Question Ryzen 9 5900X and Ryzen 5 2600 Thermals.

stchman

Distinguished
Oct 7, 2012
104
1
18,685
0
Hello all.

I recently put together a new system(System 1 in my signature).

I ran thermal stress tests on both system 1 and 2. Here are my results.

System 1:
-Windows 10
-Prime 95 with Blend
-HW Info for monitoring
After ~2 hours, the CPU reached a max temperature of ~88C. HW Info reported the CPU was drawing ~140W of power. The fans on the H100i were spinning at approx 1900RPM(far more noise than when using the system normally)


System 2:
-Windows 10
-Prime 95 with Blend
-HW Info for monitoring
After ~2 hours, the CPU reached a max temperature of ~54C. HW Info reported the CPU was drawing ~80W of power. The fan was spinning on the GAMMAX at ~1000RPM(no appreciable sound increase)

Are these typical thermals? The 5900X consumes about 60W more power than the 2600 so that is quite a bit more.

Thank you.
 

stchman

Distinguished
Oct 7, 2012
104
1
18,685
0
What case is system 1 in and where inside is the AIO oriented and what fan config?
Case is in my signature, the Cooler Master 690 II.

AIO is located at the top of the case in a push configuration with air being sucked in from the outside to the inside of the case. The case has a front 140mm intake fan and a rear 140mm exhaust fan.

From a lot of what I read, the ideal configuration with an AIO is push with air being sucked in to case.
 

punkncat

Honorable
Ambassador
I have no direct experience with the 5900X, but hitting 88C is awful close to it's thermal limit, and seems high to me for an AIO.
I think I would play with the fan configuration, perhaps set the fans to push through the radiator from the top? Perhaps push from below exhausting the case....It doesn't appear that the case offers another place to mount the radiator. I think it's what a 120 in front and a couple of optional 80s on the case panel?
 

logainofhades

Titan
Moderator
95C is the thermal limit, for a 5900x. You are still within limits. The 5900x has twice the cores, and 2 chiplets, compared to the single chiplet of a 2600. More power/heat is expected. Also, I believe the 5900x boosts higher, which in turn will mean more power draw/heat.
 

stchman

Distinguished
Oct 7, 2012
104
1
18,685
0
I have no direct experience with the 5900X, but hitting 88C is awful close to it's thermal limit, and seems high to me for an AIO.
I think I would play with the fan configuration, perhaps set the fans to push through the radiator from the top? Perhaps push from below exhausting the case....It doesn't appear that the case offers another place to mount the radiator. I think it's what a 120 in front and a couple of optional 80s on the case panel?
I already said that the AIO is in a push configuration.


95C is the thermal limit, for a 5900x. You are still within limits. The 5900x has twice the cores, and 2 chiplets, compared to the single chiplet of a 2600. More power/heat is expected. Also, I believe the 5900x boosts higher, which in turn will mean more power draw/heat.
This is obviously an extreme situation, only during stress testing will I ever have all 24 threads maxed out to 100%. Normal operating temps for the CPU during gaming are ~45-50C.


Some online have said that the stock thermal paste on the H100i is quality and that replacing it will yield no improved thermals. Others have said that the stock thermal paste is garbage and should not be used.

During the Prime95 stress test, the temperature in the room increased noticeably.
 

Phaaze88

Titan
Ambassador
You are seeing the weakness of the multi chip module design, combined with a few other things: high socket thermals.
Looks pretty bad under sustained loads the likes of Prime95...
 
Last edited:

stchman

Distinguished
Oct 7, 2012
104
1
18,685
0
Thank you for all the replies. I am still surprised that nobody said I need to use Liquid Metal or some exotic thermal paste that promises to reduce CPU temps by 40C.
 

Phaaze88

Titan
Ambassador
I am still surprised that nobody said I need to use Liquid Metal or some exotic thermal paste that promises to reduce CPU temps by 40C.
Because those who've actually used both know that's far from the case.
Liquid metal should be used under the IHS, or in direct die applications. It's not very good with the usual application on top of the IHS - paste is supreme here.
Testing has already been done with some of the best pastes, and the difference between them is minor: https://www.tomshardware.com/best-picks/best-thermal-paste
It's hard to screw up with paste, unless there's specific instructions(TG Kryonaut), or the user bought some cheapo junk off Aliexpress or something.
The pre-applied paste is OK.
 
90C

Ryzen™ 9 5900X | Gaming Desktop Processors | AMD
I'm sure the actual thermal limit of the silicon in these processors is higher, but AMD considers temperatures up to 90C as being "typical and by design for full load conditions". According to an AMD representative...

I want to be clear with everyone that AMD views temps up to 90C (5800X/5900X/5950X) and 95C (5600X) as typical and by design for full load conditions. Having a higher maximum temperature supported by the silicon and firmware allows the CPU to pursue higher and longer boost performance before the algorithm pulls back for thermal reasons.
So at least according to them, a 5900X running at 90C shouldn't be a problem. I probably wouldn't want my processor pushing those temperatures for extended periods on a daily basis, but it doesn't sound like you are experiencing that anyway. Prime95 isn't a real-world workload, and it pushes temperatures higher than you are likely to see in actual applications, so temperatures with that running don't really matter much. It's used to test a worst-case scenario in terms of power draw and heat output, and so long as the system isn't crashing with it running, it's probably fine.

As for comparing the 5900X to a 2600, there are a lot of differences. In addition to the 5900X having twice as many cores and threads, those cores are also clocked significantly higher, and are performing more computations per clock. With all threads loaded at stock settings, a 2600 will only be running its cores at a little under 3.8GHz, while a 5900X can push over 4.5GHz in a similar scenario. And while the 5900X is actually a lot more efficient due to it's updated architecture and the 7nm process it utilizes, it's ultimately performing close to 3 times as much work when all threads are loaded like that, so power draw will be higher. The 2600 also uses a monolithic design, with the cores contained on a single, larger block of silicon along with the rest of the chip, whereas the 5900X has its cores contained on two chiplets each smaller than a pinky-finger nail, and separate from the remainder of the processor on its own larger IO chip. So the cores each have less surface area to transfer their heat to the heat spreader, which can contribute to higher temperatures within them, even though the total amount of heat getting dissipated by each core is actually slightly lower.
 

ASK THE COMMUNITY

TRENDING THREADS