Ryzen 7 7700X - TjMax Query

[tpa-pr]

Hi all,

New to the forum (hello), not new to IT but am new/returning to high performance computing and hardware has never been my strong suit. I recently built my first gaming rig in ~10 years in Oct '22 (I have worked entirely off laptops since then) and I am hoping someone can give a layman an understanding of the 7700X and the "95C is the new normal" meme. I've tried my best to understand it from my own research but it's hard to work out what is correct and incorrect.

To summarise my understanding:
- When put under heavy load, the 7700X will immediately clock up and push to 95C (or as close to as it can) and stay there without throttling. Presumably throttling would start if it exceeds this temp?
- 95C is the "TJMax" of the CPU, TJMax being the max operating temperature of the chip (which older chips would normally begin to throttle at).
- TJMax is different to the max SAFE operating temperature and is not putting the chip at any risk. Older chips throttle to prevent reaching damaging temperatures.
- TJMax is a completely arbitrary temp set by the manufacturer.
- Degradation and/or damage to the chip wouldn't occur until much higher temps (I can't find anything from AMD specifically but I see a lot of people mentioning 125C?).

Have I understood the "new normal" correctly?

 

[thewan]

Its not arbitrary if AMD guarantees it for 3 years of operation at 95c. If their RnD department cant prove what is valid to be RMA to their insurance department, your RMA will never be approved on an "arbitrary" value should your CPU fail within the warranty period.

 

[tpa-pr]

Its not arbitrary if AMD guarantees it for 3 years of operation at 95c. If their RnD department cant prove what is valid to be RMA to their insurance department, your RMA will never be approved on an "arbitrary" value should your CPU fail within the warranty period.

Right, poor choice of wording on my part. What I meant by "arbitrary" was that TJMax is decided by the manufacturer based on their internal testing and benchmarking for that chip, not necessarily something more general like the peak temp for silicon as a material before it degrades. Am I correct in that?

 

[Chrispy_]

I believe long-term silicon degredation for older process nodes like 7FF were 105C. Whether that thermal threshold applies to TSMC 5nm is a different question that I don't have the answer to.

Either way, 95C is fine. Intel have been running laptop silicon at those temperatures without issues for about a decade. If it was a real problem there would be countless reports of failure and class-action lawsuits galore!

 

[Count von Schwalbe]

When put under heavy load, the 7700X will immediately clock up and push to 95C (or as close to as it can) and stay there without throttling. Presumably throttling would start if it exceeds this temp?

Throttling is a word that doesn't really apply in this case. It dynamically adjusts the clockspeeds to maintain a 95C, rather than suddenly choking down and killing performance like older processors do. If your cooling setup doesn't keep up, it could be running 95C consistently at a lower speed than the advertised boost clock, but it won't suddenly clock way down to drop temps.

 

[tpa-pr]

I believe long-term silicon degredation for older process nodes like 7FF were 105C. Whether that thermal threshold applies to TSMC 5nm is a different question that I don't have the answer to.

Either way, 95C is fine. Intel have been running laptop silicon at those temperatures without issues for about a decade. If it was a real problem there would be countless reports of failure and class-action lawsuits galore!

OK, I thought that might've been the case for the reason you've given (laptops have always run hot to some degree). It was less a concern about the temp itself and more clarification on what "max operating temp" meant. This helps, thank you.

Throttling is a word that doesn't really apply in this case. It dynamically adjusts the clockspeeds to maintain a 95C, rather than suddenly choking down and killing performance like older processors do. If your cooling setup doesn't keep up, it could be running 95C consistently at a lower speed than the advertised boost clock, but it won't suddenly clock way down to drop temps.

Understood, so it's less a hard choke and more of a "floating" shift up/down in clock speed to stay at that temp rather than pull below it. 95C is the ceiling AMD have set so in times where consistent performance is required the CPU will hit that limit and stay there with only gradual adjustments to clock speed. Performance is maintained and the temp limit is below any danger, which means no damage to the silicon, which means no need to throttle down to prevent damage. Am I on the right track?

Therefore the primary time(s) to be concerned is:
- If I am hitting 95C at a clock lower than max boost consistently (which most likely means inadequate cooling or if it's further down the lifetime of the PC, a clean and re-paste is needed).
- If the temps exceeds 95C consistently (which most likely indicates cooling system failure or something else has gone wrong).

Otherwise hitting TJMax for long periods is fine and safe. Is there anything I've missed or misunderstood?

 

[dirtyferret]

To summarise my understanding:
- When put under heavy load, the 7700X will immediately clock up and push to 95C (or as close to as it can) and stay there without throttling. Presumably throttling would start if it exceeds this temp?
- 95C is the "TJMax" of the CPU, TJMax being the max operating temperature of the chip (which older chips would normally begin to throttle at).
- TJMax is different to the max SAFE operating temperature and is not putting the chip at any risk. Older chips throttle to prevent reaching damaging temperatures.
- TJMax is a completely arbitrary temp set by the manufacturer.
- Degradation and/or damage to the chip wouldn't occur until much higher temps (I can't find anything from AMD specifically but I see a lot of people mentioning 125C?).

from the techspot review

When it comes to cooling, Zen 4 CPUs give the impression that they're difficult to cool by delivering as much performance as possible by taking full advantage of the thermal and power headroom. AMD says with the new AM5 platform and higher TDP, Zen 4 processors will run into a thermal wall before they hit a power wall. This means under heavy load they'll sit at TJMax which is about 95 degrees Celsius for the Ryzen 7000 series, and this will be particularly true for the 12 and 16-core models.

AMD has stressed that this behavior is by design and it's important to note TJMax is the maximum safe operating temperature -- not the absolute maximum temperature. AMD went on to say that 95C is not running hot, rather Zen 4 will intentionally go to this temperature under load because the power management system knows that this is the ideal way to squeeze the most performance out of the chip without damaging it. The processors are designed to run at TJMax 24/7 without risk of damage or deterioration they added.

 

[Chrispy_]

OK, I thought that might've been the case for the reason you've given (laptops have always run hot to some degree). It was less a concern about the temp itself and more clarification on what "max operating temp" meant. This helps, thank you.

95C is the target temperature. That's the easiest way to think about it.

If you're not reaching 95C at full load, then that means one of several possibilities:

• Your cooling is much better than it needs to be.
• Your motherboard has reached its own power/current/VRM temperature limit.
• Your multithreaded workload isn't truly parallel and a single core that other threads are waiting on has reached it's max voltage/frequency.

Geforce cards have had an 82C temperature target for ages now, possibly over a decade. It doesn't mean that they throttle at 83C, it just means that the opportunistic boost has run out of opporunity at that point. It's still boosting, but not quite as hard. If the temperature drops once the fans ramp up, AND the card still has power budget left to spend, it'll increase the boost again. That is why it's called opportunistic; It boosts as hard as it can when the opportunity arises.

 

[tpa-pr]

from the techspot review

When it comes to cooling, Zen 4 CPUs give the impression that they're difficult to cool by delivering as much performance as possible by taking full advantage of the thermal and power headroom. AMD says with the new AM5 platform and higher TDP, Zen 4 processors will run into a thermal wall before they hit a power wall. This means under heavy load they'll sit at TJMax which is about 95 degrees Celsius for the Ryzen 7000 series, and this will be particularly true for the 12 and 16-core models.

AMD has stressed that this behavior is by design and it's important to note TJMax is the maximum safe operating temperature -- not the absolute maximum temperature. AMD went on to say that 95C is not running hot, rather Zen 4 will intentionally go to this temperature under load because the power management system knows that this is the ideal way to squeeze the most performance out of the chip without damaging it. The processors are designed to run at TJMax 24/7 without risk of damage or deterioration they added.

It is absolutely amazing that I somehow manage to miss this review entirely when it has the explanation I need, lol. That is some Googling failure on my part.

95C is the target temperature. That's the easiest way to think about it.

If you're not reaching 95C at full load, then that means one of several possibilities:

• Your cooling is much better than it needs to be.
• Your motherboard has reached its own power/current/VRM temperature limit.
• Your multithreaded workload isn't truly parallel and a single core that other threads are waiting on has reached it's max voltage/frequency.

Great, thank you very much and thank you to everyone else who chipped in. I now have a proper understanding of how my CPU operates.

As I said in my original post, this is my first build in about a decade so I'm sure I'll have more questions to follow. But this one can be ticked off the list.

 

[dirtyferret]

95C is the target temperature. That's the easiest way to think about it.

I now have a proper understanding of how my CPU operates.

I will say this much about it, AMD states it's fine ( I have no reason to doubt them) but you still have a part that's hitting 95c in your PC case as opposed to something that was topping off around 64c while OC (W1zzards 5700x review) just six months ago. That increase of 30c may not impact the CPU for performance but it does raise the overall internal temp of your PC case that may impact other parts for example your PSU fan which now has to run at a higher RPM.

 

[Count von Schwalbe]

it does raise the overall internal temp of your PC

How? It has an identical TDP/PPT to the 5800X - which means that it is dumping the same amount of heat into the case. Temperature is not heat. Temperature is caused by the efficiency of heat transfer.

The AM5 socket is simple less efficient at transferring the 142 watts into your case than, say, LGA 1700 at transferring the 253 watts of a 13700K.

It does make setting a quiet CPU fan curve more difficult, though, as the CPU is often at TjMax.

 

[dirtyferret]

How? It has an identical TDP/PPT to the 5800X

1. TDP is an abused stat that both Intel and AMD are guilty of playing with
2. I was talking about the 5700x (hint, I mentioned "W1zzards 5700x review" which the 7700x replaces in the AMD lineup and uses twice the power consumption (according to W1zzard slightly more than twice) at stock settings while being in your own words "simple less efficient at transferring the 142 watts" and creating a higher temp.

So here is my question to you (sincere question)

It does make setting a quiet CPU fan curve more difficult

Clearly at stock fan curve settings, the fan RPM will increase due to the temp monitor. So wouldn't a higher spinning cpu fan* help spread heat similar to how increasing the blower speed on a furnace helps heat your home?

*air cooler

 

[Bill_Bright]

To summarise my understanding:
- When put under heavy load, the 7700X will immediately clock up and push to 95C (or as close to as it can) and stay there without throttling. ng temperatures.
- TJMax is a completely arbitrary temp set by the manufacturer.

No, that's not how it works.

If your CPU and case cooling are adequate and doing their jobs, even with a heavy load (assuming default clocks and voltages) the temperature should NOT immediately (or ever, for that matter) push up to 95°C. Instead, it should stay comfortably below that threshold.

Crossing a thermal threshold and throttling are bad things! For sure, you want the CPU to throttle if that threshold is crossed - otherwise the alternative is, it burns up! In effect, that throttling threshold is like a circuit breaker - a safety feature. Not a goal to achieve. And if crossed, that signals to the user, better cooling is required.

I understand what that Techspot review is saying there, but IMO, they are not explaining it correctly. They are suggesting that 95°C spec is like a car engine's radiator thermostat value. The radiator thermostat is designed to keep the engine at an ideal operating temperature - because that is where the engine works most efficiently. And that's fine.

But Techspot is suggesting CPUs operate optimally when maintained at that specific temperature too, just like a car engine. That just is not true. It is not how electronics works (except for crystal oscillators that sit in crystal ovens set to maintain specific temps in order to maintain specific oscillation frequencies). Electronics work optimally when the operating temperature is maintained within a certain "range" - essentially, anywhere comfortably within that range. Too cold is not good, too hot is definitely not good.

And as noted, it is not an arbitrary number.

95C is the target temperature. That's the easiest way to think about it.

A "target" is something one aims for. The goal is not to aim for 95°C, but to avoid reaching it. If the manufacturer rates the CPU TJmax (or Tjunction max) for 95°C we should be "aiming" to keep the CPU's temps well below that most of the time, and aim to never let it reach that threshold.

Something else to keep in mind, sustained high temperatures (even when they don't exceed thermal protection thresholds) increase aging. While the CPU, running near 95°C most of the time "may" not increase aging of the CPU itself, it "may" increase aging on the surrounding and supporting components. For example, long time users may have noticed on some inadequately cooled system, the plastic beige CPU sockets are no longer beige, but brown. That's due to long term exposure to high temperatures and can result in the plastics becoming brittle - damaging the socket.

The many heat sensitive components surrounding the CPU socket experiencing long-term exposure to high temperatures "may" age more quickly too. This is an issue users implementing water block and side firing CPU HSF solutions need to keep in mind when implementing such alternative cooling solutions.

All this is why (1) a quality case and case cooling (often overlooked ) is so critically important and (2) why notebook cases will never make good gaming computer cases.

@tpa-pr - if you are concerned about cooling - and it appears you are (a very good thing! ) then I recommend two things. (1) Look at your case cooling. If we are talking about the Corsair 7000D Full Tower listed in your specs, while a "full" tower may be overkill, nevertheless it offers excellent cooling options to ensure a great front-to-back flow of cool air through the case. And (2) monitor your temps in real-time. I use and recommend Core Temp to monitor CPU temps in real time. Under Options > Settings > Notification Area, I have mine set to display "Highest temperature" only. This displays the temperature of the warmest core, instead of a system tray icon for each of the 8 cores. If CoreTemp shows your CPU temps is sitting up close to that 95° threshold much of the time, you might consider adding another case fan, and/or lowering your ambient (room) temperature.

 

[Count von Schwalbe]

1. TDP is an abused stat that both Intel and AMD are guilty of playing with

No disagreement there.

2. I was talking about the 5700x (hint, I mentioned "W1zzards 5700x review" which the 7700x replaces in the AMD lineup and uses twice the power consumption (according to W1zzard slightly more than twice) at stock settings while being in your own words "simple less efficient at transferring the 142 watts" and creating a higher temp.

Compared to a 5700X, yes. Compared to a 5800X, not really. I was thinking of the 7700X as being the practical successor of the 5800X.

Clearly at stock fan curve settings, the fan RPM will increase due to the temp monitor. So wouldn't a higher spinning cpu fan* help spread heat similar to how increasing the blower speed on a furnace helps heat your home?

By all means. But it will reduce your case temperature (assuming a proper case fan setup), as you are reducing the temperature rise of the air. Increasing your furnace blower speed at the same heating stage will reduce the air temperature coming out of your vents, but more BTUs* are coming through with the increased air flow.

The reason I took exception to your post was because it looked like you were linking a 95C processor with higher case temperature, rather than a high power draw processor with increased case temperature.

 

[dirtyferret]

Increasing your furnace blower speed at the same heating stage will reduce the air temperature coming out of your vents, but more BTUs* are coming through with the increased air flow.

correct the air temp coming through may lower but the increased BTU's will raise the air temp in said room

The reason I took exception to your post was because it looked like you were linking a 95C processor with higher case temperature, rather than a high power draw processor with increased case temperature.

no worries

 

[Count von Schwalbe]

A "target" is something one aims for. The goal is not to aim for 95°C, but to avoid reaching it. If the manufacturer rates the CPU TJmax (or Tjunction max) for 95°C we should be "aiming" to keep the CPU's temps well below that most of the time, and aim to never let it reach that threshold.

According to AMD:

TJMax is the max safe operating temperature – not the absolute max temperature. In the Ryzen 7000 Series, the processor is designed to run at TJMax 24/7 without risk of damage or deterioration. At 95 degrees it is not running hot, rather it will intentionally go to this temperature as much as possible under load because the power management system knows that this is the ideal way to squeeze the most performance out of the chip without damaging it.

 

[HD64G]

I will say this much about it, AMD states it's fine ( I have no reason to doubt them) but you still have a part that's hitting 95c in your PC case as opposed to something that was topping off around 64c while OC (W1zzards 5700x review) just six months ago. That increase of 30c may not impact the CPU for performance but it does raise the overall internal temp of your PC case that may impact other parts for example your PSU fan which now has to run at a higher RPM.

CPU temp isn't equal to cooler temp and thus doesn't affect the case temp in such a way.

 

[Bill_Bright]

According to AMD:

I know. I saw that. And I stand by what I said.

Just because 95°C is the max "safe" temperature, that does not mean it is a "target" temperature you want to get to. If a bridge is designed to hold a maximum of 10 tons, does that mean you want to put and keep 10 tons on it?

If a tire is designed to hold a maximum of 50psi, does that mean you want to put and maintain 50psi in it?

Contrary to what many want to believe, electronics are bound to the same laws of physics as other technologies.

It seems apparent to me that AMD lawyers, marketing weenies and bean counters wrote that - not the engineers and designers. Anyone with any formal education and training in electronics knows you don't want to "sustain" your electronics at their maximum safe temperature levels.

CPU temp isn't equal to cooler temp and thus doesn't affect the case temp in such a way.

Umm, kinda, sorta, but not really - some times.

For the most part, this is true - at least if there is an adequate flow of cool air through the case - essential to exhaust the CPU's heat out the case. But certainly, if air flow through the case is inadequate, the heat from the CPU will indeed buildup inside the case and heat up the case. This is evident with just about every small case because they don't support multiple, large fans needed to create an adequate supply of cool air flowing through the case.

 

[Count von Schwalbe]

It seems apparent to me that AMD lawyers, marketing weenies and bean counters wrote that - not the engineers and designers. Anyone with any formal education and training in electronics knows you don't want to "sustain" your electronics at their maximum safe temperature levels

Two sides to the coin - that is the maximum they say it is. It could be 105C like the TPU db states, or 200C for all I know.

If a bridge is designed to hold a maximum of 10 tons, does that mean you want to put and keep 10 tons on it?

Yes, if I can transport goods across it faster that way. A bridge with a stated 10 ton weight limit can hold 50 tons. I see no reason that AMD would not attempt to avoid backlash and potential lawsuits by putting the stated TjMax well below the actual point of degradation.


I still am inclined to believe that it is designed to boost UNDER HEAVY LOAD until it reaches TjMax, and that the designers are counting on those heavy loads being relatively infrequent to protect the CPU.

Let us bear in mind that outside of networked rendering "farms" or crypto mining, all-core heavy loads for hours on end are extremely rare.

I don't necessarily agree with it, but I believe the statement - it is a handy way to get marketing slides for performance claims.

The fact is, whether you believe me or Bill, AMD has made thermal throttling a much smoother process, eliminating the sudden choking down of old.

 

[Bill_Bright]

Two sides to the coin - that is the maximum they say it is. It could be 105C like the TPU db states, or 200C for all I know.

Huh? It is not two sides. Even "IF" 95°C is somewhat arbitrary, it doesn't change the physics - especially since we don't know if the upper limit is really 105° or 200°C. The goal should never be to run electronics at their "maximum" allowed - which BTW, is the definition of TJmax.

The fact is, whether you believe me or Bill, AMD has made thermal throttling a much smoother process, eliminating the sudden choking down of old.

Totally irrelevant. Just because a safety feature works smoothly, that in no way suggests the goal is to trigger it.

I still am inclined to believe that it is designed to boost UNDER HEAVY LOAD until it reaches TjMax, and that the designers are counting on those heavy loads being relatively infrequent to protect the CPU.

This really is a contradiction. Think about what you just said. You are saying the designers intentionally designed the CPU to jump up to 95°C, but if it hits 96° it will throttle back down. Really?

Yes - under heavy load. So what? Serious gaming creates a heavy load. Nothing unusual or unexpected about that.

You are also suggesting that CPU will perform better at 95°C than it would at 85°C. Do you really believe that? Because I have to tell you, that simply is NOT true.

 

[A Computer Guy]

Let us bear in mind that outside of networked rendering "farms" or crypto mining, all-core heavy loads for hours on end are extremely rare.

Other kinds of networked cloud service farms also run CPU's all core. Not necessarily for single workloads but for example loading balancing server activity with as many users as possible or to a determined CPU threshold before a new CPU node needs to be added to a system.

 

[Count von Schwalbe]

You are also suggesting that CPU will perform better at 95°C than it would at 85°C. Do you really believe that?

Given identical amounts of cooling, a CPU that can boost up to 95 has more performance headroom than one that will only push to 85.

I believe that AMD has programmed the boost algorithm to make use of thermal headroom up to 95C, rather than limiting it in other ways (to a point, of course).

Other kinds of networked cloud service farms also run CPU's all core. Not necessarily for single workloads but for example loading balancing server activity with as many users as possible or to a determined CPU threshold before a new CPU node needs to be added to a system.

Yes. I was pointing out that long-term sustained all-core load requiring full boost is extremely rare for any PC.

 

[Bill_Bright]

Given identical amounts of cooling, a CPU that can boost up to 95 has more performance headroom than one that will only push to 85.

No! Totally different scenarios. I was not talking about two different CPUs. Not to mention, if not published, you don't know how much "headroom" either of those have. They could both have 25° for all we know.

I clearly am not explaining this well because you are not understanding the issue.

I never suggested two different CPUs. I said "that CPU" suggesting the same CPU in both scenarios.

So I say again, with your previous comment, you are suggesting "that CPU" will perform better at 95° compared to that exact same CPU running at 85°. That is not how it works.

Ironically, I have had the exact same debate with others who strive for the lowest possible temps. Some believe a CPU running at 30°C will perform better, be more stable, and have a longer lifespan than a CPU running at 50°C. Also not how it works.

 

[Count von Schwalbe]

No! Totally different scenarios. I was not talking about two different CPUs. Not to mention, if not published, you don't know how much "headroom" either of those have. They could both have 25° for all we know.

I clearly am not explaining this well because you are not understanding the issue.

I never suggested two different CPUs. I said "that CPU" suggesting the same CPU in both scenarios.

Sorry, I meant that if you limit a 7700X to 85C, you are potentially leaving a certain amount of clockspeed and therefore performance on the table, vs leaving it to being limited at 95C. This of course assumes you have cooling up to the task of cooling a 7700X.

 

[Bill_Bright]

Sorry, I meant that if you limit a 7700X to 85C

You are not limiting it. There's no choke, or governor, or current limiting device preventing it from going higher. You are just keeping it cooler most of the time.

Like I said above, I feel like marketing weenies and lawyers wrote that AMD product information and not real engineers and designers.

To me, as someone who is formally trained in electronics, and with an active 45+ year career as an electronics technician maintaining IT hardware since the mid 1970s, I believe this is nothing more than marketing hype in an attempt to rationalize those higher temps. This is to keep us users from complaining about those higher temps (and higher energy consumption rates). This is especially notable in this day and age when the rest of the world (well, much of it anyway) is concerned (and rightfully so) about global warming, deforestation, pollution, and skyrocketing energy costs.

Sorry. I don't know how to make this more understandable. If TJunction is defined as the,

the maximum thermal junction temperature that a processor will allow prior to using internal thermal control mechanisms to reduce power and limit temperature

then it is NOT desirable to push to and sustain the CPU's temperature at that maximum temperature. If you don't accept that, then by all means, cut back on your case cooling and run your CPUs to those levels. The good news is (as you indicated earlier), throttling should be a smooth process.