Question Strange H45 behavior: CPU hot, but cooler cold

[Marco Abramo]

I have a Corsair H45 on my Ryzen 5800X. In normal (light) use it stays around 35-45C (ok, it is normal). In games or high demand situation, stays around 60, 75C (normal, again)

However, something sounds very strange for me: The temperature of the block, pipes and radiator DO NOT increase, NEVER!!! Even CPU at 70C for a long time, the block, pipes and radiator stays around 25C, 30C (just few degrees above room temperature) even with the CPU at 70C for a long time, which it makes NO sense to me.

I must emphasize that I'm not complaining about the temperatures (temps are normal, ok) , but this huge difference, i can't understand this, it doesn't make sense for me. I've been building my PCs for over 15 years, and I've never seen anything like this.

This situation happens with any thermal interface: With the original thermal interface that came on the H45 itself, with Deep Cool Z5 and with MX-4. It didn't make any difference. Obviously, i checked the contact between the block and the CPU, the amount of thermal paste, the pump RPM (100%), etc.

I also have a Corsair H55 on my RTX 2080TI and in this case, things work as they should: If the GPU 60C, the block, pipes and radiator are VERYHOT to the touch, exactly what you should expect.

Well, WHY CPU hot, but H45 cold?!?!?

Thx!👍

 

[kanewolf]

I have a Corsair H45 on my Ryzen 5800X. In normal (light) use it stays around 35-45C (ok, it is normal). In games or high demand situation, stays around 60, 75C (normal, again)

However, something sounds very strange for me: The temperature of the block, pipes and radiator DO NOT increase, NEVER!!! Even CPU at 70C for a long time, the block, pipes and radiator stays around 25C, 30C (just few degrees above room temperature) even with the CPU at 70C for a long time, which it makes NO sense to me.

I must emphasize that I'm not complaining about the temperatures (temps are normal, ok) , but this huge difference, i can't understand this, it doesn't make sense for me. I've been building my PCs for over 15 years, and I've never seen anything like this.

This situation happens with any thermal interface: With the original thermal interface that came on the H45 itself, with Deep Cool Z5 and with MX-4. It didn't make any difference. Obviously, i checked the contact between the block and the CPU, the amount of thermal paste, the pump RPM (100%), etc.

I also have a Corsair H55 on my RTX 2080TI and in this case, things work as they should: If the GPU 60C, the block, pipes and radiator are VERYHOT to the touch, exactly what you should expect.

Well, WHY CPU hot, but H45 cold?!?!?

Thx!👍

Failed pump. May be reporting 100% but seems like no coolant is moving.

 

[Marco Abramo]

Failed pump. May be reporting 100% but seems like no coolant is moving.

If the pump is faulty, the rpm would not appear in the BIOS AIO_PUMP monitor, Asus Suite Monitor, HW Monitor and etc.

And even if the bomb is really failed, the block (which is in direct contact with the CPU) would be extremely hot, because the water would be standing there and heating up all the time. But it's cold. No make sense, even if the bomb was stopped .

Listening carefully, it is possible to hear the (very low) noise of the pump and fell the vibration in the pipes.

Also, the CPU temperatures are normal, at idle or high demand. If the pump was not running, the CPU temperatures would be high, and they are not. No make sense, again.

 

[kanewolf]

If the pump is faulty, the rpm would not appear in the BIOS AIO_PUMP monitor, Asus Suite Monitor, HW Monitor and etc.

And even if the bomb is really failed, the block (which is in direct contact with the CPU) would be extremely hot, because the water would be standing there and heating up all the time. But it's cold. No make sense, even if the bomb was stopped .

Listening carefully, it is possible to hear the (very low) noise of the pump and fell the vibration in the pipes.

Also, the CPU temperatures are normal, at idle or high demand. If the pump was not running, the CPU temperatures would be high, and they are not. No make sense, again.

There are only three things possible.
Bad pump.
No coolant.
Bad mounting, with insufficient contact pressure.
Bad thermal interface.
It is a pretty simple thing. Heat transfers from hot to cold. So either there is interference in the transfer (bad mount or bad grease) or you cant remove the localize heat (bad pump or no coolant).
I am not a fan of AIO coolers.

 

[Marco Abramo]

There are only three things possible.
Bad pump.
No coolant.
Bad mounting, with insufficient contact pressure.
Bad thermal interface.
It is a pretty simple thing. Heat transfers from hot to cold. So either there is interference in the transfer (bad mount or bad grease) or you cant remove the localize heat (bad pump or no coolant).
I am not a fan of AIO coolers.

I would agree, but... if bad pump, or bad mounting, bad thermal interface, etc, the CPU must be very hot in high demand or stress test, okay?

But never exceeds 65 - 75 C in high demand, and auto-throttling never occurs (90 C for 5800X). And normal desktop use, never exceeds 30, 40, 45C.

The CPUs temps are very normal, at all times, ever.

Conclusion: The the heat transfer to cooler is normal.

I really checked the contact between the WC block and the CPU, the amount of thermal paste (MX-4 is a good thermal interface), i checked 2 times. Its all fine, and the H45 is very easy to mount, very easier than an air cooler (see example photo, just 4 screw with pressure springs, which press the block on the CPU very tight)

 

[Phaaze88]

Ryzen can run far less power than the 2080Ti's core. Not really going to soak the liquid as well as the latter.
Even though it doesn't use much power, the 5000 series chip is harder to cool by design.
The multi die layout means there's heat radiating in more than one spot(unlike Intel's monoliths) and some cooler cold plates are designed around heat radiating outward in a small area from a central point.
As the process nodes continue to shrink; power efficiency improves, but the thermal density goes up, increasing difficulty of thermal transfer.
We've hit a wall where the silicon can hit their thermal limits under an AIO, at least.

 

[Marco Abramo]

Ryzen can run far less power than the 2080Ti's core. Not really going to soak the liquid as well as the latter.
Even though it doesn't use much power, the 5000 series chip is harder to cool by design.
The multi die layout means there's heat radiating in more than one spot(unlike Intel's monoliths) and some cooler cold plates are designed around heat radiating outward in a small area from a central point.
As the process nodes continue to shrink; power efficiency improves, but the thermal density goes up, increasing difficulty of thermal transfer.
We've hit a wall where the silicon can hit their thermal limits under an AIO, at least.

Hey, this makes a lot of sense to me.

High thermal density -----> increasing difficult of thermal transfer in a small surface area.

Well, i guess i should consider what's happening to be normal, despite being weird or bizarre... Because, as I pointed out, the CPU temperatures are normal.