New AMD Ryzen 2700X CPU running HOT

Using the H80i V2 cooler with a newly installed 2700X CPU (old CPU was the 1800X).

 

2700X says it's running at 193.3F but pump says temp there is 82.8F. Temp in CPUID HWMonitor also is around 190.0F.

 

Is this the right cooler for the 2700X?

That's not hot, it's a meltdown. More likely this is a sensor error. See below. If you get 3rd degree burns at 12 inches away, then there is a more serious problem.

 

http://forum.corsair.com/v3/showthread.php?t=177005

Took the 2700X out and reinstalled the 1800X and temperatures back to normal. Something is not posting correctly as my fans and pump were in turbo mode, too.

Doesn't this CPU require a BIOS update? Did you do that yet?

Yeah, I was at BIOS F6 when I installed the 1800X. Checked the specs for the 2700X and Gigabyte said to upgrade to F23d BIOS which I did.

Do you have any tools from AMD to monitor the temp? I know that temps on AMD chips have always been a challenge ... this release seems to really highlight that. But I'd think that AMD's tools would be close.

Used my current tools on my 1800X and they worked fine.

 

Someone on the AMD forum mentioned that they thought there was a problem with Windows where it could make a CPU stick in full clock mode for both AMD and Intel processors. Opened a trouble ticket on Tenforums to see if anyone there would know. Will post the results.

The other thing you need to figure out is whether this was real or phantom heat. A physical problem needs to be dealt with immediately. Even if the CPU temps were accurate, that would be a 70C+ CPU temp to coolant temp delta. That is extraordinarily high. You only get that if your voltage has gone crazy (would show in HWMonitor or Link) or the cold plate did not make good contact with the CPU. That would have been easy to spot when you changed chips, with trademark chunks of unspread TIM.

CPU-Z showed the CPU multiplier was x42 which is the chips top speed, so I imagine that it was really that hot.

 

Checked the pump plate when I removed the chip and it looked totally covered with Arctic Silver. The 1800X is currently running at 98F so I think everything was reading the temps correctly.

 

Have a ticket on Tenforums, so I'll see if they have anything to say.

The suggestion was the monitoring software has not been updated or is as of yet unable to accurately read the 2700X core temps. The proper reading of the older 1800X does not validate the other. This has been a long running theme with AMD.

 

Even if all 8 cores are locked at 4.2, you still would need something a bit out of control and the voltage to create the heat. There seem to be a very wide range of power consumption numbers out there for 2700x and I have seen as much as 250W for Prime small FFT listed. So while theoretically temps in that range are possible it would have to come with a bucket of voltage and a clear and marked rise in coolant temperature. 28C on the coolant is probably just a bit over resting idle, depending on your room and case temp. This would be a way to differentiate actual heat output (wattage) from something else. Ultimately the answer to is this enough cooler depends on your case and what you can fit. A H80i can handle the wattage, but if have an above the GPU placement you might ultimately take on some coolant temp penalties in mixed use. However, in a CPU only usage environment, this should never be an issue.

 

**EDIT: I just noticed in the other thread you posted a Link screen shot. Your pump speed is too low. The H80i v2 needs a full 12v power supply. Adjust the fans through Link, not the BIOS or other software fan controls if they are plugged into the H80i v2 pump controller. Given you fan speed is zero, they may be on the board somewhere. That is fine, but the motherboard header with the H80i v2 needs to be disabled/100%/Maximum or whatever language GA uses to describe 12v on a fan header. This may not be the entire scope of the issue being discussed, but it is certainly not good to run in this state and it will affect both cooling performance and longevity.

C-Attack:

 

The screenshot that was posted may have shown the fans at 0 speed. However, they were at turbo. I've got a trouble ticket open with Corsair about their software not reading fan speed, and I just received an update. Now the software shows the fans.

 

What may be the real issue is when I checked the Gigabyte site for my mobo for CPU compatibility the page said to use a F23d version of the BIOS, which was done. However, after this fiasco, I checked all the specs again and the page had changed the BIOS to F22. That was installed last night.

 

I'm going to try the reinstall this afternoon and will be back with the results.

There is no "Turbo" setting in Link. That sounds like Asus BIOS talk, although I am sure other brands use it as well. It doesn't really matter where you power the fans from for the radiator, but the pump connector on the H80i v2 and H100i v2 must have a 100% fan curve. This is best done by disabling CPU_FAN or wherever the H80i connected. Fan speed makes a tiny difference in short term temperatures and only has meaning over a long duration. Since your coolant temp was also not out of control yet, the pump was running but you will damage it and have terrible cooling by continuing to use it that way.

When I referred to 'turbo' I meant that they were running so fast the PC was practically rocking.

 

Anyway, re-installed the 2700X under the Gigabyte F22 BIOS and everything is running fine. 2700X temp is 110F under 15% load.

 

Seems that there's a bug in the F23d BIOS.

 

Thanks everyone for your help!

So it's an issue with the bios and not the Corsair Link? When I go into the BIOS my 2700x is running at 23 C, and I too have a Gigabyte mother board with 23d installed. Corsair Link is the one reporting the temp at 80 c, and whenever I try to run a game my computer sounds like a jet engine ready to take off. I literally don't want to blow out my cooler. So what's the solution here? How long will it take Corsair to fix this glitch? I still have seen no update for Corsair Link.

 

I remember this same problem when the 1800x came out, basically Corsair Link was wrongfully reporting the temperature as 40c higher. This time the problem is worse and is actually affecting the cooling system itself.

 

I'm using an H80i v2.

So it's an issue with the bios and not the Corsair Link? When I go into the BIOS my 2700x is running at 23 C, and I too have a Gigabyte mother board with 23d installed. Corsair Link is the one reporting the temp at 80 c, and whenever I try to run a game my computer sounds like a jet engine ready to take off. I literally don't want to blow out my cooler. So what's the solution here? How long will it take Corsair to fix this glitch? I still have seen no update for Corsair Link.

 

I remember this same problem when the 1800x came out, basically Corsair Link was wrongfully reporting the temperature as 40c higher. This time the problem is worse and is actually affecting the cooling system itself.

 

I'm using an H80i v2.

 

The immediate solution would be to use H80i Temp (coolant temp) as the control source for the fans. This is the default setting and how the cooling actually works. Using CPU temp often results in racy fans on any CPU. If you are rear mounted exhaust, you will need to make some allowance for GPU heat, but the expected range should still be predictable.

Similar issue using the Gigabyte Gaming K7 BIOS F23d with newly installed 2600x

 

Corsair link is reporting 75- 85 degC and HWMonitor is reporting 30 degC. As link controlls the rad fan its all maxed out. The H110i coolant temp is reading 26 degC and this usually rises to around 30 degC when its actually under load so would assume their is an error with the CPU package temp reporting?

 

Going to go back a BIOS level as have seen a few reports of this running hot on F23d whether sensor reporting error or not.

 

Regards

Coe.

 

edit: reinstalled link and same issue

 

also downloaded a 3rd software (HWInfo64) to check core temp and it reports the same as HWMonitor, a normal 30 degC.

Definitely pointing to an issue with the way link is taking info from the CPU package temp sensor.

The immediate solution would be to use H80i Temp (coolant temp) as the control source for the fans. This is the default setting and how the cooling actually works. Using CPU temp often results in racy fans on any CPU. If you are rear mounted exhaust, you will need to make some allowance for GPU heat, but the expected range should still be predictable.

 

Okay, how do I go about doing that?

This is a Corsair Link issue not a BIOS issue. As long as your BIOS is current on the Gigabyte motherboard, in BIOS the temps read correctly, in Ryzen Master the temps read correctly, in the newest version of HDInfo the temps read correctly. In CUE the reading is off by a factor. One thing that I did is dump managing fans in Windows and manage through BIOS only. Then all but CPU FAN #1 can be triggered by some other temperature monitor. In my case I am using coolant temp as I have a sensor on EC_TEMP#1. So even when Corsair thinks the temp is sky high, which it isn't actually, the fans still run at reasonable speeds.

Okay, how do I go about doing that?

 

In Corsair Link, select the fan from the H80/100/110/115i (or whatever i cooler you have). This should be on the right side next to the coolant temperature. Coolant/Liquid Temp is called “H115i Temp” or whatever model you have. When you click the fan, a pop up appears and will list the “group”. This is the control variable. Make sure this is set to Hxxx Temp and not CPU Package Temp.

Reply from my support ticket:

 

We have tested this against a Ryzen 2 and can verify that yes, the temperatures are reporting incorrectly in our software. We are currently working on a fix for this, and should hopefully have something out in a future update soon. We apologize for any inconvenience which this is is causing.