(Help) i3-9350KF @ 5.1 GHz w/ 1.41 Vcore, higher temps under load than expected (80+ C) with Thermalright Peerless Assassin 120 SE & Thermalright TF7

[Polaria]

Hello all,

 

I recently purchased an allegedly brand new (as the eBay seller claimed) i3-9350KF basically for the hell of it, and decided I'd pick up a cooler that's probably a bit overkill for this 91 watt processor (I figured I may as well see if I could push this thing to the moon). Installation wasn't hard, it came with their decent looking TF7 paste which was a little annoying to apply but I think I did it right, and the whole setup looks pretty slick in my case even if it's in the wrong color. Unfortunately, I don't seem to be getting anywhere near the performance that reviews across the internet promised for the PA120 - despite only drawing about 90-95 watts with AVX (but not AVX2) enabled on Prime95, I skyrocket all the way to 100 degrees within a couple of minutes and thermal throttle. Even just benching on CPU-Z shoots me to about 77-80 degrees.

 

This seems unusual. I've read a couple of reviews stating that this cooler can handle 200+ watt intel CPUs @ about 75-80 degrees celsius (at my room temperature), and Tom's Hardware had a graph that I calculated for my temps to be probably anywhere around 56-60 degrees under load (their 95 watt TDP graphs). Even when ran at stocks clocks and similar voltages to my i3-9100F it exhibits odd behavior, both with my stock cooler (which it ran on for a few days) and the PA-120, such as weird core temperature variance (core temp differences will be 35 degrees or greater sometimes between cores), very sharp spikes when under sudden load such as from 38 degrees at idle straight to 80 degrees within 1-2 seconds, and just generally what feels like lackluster performance. A lot of these behaviors don't seem to be present on my other Coffee Lake CPU, even at similar voltages.

 

What could the issue here be? Did I maybe overtighten the cooler, misapply thermal paste, etc or could there be something else wrong, like dried out/simply insufficient TIM under the IHS or simply a faulty chip? My temps are a bit better than on my old cooler, but not really satisfactory. I also only have 1 intake fan, so maybe the cooler's struggling to breathe slightly, but not sure.

[DreamCat04]

2 hours ago, Polaria said:

Hello all,

 

I recently purchased an allegedly brand new (as the eBay seller claimed) i3-9350KF basically for the hell of it, and decided I'd pick up a cooler that's probably a bit overkill for this 91 watt processor (I figured I may as well see if I could push this thing to the moon). Installation wasn't hard, it came with their decent looking TF7 paste which was a little annoying to apply but I think I did it right, and the whole setup looks pretty slick in my case even if it's in the wrong color. Unfortunately, I don't seem to be getting anywhere near the performance that reviews across the internet promised for the PA120 - despite only drawing about 90-95 watts with AVX (but not AVX2) enabled on Prime95, I skyrocket all the way to 100 degrees within a couple of minutes and thermal throttle. Even just benching on CPU-Z shoots me to about 77-80 degrees.

 

This seems unusual. I've read a couple of reviews stating that this cooler can handle 200+ watt intel CPUs @ about 75-80 degrees celsius (at my room temperature), and Tom's Hardware had a graph that I calculated for my temps to be probably anywhere around 56-60 degrees under load (their 95 watt TDP graphs). Even when ran at stocks clocks and similar voltages to my i3-9100F it exhibits odd behavior, both with my stock cooler (which it ran on for a few days) and the PA-120, such as weird core temperature variance (core temp differences will be 35 degrees or greater sometimes between cores), very sharp spikes when under sudden load such as from 38 degrees at idle straight to 80 degrees within 1-2 seconds, and just generally what feels like lackluster performance. A lot of these behaviors don't seem to be present on my other Coffee Lake CPU, even at similar voltages.

 

What could the issue here be? Did I maybe overtighten the cooler, misapply thermal paste, etc or could there be something else wrong, like dried out/simply insufficient TIM under the IHS or simply a faulty chip? My temps are a bit better than on my old cooler, but not really satisfactory. I also only have 1 intake fan, so maybe the cooler's struggling to breathe slightly, but not sure.

An i3 has way fewer cores, so those 90W are spread across just a few cores, which is not a lot of surface area. 90W on a quad-core would be comparable to like 200W on modern CPUs. It's really power dense like that, there's probably not a lot you can do except delidding and direct-die cooling (if there's anything out there for direct-die cooling 9th gen CPUs)

[DoctorNick]

2 hours ago, Polaria said:

Hello all,

 

I recently purchased an allegedly brand new (as the eBay seller claimed) i3-9350KF basically for the hell of it, and decided I'd pick up a cooler that's probably a bit overkill for this 91 watt processor (I figured I may as well see if I could push this thing to the moon). Installation wasn't hard, it came with their decent looking TF7 paste which was a little annoying to apply but I think I did it right, and the whole setup looks pretty slick in my case even if it's in the wrong color. Unfortunately, I don't seem to be getting anywhere near the performance that reviews across the internet promised for the PA120 - despite only drawing about 90-95 watts with AVX (but not AVX2) enabled on Prime95, I skyrocket all the way to 100 degrees within a couple of minutes and thermal throttle. Even just benching on CPU-Z shoots me to about 77-80 degrees.

 

This seems unusual. I've read a couple of reviews stating that this cooler can handle 200+ watt intel CPUs @ about 75-80 degrees celsius (at my room temperature), and Tom's Hardware had a graph that I calculated for my temps to be probably anywhere around 56-60 degrees under load (their 95 watt TDP graphs). Even when ran at stocks clocks and similar voltages to my i3-9100F it exhibits odd behavior, both with my stock cooler (which it ran on for a few days) and the PA-120, such as weird core temperature variance (core temp differences will be 35 degrees or greater sometimes between cores), very sharp spikes when under sudden load such as from 38 degrees at idle straight to 80 degrees within 1-2 seconds, and just generally what feels like lackluster performance. A lot of these behaviors don't seem to be present on my other Coffee Lake CPU, even at similar voltages.

 

What could the issue here be? Did I maybe overtighten the cooler, misapply thermal paste, etc or could there be something else wrong, like dried out/simply insufficient TIM under the IHS or simply a faulty chip? My temps are a bit better than on my old cooler, but not really satisfactory. I also only have 1 intake fan, so maybe the cooler's struggling to breathe slightly, but not sure.

 

Try taking the cooler off and see how your mount was?

[Polaria]

1 hour ago, DreamCat04 said:

An i3 has way fewer cores, so those 90W are spread across just a few cores, which is not a lot of surface area. 90W on a quad-core would be comparable to like 200W on modern CPUs. It's really power dense like that, there's probably not a lot you can do except delidding and direct-die cooling (if there's anything out there for direct-die cooling 9th gen CPUs)

There is! From a company called RockIt Cool (though it might be hard to FIND one - not sure they sell them on their website anymore). Would direct-die cooling a 9th gen overclockable i3 be overkill? Yeah probably. But also it'd be hilarious so I might do it

 

What do you think the temp differences might be like? Should I go LM or just use paste? And what kind of compatibility would a direct die kit have? Can I just slap any old air cooler ontop? Never done it before

 

1 hour ago, DoctorNick said:

Try taking the cooler off and see how your mount was?

Will do, but that'll be tomorrow cause it's like 4 AM where I live. Will update with pictures

[DreamCat04]

4 hours ago, Polaria said:

Even when ran at stocks clocks and similar voltages to my i3-9100F it exhibits odd behavior, both with my stock cooler (which it ran on for a few days) and the PA-120, such as weird core temperature variance (core temp differences will be 35 degrees or greater sometimes between cores), very sharp spikes when under sudden load such as from 38 degrees at idle straight to 80 degrees within 1-2 seconds, and just generally what feels like lackluster performance. A lot of these behaviors don't seem to be present on my other Coffee Lake CPU, even at similar voltages.

I didn't see that section initially. 35 degrees between cores is quite a lot. Maybe the TIM under the heatspreader is really old? (though if I remember correctly 9th gen CPUs should be soldered and technically shouldn't go bad). I had some similar behavior where my laptop's CPU would have two cores pinned to 95°C while the others were in the low 80s. Turns out that there was an oxidized spot in the factory-applied LM. Re-spreading it fixed that huge temperature difference. Personally, I don't think a bad mount should cause such huge imbalance, it should just raise the temperature in general (that's just my thinking, it may be different in reality, idk)
Though now that I thought of it a little more, I remember testing an old PC that relatives gave to us as they no longer needed it. It had one of the small stock heatsinks cooling a 4th gen i5. This cooler could easily do the 87W that this quad-core CPU drew under prime95 combined with FurMark PASSIVELY for a few minutes. There's definitely some contact issue somewhere in your case, I no longer think that power density is the main problem

Edited May 20 by DreamCat04

[podkall]

11 minutes ago, DreamCat04 said:

I don't think a bad mount should cause such huge imbalance, it should just raise the temperature in general (that's just my thinking, it may be different in reality, idk)

Bad mount can leave spots with more and less paste, and if those spots aren't connecting the cooler with CPU, the heat has to travel sideways through IHS/thermal paste to reach the connected cross point. Which can be on top of another core.