Question Uneven temps when under max load

[mantinziza]

I'm currently testing my overclocks and trying to find my best numbers but I noticed that when my computer is under a full load when using prime 95, I start to thermal throttle

I get

core0@ 68 C it never goes above
core1@ 95 C never goes above that
core2@ 105C which is causing the throttling but I am stable, and am having no issues, again this is at full load
core3@87C and never above that
Package @ 105C right with core 2

at idle my cores never get above 57C, and under normal load all of them are within 5 C of each other in the 80-90C range

It's an old I7-3770k on z97 OC formula

I am well aware that I am extremely pushing this equipment, and I have seen uneven temps in this same pattern (core 0,1,2,3 wise) on my old rig, but not to this extreme is this just something with the hardware design of the chips?

Ive reseated my coolers multiple times even used a micro gram torque wrench to make sure my screw pressure is all equal so my cooler is definitely flat and level and I know i seated my thermal material right.

I'm not super concerned because I never reach these temps under normal load but, I can't help but wonder if I'm missing something, or just being paranoid.

Thanks in advance

 

[TJ Hooker]

If you've reseated the cooler multiple times and there's always the same one or two cores that are significantly hotter, it's probably a result of the TIM (thermal interface material) between the CPU die and heatspreader. Either it dried out or wasn't applied evenly such that the die are associated with those cores isn't making good contact with the heatspreader. I had the same issue with my 6700k, although the temperature delta wasn't nearly as bad as what you're seeing. If this was a newer CPU you could possibly RMA it but I think your only option would be to delid the CPU and replace the TIM.

 

[mantinziza]

My rig is delidded, and I have direct contact with copper/liquid cooler installed. The thermal material cant be dried, I just reapplied last week (thermal grizzly), I guess I could try to check and make sure that I have even pressure again, but considering the tools I used and the triple double quadruple checks, i just really don't know.

strange part is when I'm using it with anything other than prime 95, core 1 is my hottest core, not 2. and only by 2 or 3C at best.
Which makes me really wonder it's just the way the chip is when under prime 95's massive loads.

I am pushing 4.8 ghz on each core, 4.9 with speed step, so I'm not entirely surprised, it's jsut hard to tell if it's jsut the way the chip is, or if something isn't seated right, thanks for your input

 

[TJ Hooker]

Hmm, interesting. Maybe it is related to the structure of the chip somewhat. Given that cores 1 and 2 are in the middle of the die and therefore have other cores on both sides of them also generating lots of heat, it might make sense that they'd be the hottest. I wouldn't have thought it'd be that much of a difference though.

 

[mantinziza]

Hmm, interesting. Maybe it is related to the structure of the chip somewhat. Given that cores 1 and 2 are in the middle of the die and therefore have other cores on both sides of them also generating lots of heat, it might make sense that they'd be the hottest. I wouldn't have thought it'd be that much of a difference though.

That truly makes sense, Ive been useing this same processor for a long time, and I have multiple cooling rigs hooked to it over it "evolution". And the "heat spread pattern",
(I guess that's the right way to say it, lol). Has always been the same between the 4 cores.
Finally multitudes of years of trying to figure this thing out, and I feel like I've finally come to a conclusion. It's just the processor.

When I had my old air cooling rig/s, I let it hit 105C as peak under normal loads for years+, never ruined it or fried it, I even let it run for a week 1 time, while it thermal throttled at 105C down to 3.3 ghz from 4.2, lol

And it was stable, and ran great, and that was with the lid and the crappy stock paste. And the 105C was on core 2 with the same heat spread. Awesome, thanks again. Nice to finally have an answer.