Jump to content
Corsair Community

H100i RGB Platinum 5930K too hot?


Recommended Posts

I have had this system for about 5 years now and I switched from a 5-year-old H100i original to the new RGB Platinum H100i. Overall this seems off. I am pretty sure I used the correct stand-offs and everything seems correctly tight. I can feel the pump doing something when I touch the block and if I quickly unplug/plug the SATA power, the CPU immediately starts heating up then cools down again. So it is working.

 

I have two fans in the front of the case (Obsidian 750D) as an intake, a single fan in the back facing out, and the two radiator fans set facing out. Also I cleaned up the case and fans.

 

I have an ASUS Rampage V Extreme and whenever I reinstall the CPU, it says 'new CPU detected' and it seems to turn off OC at least for that boot. In this scenario, the CPU can only go up to 3.5 GHz and stays around 30-35 C.

 

I also re-did the paste in a vertical line form as Arctic Silver suggests.

 

I reduced the VCore to 1.2 V as ASUS' gamer's OC profile for 6 core CPU sets this to 1.3. This seems to have led to slightly lower temperatures but not by much.

 

Note on that profile: it sets cores 1-3 to 45 multiplier and the other 3 to 43. It seems this makes the CPU throttle everything to 4.3 GHz.

 

I have ways to test (e.g. compiling Clang), but when I run at full speed OC to ~4.3 GHz I am easily hitting 90-100 C and this does not seem right. The machine is stable, but I feel I should be able to get better temperatures. Maybe I need to tighten the block down a bit more?

 

I am running Linux, but I've set up iCUE in a virtual machine and set the pump and fans to EXTREME. This doesn't seem to make much of a difference. (Yes I know I have to do this every cold start.)

 

Attached is a screenshot of temperatures and clock frequency. This was taken while starting up Sims 4 (in Wine). Not very heavy but adds some heat to the case due to GPU use.

20200614-screenshot0001.thumb.png.27c72852b954b0bba9fb76286390a3fa.png

Edited by Tatsh
set image width/height
Link to comment
Share on other sites

I have an ASUS Rampage V Extreme and whenever I reinstall the CPU, it says 'new CPU detected' and it seems to turn off OC at least for that boot....

 

Maybe I need to tighten the block down a bit more?

 

 

No. I would avoid doing that at the moment. The 'new cpu detected' bit has me worried the CPU is either already under too much pressure or is sliding around in the socket for some reason. I would take the pump head off so you can see: 1) if the TIM spread evenly on the CPU and cold plate. If it did not, then we know there was a contact issue. 2) Maybe check the CPU's position in the socket, reset it if needed.

 

2011(3) is very straightforward with the asymmetrical posts and simple bolt down system. Adequate pressure and contact is rarely an issue. The stand-offs have a stubbier end on one side and I doubt you could even get the 115x posts to stand up in the thread holes.

 

I ran my 5930K at 4.5GHz/1.30v with a 280mm cooler for a long time. I am not sure I ever broke 80C on a hot day. Something is off, but I can't see a clear indicator of where at this point. Another way to determine if CPU heat is effectively transferred is the rise in coolant temperature (H100i Temp) when under load. However, at the present it would be better to avoid using stress tests.

Link to comment
Share on other sites

The 'new cpu detected' bit has me worried the CPU is either already under too much pressure or is sliding around in the socket for some reason.

 

It seems there might be a sensor on the board that checks if the latches have moved?

 

Regarding the stand-offs, this motherboard came with a bracket as far as I can remember. The manual to the H100i RGB Platinum says for 2011-3 bracket installation is not required.

 

1) if the TIM spread evenly on the CPU and cold plate. If it did not, then we know there was a contact issue.

 

If I check this and it seems fine, do I have to redo the TIM?

 

2) Maybe check the CPU's position in the socket, reset it if needed.

 

CPU doesn't seem to move at all in the socket with the posts down.

 

I ran my 5930K at 4.5GHz/1.30v with a 280mm cooler for a long time.

 

I have been running this on the old Corsair since 2015 but lately that one has been making grinding noises in one of its radiator fans and it's that old, so it's likely the coolant isn't so good anymore. That's why I decided to replace and upgrade.

Link to comment
Share on other sites

It seems there might be a sensor on the board that checks if the latches have moved?

 

Regarding the stand-offs, this motherboard came with a bracket as far as I can remember. The manual to the H100i RGB Platinum says for 2011-3 bracket installation is not required.

 

 

No sensor on the swing bar that locks the processor down. No, you don't need a bracket because the motherboard has its own (all Intel sockets should be like this). However, make sure you are using the new mounting posts that came with the Platinum and not the old ones from prior cooler.

 

 

 

If I check this and it seems fine, do I have to redo the TIM?

 

No, not for a quick check.

Link to comment
Share on other sites

Thanks for the quick replies.

 

I have attached pictures of the spread TIM on the CPU and the bottom of the block. Does this seem okay? This was done with the vertical line method. I put this back in without re-doing the TIM. Made sure it was tight again.

 

I switched the stand-offs as I think I may have been using the old ones. It appears the old ones have no notches on them, and the new ones do. I hope this is correct. I can't verify this easily as the old manual and the new manual don't have pictures with notches. Only the grayed TR4 image on the AMD page shows notches in the stand-offs.

 

Also I am pretty sure the thumb screws are exactly the same. Both sets are able to fully screw on to the stand-offs.

 

I am compiling Clang as a way to test and things seem stable, and though it takes some time, it still reaches 100 C package temperature (load average of 10). I can reduce my desired load for the time being just to keep the temperatures from going so high, at the cost of time of course.

 

After I stop this process, it goes back down to about 60 C but it takes about a minute. See the second screenshot. Eventually gets down to 55. And 45.

 

Should I change my cooling scheme? It seems when I open the front of this case and remove the filter, it cools down a lot faster once the load is taken off.

 

Something is still making this unable to cool effectively for that kind of load.

 

I used to live on a XMP profile that would make the CPU go to about 4.1 GHz and I'm wondering if I should go back to that. I switched to the gamer's OC profile because I really cannot notice the speed-ups from memory OC'ing.

20200614-screenshot0001.thumb.png.471de29f7fde1be8692327a972a18dfa.png

IMG_20200614_201815.thumb.jpg.c3c138308cf9b84208ac2a740cac1b8a.jpg

IMG_20200614_201827.thumb.jpg.cd779ce24f9d08426486c0f2a77dd214.jpg

20200614-screenshot0002.thumb.png.a7c7441e7de57ece1d2bc0bfddd12548.png

Link to comment
Share on other sites

The TIM coverage is fine and I think the last picture after you stop the load shows the issue. When you are running a load and you have high temps because of excessive voltage or bad contact, the CPU temps fall of a cliff the moment you stop. 90-90-90-40-40-38-36.. etc. I don't know anything about the program you're running, but clearly you get a 20C drop and then all the rest is very gradual, even thought he clocks have bottomed out. That reads like a cooler issue and suggests the coolant temp is highly elevated -- well beyond what it should ever be.

 

It would help if you could see the H100i Temp (coolant temperature) in iCUE or Link on the VM. Other than that, there should also be physical signs. The exhaust coming off the radiator should be distinctly warm. If it feels cool after 10 min of load, something is wrong. If one hose on the cooler is very warm and the other is cool to the touch, there is a flow problem. If you cold boot the PC and can watch your CPU temp continually creep up 35...40..45..50...55C etc, there is a flow problem. The same is true in reverse. Once you have loaded up the CPU and it heats up, it takes forever to cool back down. That is what I am seeing in the graph above.

 

Most flow problems are fatal and not user addressable. Pump or fan speed low or high should never affect the cooler in this manner to this degree. The only other thing that might mimic these CPU temp behaviors if if you block the radiator exhaust and it can't get rid of the heat. However, it would take distinctly longer to reach this state and other symptoms like hot/cold hoses won't exist.

Link to comment
Share on other sites

Also I have one of those infrared thermometers and I've checked the following:

 

Block - 45C

Top tube from the block - 48C

Bottom tube - 47C

Fans on radiator - 33C

Radiator on tube side - 50C

Top of case / radiator top - 45-57C

 

Here is a video of things going not at load just in case anything I missed can be spotted in terms of my installation:

 

[ame]https://www.youtube.com/watch?v=xQ1zZ50qJHo&feature=youtu.be[/ame]

 

As for q-code 00, I don't know why it happens but when I boot up it shows AA when it hits the boot loader and once Linux starts it flips to 00. It says unused in the ASUS manual so I don't think it's anything to worry about.

Edited by Tatsh
Link to comment
Share on other sites

Here is a side by side screenshot with iCUE. The temperature of the coolant gets to be 50 C and greater at full load.

 

 

50C is out of bounds. Normal coolant rise on a 5930K with a 240mm should be +8C at medium fan speed. That means you would need to be in a 35C room to make 50C. However, the shown fan speed is 400 rpm. Is that accurate? Did you set them that low? It doesn’t sound like 400 rpm in the video.

Edited by c-attack
Link to comment
Share on other sites

I didn't set the fans that low. I don't see a place in iCUE where I can directly set the fan speed. I have the fans and the pump set to EXTREME. This only stays until I do a cold reboot. Is there something to make this permanent ?

 

Have to admit, the iCUE software is a bit confusing.

 

When the cooler reverts on a cold start, what mode is it in? I know it's reverted because it does not have the color settings I picked. I would presume it would be in a mode where it 'just works' for most people without needing configuration in iCUE.

Link to comment
Share on other sites

I looked at some others screenshots of iCUE with this same cooler and they had fan speeds over 2000. I think my primary issue is these fan speeds. I cannot get them to be above 350 RPM. Have no idea why.

 

I reconnected the power plugs on each fan just in case. Still no change.

Link to comment
Share on other sites

Can you show me a screenshot of the H100i Platinum -> Performance tab in iCUE?

 

The thing is this still shouldn’t happen on any of the presets. On all three Settings should slam to maximum after you cross the 40C threshold. The dashboard indicates they are connected to the pump for control. Are they set to fixed PWM%?

Link to comment
Share on other sites

Attached screenshot of H100i Platinum -> Performance. I clicked on EXTREME and I guess this just highlights those that are in 'extreme' preset.

 

I'm not sure if it's because of the VM that the numbers would be off. Or that any other data is missing. I can try booting Windows I guess to see what it says but it will take a little while because I do not have that installed at the moment. The VM is not in any way directly attached to my motherboard so it cannot report things like CPU temperature. I don't know if this messes up how iCUE is supposed to work.

Capture.thumb.PNG.25781d83f3df7a7659d5eecfca2b7ba8.PNG

Link to comment
Share on other sites

I don't know where I can look with the default presets to see if the fans are set to fixed PWM%. If I make a custom profile with the the PWM% set to 100 and apply it to a fan, the fan speed still stays at its 'maximum' of ~350.

 

See screenshot.

custom-fixed-rpm-3000-no-go.thumb.png.82b2eb7b305df2e2086e9fbc306e24fd.png

Link to comment
Share on other sites

The temp sensor for coolant is internal and even if you never installed iCUE, the fans should aggressively respond to coolant temps > 40C. The next question is are those fans really at ~400 rpm. That would be barely moving with almost no detectable air through the radiator.

 

Don’t install Windows. I’ll ask others on VM’s to see what they know, but either way the solution would be to move the fans to the MB if the cooler is not able to self-regulate.

Link to comment
Share on other sites

Whoa.

 

I did a forced firmware update (despite it saying I'm already up to date), and now my fan speeds are finally normal. I can actually hear them. Temperature is actually going down finally.

normal-balanced.thumb.PNG.7fa4ef3892abc6f4c3cf79910f039836.PNG

Link to comment
Share on other sites

OK, good. Although this is one of the stranger issues I've seen. Be careful with repeated firmware flashes. They don't always go as planned. If this kind of thing occurs again, I would reach out to Corsair Support with an eye toward replacement.
Link to comment
Share on other sites

Also, do you have any input on how to achieve 4.5 GHz (without OC'ing memory)?

 

I tried but it was not stable. It just kept hanging on high load.

 

CPU core voltage - 1.3

All cores set to 45 multiplier

 

I also tried with voltage 1.35. I don't want to go any higher.

Link to comment
Share on other sites

Not all 5930/5820K will be able to do 4.5. There was a pretty clear hurdle going from 4.4 to 4.5. The X99 BIOS was pretty simple in terms of overclocking. LLC level 1-2 down from max. Set the voltage. If you can't get 4.5 all core at 1.30, I would split the difference and try per core 4.5 for 2 and 4.4 for 3-6. Voltage can stay at 1.30 and see how that goes. The difference between 4.4 and 4.5 is not going to be visible on anything other than a bench mark.
Link to comment
Share on other sites

  • 8 months later...

Curiously I have just had an issue installing an H100i on a maximus V MB. Closely similar to you're MB generationaly speaking.

I discovered that the standoffs when secured to the backplate leave a gap so the back plate is loose enough to reposition. However with the pump mounting tabs on the H100i being an integral part to the pump, unlike my older H80 unit with a separate mounting bracket, the shorter standoffs in the new kit are still to long and leaves the pump actually loose to the chip heat spreader, even as the thumbscrews tightened down as evidenced by being able to move the backplate about after the install.

I finally had to mod the pump mounting brackets by slightly bending them back such that when tightening the thumbscrews some contact pressure could be applied by drawing the backplate up snug to MB rear which then allowed the pump block to be better secured to the heat spreader.

 

I am sure that voids the warranty but at least when I run intel's processor diagnostic tool it supplies 100% load and my 3770K, notorious for heat issues, tops out at about 72C. Previously it would shut down as the temp just kept rising past 105C halfway through the temp test.

It's not a obvious thing, it took me two days to ferret it out.

Hope this can help out in some way.

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

×
×
  • Create New...