i7 8700 reaching 100c

[Bennedict 929]

hello, i just upgraded my rig from i5 6500 to i7 8700. the cpu is working like normal, but when i stress test the cpu it went to 100c pretty quickly. i thought my cooler (thermaltake contac silent 12) wasn't enough, so i bought a Cryorig H5 (which should be more than enough for cooling an i7 8700). after installation, temp didn't change under load. It's still 100c under load and is throttling. when i touch the cooler, it's not as warm as my old i5 6500 (with the same contac silent 12) i've tried reapplying the thermal paste (mastergel pro) many times, with good coverage, but it didn't improve the temps a bit. Any idea what's wrong?

 

[J_E_D_70]

What motherboard?

What is the vCore under load?

Edit: whats the rpm on the cpu fan?

 

[Bennedict 929]

What motherboard?

What is the vCore under load?

Edit: whats the rpm on the cpu fan?

the motherboard is asrock z370 killer sli
vcore is 1.399v (completely stock)
fan rpm is around 1350 (cryorig h5)

 

[J_E_D_70]

That is way, way too much vcore for a stock 8700 (assuming you meant non-k).

Did the bios allow you to select something like multicore enhancement? That will crank the vcore up for no good reason.

I’m at 4.7 all core with 1.26. If you are fully stock you should probably be down at 1.2 or so. Check your manual on how to manually reduce vcore.

Edit: what stress test are you using? Prime95 after 26.6 puts an unreasonable load (for a gaming system) on the cpu with AVX instructions.

FWIW, my H5 stays under 60c at full gaming load so you should be under that.

 

[Bennedict 929]

That is way, way too much vcore for a stock 8700 (assuming you meant non-k).

Did the bios allow you to select something like multicore enhancement? That will crank the vcore up for no good reason.

I’m at 4.7 all core with 1.26. If you are fully stock you should probably be down at 1.2 or so. Check your manual on how to manually reduce vcore.

Edit: what stress test are you using? Prime95 after 26.6 puts an unreasonable load (for a gaming system) on the cpu with AVX instructions.

FWIW, my H5 stays under 60c at full gaming load so you should be under that.

i'm using the aida64 extreme fpu test

 

[J_E_D_70]

Probably not a good test:
http://www.tomshardware.com/answers/id-3159716/4790k-aida64-cpu-fpu-test-showing-100-degrees.html

Try OCCT large or medium datasets. If those are reasonable temps then you’re good. The small datasets will also punish the cpu if you just want to see how it compares to AIDA.

No matter, that vCore needs to be reduced. If

 

[Bennedict 929]

Probably not a good test:
http://www.tomshardware.com/answers/id-3159716/4790k-aida64-cpu-fpu-test-showing-100-degrees.html

Try OCCT large or medium datasets. If those are reasonable temps then you’re good. The small datasets will also punish the cpu if you just want to see how it compares to AIDA.

No matter, that vCore needs to be reduced. If

i didn't find anything like multicore enhancement in the bios. so should i set the voltage manually? at what load line calibration?

 

[J_E_D_70]

You can leave LLC at auto and if it keeps bumping up the voltage too much can set it in thr middle and tweak from there.

Ya - manual at like 1.25 and stress thrn start working down til not stable then up one .01v and all set.

 

[Bennedict 929]

You can leave LLC at auto and if it keeps bumping up the voltage too much can set it in thr middle and tweak from there.

Ya - manual at like 1.25 and stress thrn start working down til not stable then up one .01v and all set.

thanks. i got it at 1.150 stable now with max temp around 83c

 

[J_E_D_70]

1.15 is great voltage but an h5 should be way cooler than that (your temps are still safe tho).

Do you have a fan or two drawing air in and at least one as exhaust in the back?

What is your ambient room temp?

 

[Bennedict 929]

1.15 is great voltage but an h5 should be way cooler than that (your temps are still safe tho).

Do you have a fan or two drawing air in and at least one as exhaust in the back?

What is your ambient room temp?

i'm using 1.2 currently because peak power issues
i have 3 intake fans and 1 exhaust
my ambient is around 27c

 

[J_E_D_70]

I’d peg your max temp at 65c then since you’re 5c - 6c hotter than my ambient.

Could be a faulty sensor I suppose. If not, your H5 isn’t on quite right. Did you slowly tighten one side then the other, back and forth, until snug? Use a pea-sized amount of paste (or a thin X shape)?

 

[Bennedict 929]

I’d peg your max temp at 65c then since you’re 5c - 6c hotter than my ambient.

Could be a faulty sensor I suppose. If not, your H5 isn’t on quite right. Did you slowly tighten one side then the other, back and forth, until snug? Use a pea-sized amount of paste (or a thin X shape)?

yes, i did tighten both side evenly
i used pea sized paste with mastergel pro
with the latest edition of prime95, the maximum heat mode,
max temp is 75 with 1.25 vcore
i think i will delid later

 

[J_E_D_70]

Ok - you are good at 75 if you’re doing p95 burn. I wouldnt delid as you should not see anything that hot in real world use.

 

[Bennedict 929]

Ok - you are good at 75 if you’re doing p95 burn. I wouldnt delid as you should not see anything that hot in real world use.

if i delidded my cpu is there a lot of temp difference between using liquid metal and thermal paste? i read that the biggest problem is the silicon thickness used between cpu and ihs not the TIM.

 

[J_E_D_70]

There could be 10-15c difference but since you’re well under the throttle point I wouldn’t risk it.

 

[Bennedict 929]

There could be 10-15c difference but since you’re well under the throttle point I wouldn’t risk it.

i mean, the temp difference if i used thermal paste again (mastergel maker nano) vs conductonaut between the die and ihs after delidding

 

[J_E_D_70]

That’s what I mean .

Edit: from SL but I dont believe anyone keeps them long enough to make a real workd difference;

Peak core temperatures under a heavy overclocked load typically decrease anywhere from 10°C to 20°C when delidded, ensuring greater stability and a longer component lifespan.

https://siliconlottery.com/collections/all/products/delid

 

[Bennedict 929]

That’s what I mean .

Edit: from SL but I dont believe anyone keeps them long enough to make a real workd difference;

Peak core temperatures under a heavy overclocked load typically decrease anywhere from 10°C to 20°C when delidded, ensuring greater stability and a longer component lifespan.

https://siliconlottery.com/collections/all/products/delid

does that mean there will be no temperature drops if i delidded my cpu and use thermal paste instead of liquid metal between? i will also clean the silicon so that the gap between cpu die and ihs will decrease

 

[J_E_D_70]

No idea. Sorry! Either way there’s risk for no real life gain.

 

[Bennedict 929]

No idea. Sorry! Either way there’s risk for no real life gain.

update: turns out i read the wrong voltage. 1.39 is the VID and the vcore is actually 1.28 max on stock

 

[Demonic Heart]

No idea. Sorry! Either way there’s risk for no real life gain.

update: turns out i read the wrong voltage. 1.39 is the VID and the vcore is actually 1.28 max on stock

i had a feeling you read it wrong lol it happens a lot

what is now your current temps?