it's tooooo offensive.

Buy a 13900K let's see you and others explain away 100°C

My laptop an 8750H regularly sits at 90/95 while as ever, Actually doing something.

@Max(IT) your Done son, your upgrade days are over, you just got schooled by reality, go buy that 13900k while shit posting about Black's law hahaaaaaa.

 

[marios15]

My old Athlon 64 X2 4200+ overclocked to 2.7-2.8Ghz was running up to 120C without issues But the package temp remained below 70
Back then the package temp was just the socket temperature and very few CPUs had actual core temperatures reported

 

[OkieDan]

If only AMD could make the PCB thicker so the IHS could be thinner. Not sure how feasible it would be but doesn't sound totally stupid.

 

[CapNemo72]

Thanks for a great article.

However, this does not really answer the question for majority:

We know it will be water cooling, but 240 or 360?

I will not go air cooler (even though that is what I have on my TR1950x) and do not want big case just to be able to install that 420 monster.

I am absolutely torn between 240 or 360 AIO.

 

[bl4C3y3]

Hi, thanks a lot for this deep dive !
This article is what made me consider buying this new generation, because the technology and behavior is interesting
For testing, I'm now running a 7950X on an MSI B650 motherboard with my old Wraith Prism cooler (from a 3900X) and it's nicely rendering with Cineware (Cinema 4D) at a constant 95° and hovering around 5000MHz CCD1 - 4800MHz CCD2 at 100% usage (while typing this )

We know it will be water cooling, but 240 or 360?

I just did a quick render comparison between my current system "5900X with EKWB block + S360 radiator" and my new system "7950X + Wraith spire air cooler" (still under construction):

• 5900X + 360mm WC = 20min01sec (@ +/-145W)
• 7950X + low end AC = 10min28sec (@ +/-195W)

So, the 7950X is about 1.4x more efficient in power usage, and almost 2x more time efficient !

The 7950X is 'similar' to your TR1950X (in cores 16c/32t, and TDP 170/180W), and with the Wraith spire at 100% fan speed, the 7950X hovered around 5.1GHz CCD1 / 5.0GHz CCD2 at a stable 95°C during this short render test. It started with higher frequency (5.4 ?), but dropped in few seconds to 5.1.
I will move the EKWB block + S360 radiator to the 7950X, and will do the test again, to see what the difference is ... but will only get to it in a couple of weeks.
But now I realize that I also have a EKWB + 240 radiator (from my old 3900X system), which I 'unmounted' without draining, so I could put that on the 7950X first , maybe next week, so I can compare low end air to 240mm (aluminum, so similar to most AIO) to 360mm (copper).
W1zzard also has the Wraith spire as data point, so this could be an interesting addition ... although when looking at W1zzard results, I don't think that there wil be a big difference between 240 and 360 ?

 

[bl4C3y3]

When doing this same image reder test in Cinema 4D with the 'alu EKWB + 240 radiator', and comparing it to the Wraith spire, then I would say the biggest factor is noise
So, I can imagine that a 360 radiator could offer slightly better performance at lower noise ...

Nothing special in the results, they are exactly as in the review, you seem to lose little performance by toning down the cooling, which is great:

• Wraith spire @ 100% fan speed > 10:45 (5.0-5.1/4.9-5.0, 195W, 95°C)
• Alu 240 rad @ 100% fan speed > 10:25 (5.1-5.2/5.0-5.1, 210W, 95°C)
• Alu 240 rad @ 50% fan speed > 10:35 (5.1-5.2/5.0-5.1, 200W, 95°C)
• Alu 240 rad @ 25% fan speed > 10:55 (5.0-5.1/4.9-5.0, 190W, 95°C)

With the Alu 240 rad water cooling, you gain 20sec (about 5%) by going from 25% fan to 50%, and 10sec by going from 50% to 100% fan speed.
The difference in GHz and time are not so dramatic, but the difference in noise is quite big ...
I'll test with a copper 360 radiator later ... and if i can make a prediction, I think it will do 10:20 at 50% fan speed

 

[CapNemo72]

When doing this same image reder test in Cinema 4D with the 'alu EKWB + 240 radiator', and comparing it to the Wraith spire, then I would say the biggest factor is noise
So, I can imagine that a 360 radiator could offer slightly better performance at lower noise ...

Nothing special in the results, they are exactly as in the review, you seem to lose little performance by toning down the cooling, which is great:

• Wraith spire @ 100% fan speed > 10:45 (5.0-5.1/4.9-5.0, 195W, 95°C)
• Alu 240 rad @ 100% fan speed > 10:25 (5.1-5.2/5.0-5.1, 210W, 95°C)
• Alu 240 rad @ 50% fan speed > 10:35 (5.1-5.2/5.0-5.1, 200W, 95°C)
• Alu 240 rad @ 25% fan speed > 10:55 (5.0-5.1/4.9-5.0, 190W, 95°C)

With the Alu 240 rad water cooling, you gain 20sec (about 5%) by going from 25% fan to 50%, and 10sec by going from 50% to 100% fan speed.
The difference in GHz and time are not so dramatic, but the difference in noise is quite big ...
I'll test with a copper 360 radiator later ... and if i can make a prediction, I think it will do 10:20 at 50% fan speed

Thank you. This is so important.

In the meantime, I got Thermaltake TH240 Snow edition. I prefer quiet systems and it is OK if I lose some performance.
Having a lot to finish (job projects), I still did not had a time to start my build.
My last element just arrived (Radeon 7900 XTX), but not sure if I want to keep it. The shop was playing "scalper" mode on their price.

 

[bl4C3y3]

Thermaltake TH240 Snow edition

Yes, this should be sufficient. Same here about noise ... for some testing, I don't mind it being loud, but for general usage, I also like it quiet.
I just finished my project, moving the EK classic S360 set from X570+5900X to B650+7950X (and the EK FG A240R set from X370+3900X to X570+5900X), and after some quick test, i don't really see any difference yet in time/frequencies/power/temperature/noise of the 360 vs the 240.
I have a Vega64 in the 240 loop, and a 6900XT in the 360 loop, but for these rendering tests, they were more or less idle I guess.

About the 6900XT, I also didn't want to support scalper prices, so I checked the weekly drops on amd.com shop, and it took 7 months before I was able to get an 6900XT.
Just out of curiosity for the 7900XTX, I checked the amd.com shop during the launch time and i was able to get one in my cart for MSRP (but didn't pull the trigger) ... I see this same scalper mode for the 7900XT: it's still available at MSRP on amd.com shop, while some online retailers ask +30% for the same reference card ...

 

[alceryes]

I've got the Dark Rock TF 2 on my 7950X. With default PBO (B650E Aorus Master) the CPU will march right up to 95ºC in all-core workloads/benchmarking but still gives me a burst Cinebench R23 score of 38686.

Once I set a 125W PPT limit it stays cool and quiet. According to 3DMark Time Spy I'm only losing 3-4% peak (not all the time) gaming performance vs no PPT limit. I play at 4K anyway so my 6900 XT will always be the limiting factor.