AMD Ryzen 9 7950X Cooling Requirements & Thermal Throttling

[thestryker6]

I disagree with you still, surprise.

And I have to ask are you going to work on Hedt or just game?!.

With either supplier your future looks bleak I'd say, newer chip's are not going to be cooler.

And when Intel kicks in with chiplets and stacks your ideology is going to get wrecked IMHO.

What do you disagree with? AMD literally made a worse product due to maintaining cooler compatibility than if they didn't. The efficiency was also higher with direct die on top of the temperature going down.

I do both gaming and work, but AMD priced HEDT way too high for my needs (PCIe/memory bandwidth, I don't need cores) so I'm keeping an open mind that Intel may decide to put out some cheaper lower core count models.

I too doubt it's going to get better, but I don't need something immediately so Intel having high load power consumption and AMD making their own efficiency worse (and having pretty high power consumption) makes it easy to wait and see.

 

[AusWolf]

All this yadda yadda about temperatures, and I'm still on the side of "just slap your big-ass cooler that you already have on it, and enable Eco mode". Problem solved. Based on test data, I honestly believe that there is absolutely no reason to run these CPUs at factory default TDPs.

 

[TheoneandonlyMrK]

What do you disagree with? AMD literally made a worse product due to maintaining cooler compatibility than if they didn't. The efficiency was also higher with direct die on top of the temperature going down.

I do both gaming and work, but AMD priced HEDT way too high for my needs (PCIe/memory bandwidth, I don't need cores) so I'm keeping an open mind that Intel may decide to put out some cheaper lower core count models.

I too doubt it's going to get better, but I don't need something immediately so Intel having high load power consumption and AMD making their own efficiency worse (and having pretty high power consumption) makes it easy to wait and see.

See bold, your mentioning direct die AGAIN, is that what you think reasonable to do!?.

The IHS is there for a reason, no one's selling consumer's direct die cooling opportunity's.

AMD made a worse product!?, As an engineer, and a true computing enthusiast, I think your wrong and deluded.

That there was any other option for AMD in real terms given Intel's push on core's and lack of concern about power use.

And that somehow you as an armchair engineer know more than AMD know after many thousands of scientific testing hours , validation, qualification and endurance /stress testing.


To be specific.

And in a thread where the OP proved AMD right no less.

see post 1.

 

[thestryker6]

See bold, your mentioning direct die AGAIN, is that what you think reasonable to do!?.

The IHS is there for a reason, no one's selling consumer's direct die cooling opportunity's.

AMD made a worse product!?, As an engineer, and a true computing enthusiast, I think your wrong and deluded.

That there was any other option for AMD in real terms given Intel's push on core's and lack of concern about power use.

And that somehow you as an armchair engineer know more than AMD know after many thousands of scientific testing hours , validation, qualification and endurance /stress testing.


To be specific.

And in a thread where the OP proved AMD right no less.

see post 1.

You are literally ignoring the fact that there are significantly higher temperature deltas when delidding these processors. That means their IHS is having an outsized impact on the thermal performance. I don't know why you cannot seem to grasp it and keep repeating the same nonsense. Maybe you don't understand the point: if AMD had made a thinner IHS and tossed drop in cooler compatibility the thermal performance would have been better.

 

[AusWolf]

You are literally ignoring the fact that there are significantly higher temperature deltas when delidding these processors. That means their IHS is having an outsized impact on the thermal performance. I don't know why you cannot seem to grasp it and keep repeating the same nonsense. Maybe you don't understand the point: if AMD had made a thinner IHS and tossed drop in cooler compatibility the thermal performance would have been better.

AMD could have made a thinner IHS and a new cooler mounting system. They didn't. So what?

Why is it such a pain for some of you guys to enable Eco mode and lose maybe 2% performance?

 

[thestryker6]

AMD could have made a thinner IHS and a new cooler mounting system. They didn't. So what?

Why is it such a pain for some of you guys to enable Eco mode and lose maybe 2% performance?

They chose to make their product worse just like I'm choosing not to buy it in part because of that. What's so hard to understand about that?

 

[AusWolf]

They chose to make their product worse just like I'm choosing not to buy it in part because of that. What's so hard to understand about that?

How is it worse? By the user having to enter the BIOS to enable Eco mode?

 

[TheoneandonlyMrK]

They chose to make their product worse just like I'm choosing not to buy it in part because of that. What's so hard to understand about that?

Your perspective is yours, I don't share it.

Rant on about a IHS of a part you weren't going to buy I'm out, the reasons you give are not the be all of the situation and you think you can do better, right crack on then let's see your next rig when it happens, should be interesting.

 

[thestryker6]

How is it worse? By the user having to enter the BIOS to enable Eco mode?

It drops performance to work around a chosen design decision? That makes the product worse than if the decision hadn't been made in the first place.

 

[TheoneandonlyMrK]

It drops performance to work around a chosen design decision? That makes the product worse than if the decision hadn't been made in the first place.

Reading comprehension fail.

Double down and keep repeating.

Wtaf is this tactic, we heard your opinion, how many times do you want us to hear it.

I quote

"What you might not be noticing immediately is that with Zen 4 you will now get higher performance in return, automatically, without doing anything, that will be your benefit of upgrading—more perf, not lower temps. As we've seen in this article, if you already have very decent cooling, then the gains from going water are relatively small, and probably not worth it, especially in games. Also, infinitely good cooling will not scale clocks and performance up automatically. Rather there are some hard limits, above which the CPU will not boost, even with excellent cooling."

So baring extreme cooling methods and effort, it's not possible to get more performance than what you have anyway.

A thinner IHs would not boost higher and due to the way it's designed it would , work load allowing get to 95 anyway, because it's designed to aim for that.

And that's a better temperature than 65 because. The differential in temperature between that and your typical ambient temperature.

Actually allows more heat to be removed for the same effort.

 

[A Computer Guy]

The thermal envelope Am5 operates in is directly caused by the node not the heatsink and all you saying otherwise need to show your qualies and test data because your chatting total balls.

And clearly do not understand the design choices made were to optimise the performance in every scenario, automatically.

A lower Z hight on some KS chip's was not the heatsink, the actual chip was ground thinner, which will have meant that the heatsink on those would be thicker than standard to align height with the cooler's.

It isn't the IHS it's the power and frequency curve control circuitry causing the DESIREd high temperatures.

Delidded chips with direct die cooling are always upwards of 20° c cooler because there's two less layers for heat to pass through.

Arm chair engineering= asss brained assuming IMHO.

And I can't wait to hear some of the yeah butting, and whataboutism in the 13900K review thread.

I don't get what all the hubbub is about. If direct die contact can't cool the CPU enough to provide realistic performance gains above 200W (from a buildzoid chart that I saw earlier this week presumably with the IHS on) then the whole thickness of the IHS for heat dissipation is a moot argument. I've been following the thread long enough to realize 200W is 200W regardless of the temperature. Also the larger temperature differential between the cpu and cooler can be a better scenario for cooling.

So is everyone worried about the excitement of atoms at 95c causing a breakdown of the structures of the CPU like over 15 years beyond warranty?

All this yadda yadda about temperatures, and I'm still on the side of "just slap your big-ass cooler that you already have on it, and enable Eco mode". Problem solved. Based on test data, I honestly believe that there is absolutely no reason to run these CPUs at factory default TDPs.

I bet you can just dial down EDC a bit and reduce temps just like with 5950x without having to go full ECO mode.

 

[AusWolf]

It drops performance to work around a chosen design decision? That makes the product worse than if the decision hadn't been made in the first place.

It doesn't drop performance. It runs well above base clock unless you use the most basic cooler in existence with the lowest fan rpm possible. I get the feeling that you didn't read the article and didn't watch HU's video testing Eco mode.

My i7 11700 non-K runs at 4.9 GHz in single core and 4.4 GHz in all-core workloads. Is it a 500 MHz performance drop, or what? Not to mention the 2.8 GHz I see when the 65 W TDP is enforced as PL1.

I bet you can just dial down EDC a bit and reduce temps just like with 5950x without having to go full ECO mode.

You probably can, but as long as Eco mode drops performance only by a couple percent (literally), and power consumption as well as heat by a considerable margin, I'm fine with it.

 

[Dirt Chip]

Reading comprehension fail.

Double down and keep repeating.

Wtaf is this tactic, we heard your opinion, how many times do you want us to hear it.

I quote

"What you might not be noticing immediately is that with Zen 4 you will now get higher performance in return, automatically, without doing anything, that will be your benefit of upgrading—more perf, not lower temps. As we've seen in this article, if you already have very decent cooling, then the gains from going water are relatively small, and probably not worth it, especially in games. Also, infinitely good cooling will not scale clocks and performance up automatically. Rather there are some hard limits, above which the CPU will not boost, even with excellent cooling."

So baring extreme cooling methods and effort, it's not possible to get more performance than what you have anyway.

A thinner IHs would not boost higher and due to the way it's designed it would , work load allowing get to 95 anyway, because it's designed to aim for that.

And that's a better temperature than 65 because. The differential in temperature between that and your typical ambient temperature.

Actually allows more heat to be removed for the same effort.

The only problem that left is noise.
If you are not sensitive for it, it will be hard to explain.
By choosing 95 degree (at lest part of it a result of maintaining cooler compatibility\thick IHS) you get very high fan speed (read- noise) for no good reason. You need to work around it by testing and capping the fan at 50-80% rpm for it to operate in reasonable noise level in day to day usage, so you loose the 100% cooling capacity in the cases you need them the most (and manually changing the fun curve profile according to application usage will no happen...)

Maybe the amount of people that use fan coolers and are sensitive to noise is small, but still- from noise perspective AMD`s design choice to maintain compatibility lead to thick IHS, lead to set 95 degree as default to work around the thick IHS side effects (and if it`s not the reason then it`s pure stupidity) which lead to a less flexile fan curve that much harder to tune.
By that metric, AMD`s design choice is a design flaw, one that is not so big because you can workaround it`s limitation.
With future AM5 CPUs on smaller process this "problem" might get bigger.

 

[TheoneandonlyMrK]

The only problem that left is noise.
If you are not sensitive for it, it will be hard to explain.
By choosing 95 degree (at lest part of it a result of maintaining cooler compatibility\thick IHS) you get very high fan speed (read- noise) for no good reason. You need to work around it by testing and capping the fan at 50-80% rpm for it to operate in reasonable noise level in day to day usage, so you loose the 100% cooling capacity in the cases you need them the most (and manually changing the fun curve profile according to application usage will no happen...)

Maybe the amount of people that use fan coolers and are sensitive to noise is small, but still- from noise perspective AMD`s design choice to maintain compatibility lead to thick IHS, lead to set 95 degree as default to work around the thick IHS side effects (and if it`s not the reason then it`s pure stupidity) which lead to a less flexile fan curve that much harder to tune.
By that metric, AMD`s design choice is a design flaw, one that is not so big because you can workaround it`s limitation.
With future AM5 CPUs on smaller process this "problem" might get bigger.

Really. Seriously this again.

We covered noise bro, fit the best cooler you got, set up a quiet fan max speed, deal with whatever performance you get in relative quite term's.

 

[Dirt Chip]

Really. Seriously this again.

We covered noise bro, fit the best cooler you got, set up a quiet fan max speed, deal with whatever performance you get in relative quite term's.

Yes, this is the simple solution to the problem presented by zen4.
although minor one, but still a problem. I`ll give it a 5% weight when compering cup's, you might ignore it completely- a subjective matter.
Just note that the root cause of it might (hopefully not) turn it into a bigger problem in yet to come AM5 cup's (zen4 3D variant, zen5 on smaller node and so on).

 

[Mussels]

Imagine this scenario:

- with the best cooler @ it's highest fan speed, the CPU takes 6 minutes to complete task A @ 5.7 GHz
- with the worst cooler @ it's lowest fan speed, the CPU takes 7 minutes to complete task A @ 5.0 GHz

If in the extra time it takes for the CPU with the worst fan speed to complete task A, the CPU ends up consuming more power than the difference in power they both had @ 6 minutes, then it ends up consuming more total power for that task, which means less efficiency. If not, then it's more efficient.

Efficiency curves - you're assuming that they use the same wattage at both tasks, but at lower clock speeds they will be more efficient

 

[HTC]

Efficiency curves - you're assuming that they use the same wattage at both tasks, but at lower clock speeds they will be more efficient

Like i said, it depends on the power used in the extra time taken for the CPU with the worst cooler to perform the task.

Lets re-use the same scenario but this time lets say, for the sake or argument, that the CPU with the best cooler uses 200W/H:

- 6 minutes @ 200W/H, that's 20 watts

For the CPU with the worst cooler to be more efficient in that 7 minute task, it would have to use @ most 171.428W/H: less than that means it's more efficient, but more than that means it's less efficient.

Or is my reasoning flawed?

 

[AusWolf]

Like i said, it depends on the power used in the extra time taken for the CPU with the worst cooler to perform the task.

Lets re-use the same scenario but this time lets say, for the sake or argument, that the CPU with the best cooler uses 200W/H:

- 6 minutes @ 200W/H, that's 20 watts

For the CPU with the worst cooler to be more efficient in that 7 minute task, it would have to use @ most 171.428W/H: less than that means it's more efficient, but more than that means it's less efficient.

Or is my reasoning flawed?

I think your reasoning is excellent.
The only thing is that evidence shows that there is very little difference in performance when you lower the 7950X's power consumption from 230 W to about 170-ish. So that 6-minute task doesn't take 7 minutes. More like 6 minutes and 10 seconds at most.

 

[HTC]

I think your reasoning is excellent.
The only thing is that evidence shows that there is very little difference in performance when you lower the 7950X's power consumption from 230 W to about 170-ish. So that 6-minute task doesn't take 7 minutes. More like 6 minutes and 10 seconds at most.

That would depend on the task. According to @W1zzard 's own frequency scaling chart:

The 32 thread CPU speed goes from roughly 5.2 GHz with the AIO down to roughly 3.2 GHz with the Wraith @ 20%, which is a very VERY significant drop.

 

[A Computer Guy]

That would depend on the task. According to @W1zzard 's own frequency scaling chart:

The 32 thread CPU speed goes from roughly 5.2 GHz with the AIO down to roughly 3.2 GHz with the Wraith @ 20%, which is a very VERY significant drop.

Noctua 40% fan speed is pretty good.

 

[AusWolf]

That would depend on the task. According to @W1zzard 's own frequency scaling chart:

The 32 thread CPU speed goes from roughly 5.2 GHz with the AIO down to roughly 3.2 GHz with the Wraith @ 20%, which is a very VERY significant drop.

But you have the Noctua at 20%, which indicated a max power consumption drop to roughly 160-170 W, with a 97.4% average performance compared to the AIO result. Frequency and performance don't always correlate. At least not linearly.

 

[TheoneandonlyMrK]

But you have the Noctua at 20%, which indicated a max power consumption drop to roughly 160-170 W, with a 97.4% average performance compared to the AIO result. Frequency and performance don't always correlate. At least not linearly.
View attachment 264389View attachment 264390

Yeah but that's reasonable, I want it to be silent, perform at least 50% better than anything else and cost 6p.


: D

 

[HTC]

But you have the Noctua at 20%, which indicated a max power consumption drop to roughly 160-170 W, with a 97.4% average performance compared to the AIO result. Frequency and performance don't always correlate. At least not linearly.

Yes, but Noctua @ 20% isn't the worst cooler.

Besides, that percentage is the average: look closer @ both Cinebench nT and MP3's bars.

 

[AusWolf]

Yes, but Noctua @ 20% isn't the worst cooler.

Besides, that percentage is the average: look closer @ both Cinebench nT and MP3's bars.

Why would you want the worst cooler with a 7950X? My point is that by lowering the power consumption by some ~30% (with any cooler), you lose 3-5% performance at most.

Yeah, Cinebench nT shows something like 90-92% performance? Using your analogy, that's 36 seconds more for the 6-minute task.

230 W x 6 minutes (0.1 hours) = 23 Wh.
170 W x 6.6 minutes (0.11 hours) = 18.7 Wh.

Like I said, your logic is good, but that's exactly why the 7950X is more efficient at lower power.

 

[Max(IT)]

TLDR (TLDW): Enabling Eco mode to bring down the TDP from 105 W to 65 W (PPT from 142 W to 88 W) on the 7600X resulted in basically no performance loss, but a huge reduction in power and thermals.

this is telling us even more how idiotic was this default choice by AMD. Zen 4 would have been good without skyrocketing TDP that way...