Question Ryzen 7600X 95°C Prime95. Normal?

[IDProG]

The title.

Case is NR200.
CPU cooler is Thermalright Peerless Assassin, but with no fan. Instead, I use the 92mm fan at the back and the 120mm fan at the top as intakes to cool the heatsink.

I did a Small FFT All Core with SMT Prime95 test because I only wanted to test the CPU, not the RAM. It reached 95°C in, like, 3 seconds.

The heatsink isn't warm, either.

Can it be because I applied too much thermal paste?

 

[Lafong]

Highly unlikely thermal paste has anything to do with it, assuming you re-attached the cooler correctly.

What are your temps when doing your traditional tasks, under a good load?

Prime 95 is rarely representative of what you'd encounter when using the PC for its intended purposes, but you might be highly anxious about Prime 95 temps for some reason?

 

[IDProG]

Highly unlikely thermal paste has anything to do with it, assuming you re-attached the cooler correctly.

What are your temps when doing your traditional tasks, under a good load?

Prime 95 is rarely representative of what you'd encounter when using the PC for its intended purposes, but you might be highly anxious about Prime 95 temps for some reason?

I am at the beginning of the PC usage, if that makes sense. I haven't really done anything yet, though I am planning to use it for Unity game development.

At idle, the CPU temp is around 43°C (the first CPU temp in HWINFO64). I think a good CPU idle temp is at below 40°C (well, at least this is true for my previous 1500X CPU).

Prime95 is an overclocker's best friend alongside HWINFO. While I am not an overclocker, I am an undervolter, meaning that I need them.

 

[Zerk2012]

I am at the beginning of the PC usage, if that makes sense. I haven't really done anything yet, though I am planning to use it for Unity game development.

At idle, the CPU temp is around 43°C (the first CPU temp in HWINFO64). I think a good CPU idle temp is at below 40°C (well, at least this is true for my previous 1500X CPU).

Prime95 is an overclocker's best friend alongside HWINFO. While I am not an overclocker, I am an undervolter, meaning that I need them.

Why no fan on the cooler

 

[Phaaze88]

Prime95 is an overclocker's best friend...

It isn't. The loads it applies are too flat for that.
For Core i, P95 is thermal stress test - makes them go balls to the wall on power use.
Non-7000 Ryzens, on the other hand, behave differently with that app. Ryzen 7000 differs further, as AMD designed them to run at 95C specifically when the user loads all the cores - then it's just a matter of how far the cooler can keep the temperature away from that limit. The further away it is, the higher the average clock speeds will be.

 

[Deleted member 2838871]

Ryzen 7000 differs further, as AMD designed them to run at 95C specifically when the user loads all the cores - then it's just a matter of how far the cooler can keep the temperature away from that limit. The further away it is, the higher the average clock speeds will be.

Interesting... didn't know that.

I've ran 30 min all core burn tests and hit 89C after 1 minute and then stayed there for 29 mins. The D15 does a very nice job.

Haven't approached those temps in every day use yet but am looking forward to seeing the numbers as soon as I start processing my next batch of video footage.

 

[Phaaze88]

Interesting... didn't know that.

I've ran 30 min all core burn tests and hit 89C after 1 minute and then stayed there for 29 mins. The D15 does a very nice job.

Haven't approached those temps in every day use yet but am looking forward to seeing the numbers as soon as I start processing my next batch of video footage.

I got that from this:

View: https://www.youtube.com/watch?v=OGuFq3jm9hM

 

[IDProG]

It isn't. The loads it applies are too flat for that.
For Core i, P95 is thermal stress test - makes them go balls to the wall on power use.
Non-7000 Ryzens, on the other hand, behave differently with that app. Ryzen 7000 differs further, as AMD designed them to run at 95C specifically when the user loads all the cores - then it's just a matter of how far the cooler can keep the temperature away from that limit. The further away it is, the higher the average clock speeds will be.

I still don't get it.

Is HWINFO lying about the temps?
The Thermalright Peerless Assassin can cool >200W TDP CPUs. 7600X has, like, 105W TDP.

Is it even possible (and safe) for a CPU to reach 95°C so quick before the cooler can dissipate the heat?

Will the temps go down if I just wait for the heat to dissipate for, like, 3 minutes?

Will the temps not reach 95°C if I set a manual voltage and clock speed?

 

[Deleted member 2838871]

I got that from this:

Ahhhh... yeah when it says runs HOT I figured they were talking about the non-x3D chips. The TDP is 170 vs 120 on the x3D... another reason I pulled the trigger on it.

 

[Phaaze88]

Is HWINFO lying about the temps?
The Thermalright Peerless Assassin can cool >200W TDP CPUs. 7600X has, like, 105W TDP.

No, it's not.
You have to understand that cpus are complex - there's more to it than that. There's factors such as thermal density, how the IHS will flex under thermal cycles, how many 'layers' are involved in heat transfer...

Is it even possible (and safe) for a CPU to reach 95°C so quick before the cooler can dissipate the heat?

Yes. The adages about older Intel chips that most are used to don't apply to these. You also applied an all core load, of which they are specifically designed to act that way.
Look at it another way: the actual dangerous thermal territory for Ryzen 7000 would require you to raise thermal limit up from 95C.
These companies already know that if they give users too much freedom, there will be extra cases of folks breaking stuff. The 95C starting line wouldn't be a thing if AMD's own engineers didn't believe it was safe.

Will the temps go down if I just wait for the heat to dissipate for, like, 3 minutes?

If you stop the all core load, of course the temperature will decrease.

Will the temps not reach 95°C if I set a manual voltage and clock speed?

This comes up so much, that I think AMD should've included a disclaimer with these cpus. Most PC customers are not tech savvy or look at few, if any reviews, and will get caught off guard(overreact) by these cpus.
Thus, they go and take a chance on reducing/increasing cpu performance because they let their personal feelings get in the way.
Some folks also panicked with Ryzen 3000 and 5000, since they did something different from the 'norm': they had higher low load thermals...

Let the cpu do its thing, and maybe get more fans in there; the cpu will push higher clocks on its own the better cooling you have. Ryzen 3000 and 5000 did it too, minus the 95C thing.

 

[IDProG]

This comes up so much, that I think AMD should've included a disclaimer with these cpus. Most PC customers are not tech savvy or look at few, if any reviews, and will get caught off guard(overreact) by these cpus.
Thus, they go and take a chance on reducing/increasing cpu performance because they let their personal feelings get in the way.
Some folks also panicked with Ryzen 3000 and 5000, since they did something different from the 'norm': they had higher low load thermals...

Let the cpu do its thing, and maybe get more fans in there; the cpu will push higher clocks on its own the better cooling you have. Ryzen 3000 and 5000 did it too, minus the 95C thing.

Well, I am a bit more tech-savvy than the average consumer. I know how undervolting works. Honestly, I always consider undervolting the CPU/GPU a must, because stock voltage is made with the absolute worst silicon in consideration. So, average silicon should be able to reduce some heat by undervolting it.

I just found out that I can get away with 4.7GHz @1.1V. And it's not that hot, around 87°C max. Now that I think about it, it does make sense, because my 1500X got away with 3.5GHz (the base clock speed) @1.1V. Then again, the 1500X used Wraith Spire and the temp was around 78°C max.

MAYBE, hopefully I can reduce the voltage even more before it gets unstable. Then I'll move on to RAM overclocking.

Thank you for the insight on how hot the Ryzen 7000 is.

 

[Phaaze88]

I just found out that I can get away with 4.7GHz @1.1V. And it's not that hot, around 87°C max. Now that I think about it, it does make sense, because my 1500X got away with 3.5GHz (the base clock speed) @1.1V. Then again, the 1500X used Wraith Spire and the temp was around 78°C max.

The first 2 sentences are exactly what I'm referring to - letting personal feelings get in the way. So you know better about the cpu's limits than what the folks who actually designed the cpu do?
4.7ghz at 87C? In the video I linked TravisPNW to further above, Hardware Canucks was getting 5.253(technically 5.3)ghz out of the 7600X at that same temperature, and they didn't do anything to it.
A ~13% difference in frequency. Even at 95C, it was at 5.1ghz.


These 2 cpus are so different, that's not a fair comparison.
-1500X has lower thermal density present. The higher thermal density is, the harder it is to cool; the on die hot spots are focused at smaller points, making heat transfer more challenging.
-The primary CCD of the 1500X is in the center. 7600X is offset.
-The IHS on the 1500X is thinner than the 7600X's.
-Auto clock boost behavior on 1500X isn't nearly as aggressive as 7600X.


Good luck.

 

[IDProG]

So you know better about the cpu's limits than what the folks who actually designed the cpu do?

I don't know how to answer this. The premise itself is already faulty.

Like I said, when designing stock voltage, they take the absolute worst silicon into consideration. Otherwise, they will have to measure every single CPU they produce to find out the best voltage each CPU can do.

Yes, they know more than I do, but that does not mean anything in this particular context about voltage.

4.7ghz at 87C? In the video I linked TravisPNW to further above, Hardware Canucks was getting 5.253(technically 5.3)ghz out of the 7600X at that same temperature, and they didn't do anything to it.
A ~13% difference in frequency. Even at 95C, it was at 5.1ghz.

That's right, 5.253GHz at 87°C and 5.0-ish GHz at a higher temp, even though they should be the other way around. The chart clearly was describing something else.

I just finished undervolting my CPU. 4.7GHz @ 1.0V, around 63W package power and 76°C max when stress-tested. Very happy with the results.

As a bonus, it can do 5GHz @ 1.1V and 5.2GHz @ 1.2V, though the latter reaches 105°C when stress tested.

 

[Deleted member 2947362]

According to AMD's website max temp for CPU.

Max. Operating Temperature (Tjmax)
95°C

I'm not 100% sure but I think once the CPU reaches 105c it thermal throttles it's self to prevent any damage to the CPU.

you can google what temp the CPU thermal throttle kicks in for that CPU

To give you an idea of how hot that is

Soft solder typically has a melting point range of 90 to 450 °C

Im not saying it's going to melt the solder on the CPU or motherboard but blimey that's a toasty CPU!

I'm sticking with the older R5 5600 76c max when stress testing it, but in normal usage like gaming never goes much above 64-66c

EDIT
The real reason I'm sticking with the R5 5600 is because I cant afford to upgrade to a whole new platform and the lower temps just comfort me now (lol)

https://www.amd.com/en/products/cpu/amd-ryzen-5-7600x