Is this normal for 10900k?

[Apollo190]

I'm getting as high as 96 degrees when gaming or doing some other cpu intensive tasks all at once. I'm using Dark Rock Pro 4 and thermal grizzly kryonaut for the cooling. Both the cpu, paste and cooler are pretty new (been only a few weeks since I got them) and there's absolutely no dust inside.
I did change some settings in bios however (motherboard is TUF Z490 Plus). For example I changed ASUS MultiCore Enhancement from auto to "remove all limits", and I also changed Power-saving from auto to performance mode. Didn't overclock or undervolt at all though, not yet at least.
So, is this the expected temperature for my setup or did I do something wrong?

[Jurrunio]

12 minutes ago, Apollo190 said:

For example I changed ASUS MultiCore Enhancement from auto to "remove all limits", and I also changed Power-saving from auto to performance mode. Didn't overclock or undervolt at all though, not yet at least

Turning on MCE is already overclocking, and performance mode means setting power limit really high... I'm not surprised with that temperature, at all.

[Apollo190]

9 minutes ago, Jurrunio said:

Turning on MCE is already overclocking, and performance mode means setting power limit really high... I'm not surprised with that temperature, at all.

Oh really? I only made those changes because I noticed non-stop power limit throttling, I didn't know it would affect it that much. Is there anything I could do to lower the temps, but avoid power limit throttling?

Btw the case I'm using is TUF GT501, there's 3 intake fans at the front and 1 exhaust fan on the back, and panels are perforated so airflow should be sufficient.

[VeganJoy]

temps are directly correlated to the wattage you put through there, by default the boost tdp is 250w and with MCE it probably completely unlocks the chip power limit, pulling 250-300w. leaving it at stock will use just 125w after the boost period is over, which is only a few seconds for intel's actual default limits. most of today's air coolers were designed to deal with 4-6 haswell or skylake cores running well under 5 ghz, so putting 10 skylake cores under there and tryna run at the default 4.9-5.0ghz boost just aint feasible. my noctua d15s had trouble with 9th gen chips pulling ~200w.

 

so if you wanna run that chip without power limits, you'll need a robust liquid cooler. a 280-360mm radiator would keep temps decent, my buddy's 10900kf pulling 300w peaked around 85C or so with a kraken x63 running at max pump speed and with arctic p14s instead of the stock fans. arctic, corsair, and nzxt make some of the more popular aios and are good choices. a 240mm aio would be very similar to the drp4 you have right now so i'd try for larger sized aios.

[Jurrunio]

55 minutes ago, Apollo190 said:

avoid power limit throttling

impossible even with undervolting... I hope you didnt buy Intel for efficiency, 125w PL1 will be violated all the time.

 

55 minutes ago, Apollo190 said:

Btw the case I'm using is TUF GT501, there's 3 intake fans at the front and 1 exhaust fan on the back, and panels are perforated so airflow should be sufficient.

air cooler is not really sufficient for MCE and blasting the power limit up high when we're talking about a 14nm 10 core. Undervolt and disable MCE first, you can raise power limit if you dont want it to power throttle so much but keep it to something reasonable, say 200w.

[Apollo190]

5 hours ago, Jurrunio said:

impossible even with undervolting... I hope you didnt buy Intel for efficiency, 125w PL1 will be violated all the time.

 

air cooler is not really sufficient for MCE and blasting the power limit up high when we're talking about a 14nm 10 core. Undervolt and disable MCE first, you can raise power limit if you dont want it to power throttle so much but keep it to something reasonable, say 200w.

Linus said in this video that nh-d15 should be getting 80-high/90-low with thermal pad, and even less with quality thermal paste. In general, most reviewers say there's not much difference between nh-d15 and drp4, maybe 2-3 degrees at most. Linus even compared it to a 280mm aio here, it had pretty much the same results.

[CommanderAlex]

2 hours ago, Apollo190 said:

Linus said in this video that nh-d15 should be getting 80-high/90-low with thermal pad, and even less with quality thermal paste. In general, most reviewers say there's not much difference between nh-d15 and drp4, maybe 2-3 degrees at most. Linus even compared it to a 280mm aio here, it had pretty much the same results.

Have you messed around with the fan curves of the CPU cooler and intake fans to increase RPM when increasing temperatures? For example, I have my Noctua NH-D15 as well as the case fans go to 100% at 70 Celsius and during Cinebench benchmarks, they reach about 75 C and nothing higher unless I use a crappier thermal paste.

[Apollo190]

52 minutes ago, CommanderAlex said:

Have you messed around with the fan curves of the CPU cooler and intake fans to increase RPM when increasing temperatures? For example, I have my Noctua NH-D15 as well as the case fans go to 100% at 70 Celsius and during Cinebench benchmarks, they reach about 75 C and nothing higher unless I use a crappier thermal paste.

I haven't done that yet, I'll give it a try, it might actually help. 75 sounds really good, well done man

 

[CommanderAlex]

21 minutes ago, Apollo190 said:

I haven't done that yet, I'll give it a try, it might actually help. 75 sounds really good, well done man

 

Thanks, I'm cooling a 9800X and using Noctua's NH-T1 thermal paste in case you wanted to know. 

[Jurrunio]

5 hours ago, Apollo190 said:

Linus said in this video that nh-d15 should be getting 80-high/90-low with thermal pad, and even less with quality thermal paste. In general, most reviewers say there's not much difference between nh-d15 and drp4, maybe 2-3 degrees at most. Linus even compared it to a 280mm aio here, it had pretty much the same results.

They are on an open test bench and mid 20s for ambient temperature if I have not mistaken, maybe your environment is worse than that? Taking off the side panel of the case can improve case temperatures a lot so try that first

 

Also I dont know what voltage Linus and your system runs when power limit is effectively gone. Possible that your board just made stupid decisions there (too high) while their board didnt.

[Energycore]

Three things to keep in mind.

 

1) Are you running fans at full speed? Your fan curve in BIOS might not be ramping them all the way up.

2) Do you live in a hot environment? CPU and GPU temps increase with temperature, so a setup that peaks at 65C if ambient is at 20 will peak at 75C if ambient is at 30.

3) What case do you have? If it has poor airflow, you'll be recirculating hot air instead of moving cool air into your CPU cooler, which will cause it to have abnormally high temperatures.

[Apollo190]

4 hours ago, Jurrunio said:

They are on an open test bench and mid 20s for ambient temperature if I have not mistaken, maybe your environment is worse than that? Taking off the side panel of the case can improve case temperatures a lot so try that first

 

Also I dont know what voltage Linus and your system runs when power limit is effectively gone. Possible that your board just made stupid decisions there (too high) while their board didnt.

It was 26°C in my room when I tested it. I'm not sure about the voltage, I'll have to try undervolting to see if it helps.

 

2 hours ago, Energycore said:

Three things to keep in mind.

 

1) Are you running fans at full speed? Your fan curve in BIOS might not be ramping them all the way up.

2) Do you live in a hot environment? CPU and GPU temps increase with temperature, so a setup that peaks at 65C if ambient is at 20 will peak at 75C if ambient is at 30.

3) What case do you have? If it has poor airflow, you'll be recirculating hot air instead of moving cool air into your CPU cooler, which will cause it to have abnormally high temperatures.

1) Yeah the fans were at full speed when it hit 96°C.

2) It was 26°C in my room when I tested it.

3) The case is TUF GT501, I said in a previous post that it has 3 intake fans at the front and 1 exhaust fan on the back, and panels are perforated too.

[Energycore]

15 minutes ago, Apollo190 said:

1) Yeah the fans were at full speed when it hit 96°C.

2) It was 26°C in my room when I tested it.

3) The case is TUF GT501, I said in a previous post that it has 3 intake fans at the front and 1 exhaust fan on the back, and panels are perforated too.

Hmm. Not sure if airflow on it is amazing. I can't tell from pictures how wide the gap of air intake is, but the perforations are quite a bit of steel, probably more than half.

 

I would test with the side panel off. See if you gate the same temps. If not, then the cause is poor airflow of the case design.

 

96C at 26C room temperature is pretty damn high, so let's hope it's that since you've ruled out most other things.

 

Looks like the 10900K draws about 120W in fully threaded workloads, I would expect a ballpark temperature under 75C with the DRP4 fan running at 100%.

 

If the case test doesn't work, I'd reapply thermal paste, being careful to add enough and to put enough mounting pressure on the cooler.

[-rascal-]

12 minutes ago, Energycore said:

Hmm. Not sure if airflow on it is amazing. I can't tell from pictures how wide the gap of air intake is, but the perforations are quite a bit of steel, probably more than half.

 

I would test with the side panel off. See if you gate the same temps. If not, then the cause is poor airflow of the case design.

 

96C at 26C room temperature is pretty damn high, so let's hope it's that since you've ruled out most other things.

 

Looks like the 10900K draws about 120W in fully threaded workloads, I would expect a ballpark temperature under 75C with the DRP4 fan running at 100%.

 

If the case test doesn't work, I'd reapply thermal paste, being careful to add enough and to put enough mounting pressure on the cooler.

 

Remember, OP has Multi-Core Enhancement enabled, AND limits removed / set to unlimited (e.g. Intel spec'd PL1 and PL2 times omitted).

Power options is also set to Performance Mode.

Not sure if those settings have been disabled / dialed back yet.

 

If MCE is enabled, and limits ignored, wouldn't the i9-10900K boost max / near max (e.g. 5.3 GHz) across all cores...basically forever ?

In that case, that is waaay beyond the advertised so-called "120W" spec.

[Jeppes]

6 minutes ago, -rascal- said:

 

Remember, OP has Multi-Core Enhancement enabled, AND limits removed / set to unlimited (e.g. Intel spec'd PL1 and PL2 times omitted).

Power options is also set to Performance Mode.

Not sure if those settings have been disabled / dialed back yet.

 

If MCE is enabled, and limits ignored, wouldn't the i9-10900K boost max / near max (e.g. 5.3 GHz) across all cores...basically forever ?

In that case, that is waaay beyond the advertised so-called "120W" spec.

Even with stock settings its about 200W:s in CB r20 as the max boost duration is something like 50-60s so that they get a nice score with stock settings too.

[Energycore]

1 hour ago, -rascal- said:

 

Remember, OP has Multi-Core Enhancement enabled, AND limits removed / set to unlimited (e.g. Intel spec'd PL1 and PL2 times omitted).

Power options is also set to Performance Mode.

Not sure if those settings have been disabled / dialed back yet.

 

If MCE is enabled, and limits ignored, wouldn't the i9-10900K boost max / near max (e.g. 5.3 GHz) across all cores...basically forever ?

In that case, that is waaay beyond the advertised so-called "120W" spec.

Right, I was under the assumption that these were disabled but if they are, 70C Delta T over Ambient is still way too high for a DRP4 I would say.

 

I'd expect a 10C drop from solving whatever issues there are.

 

I guess MCE counts as a 5.3GHz Overclock huh

[Apollo190]

I had some spare time and tested with blender classroom this time (Multi-core enhancement still enabled), just like linus in that video, and peak temperature was 90C, with average around 87C (room temperature 25C). Not bad, but I was expecting a few degrees cooler, like mid 80s or something. Btw I also checked core clocks during the test, and all cores were running at 4.9 GHz throughout the whole test, so I guess multi core isn't pushing them to 5.3 for some reason.

[Apollo190]

On 10/30/2020 at 10:27 PM, Energycore said:

Right, I was under the assumption that these were disabled but if they are, 70C Delta T over Ambient is still way too high for a DRP4 I would say.

 

I'd expect a 10C drop from solving whatever issues there are.

 

I guess MCE counts as a 5.3GHz Overclock huh

I just noticed there are 4 different CPU temperature readings in hwinfo:

Which one is the "real" one that I should monitor ?
(The third column in this picture shows the max temperature after blender benchmark)

EDIT: I also noticed in hwinfo that "RING: Max VR Voltage, ICCmax, PL4" and "IA: Electrical Design Point/Other (ICCmax,PL4,SVID,DDR RAPL)" were showing "YES" during my testing, which is probably why I had occasional 100MHz drops. Then I changed Processor Core IccMax from 255.500A to 255.750A in Intel XTU and both of those were now showing "NO" during the test. The power draw was 200W in both cases, and voltage at ~1.26V (even though I set it to 1.35V in bios and LLC to 4). Peak temperature was 84C this time (I guess it's lower because I changed voltage from auto to 1.35V).

Edited November 3, 2020 by Apollo190