• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

is Intel stock cooler really this bad?

Associate
Joined
11 Dec 2013
Posts
4
So I just got a new cpu, and decided for now to wait a while before I enter the watercooling world. As such i wanted to see what my cpu was made of with everything stock just to see if there's nothing wrong.

So I ran Prime95 on the standard setting (not the one that generates extra heat) set my system to high performance (idle temps go from 30 to 40C) and then engaged the torture test. to my shock the temperature goes up to 96C. with the cpu fan at 100%.

The heatsink is seated properly, so that's not the problem, the side of my case is open as I'm measuring for future watercooling parts and where to possibly put them.

But really is the intel cooler that bad, that it can't even cool decently on stock? I wasn't expecting much, more around 85C at most, but 96-98C makes me worry.

Now I've been told that as long as it doesn't throttle that this is fine!? What is this throttling based on anyway, which temperature does it take? I've come accros 3 different temperatures for my cpu, AI suite from asus reports 69C while hardware monitor reports 84C under the motherboard section for cpu, and it reports 96C when looking at the cpu section. On what temperature would the throttling be based? If it's based on the 69C for the motherboard it won't throttle ever. and there's a 27C delta between that value and the cpu.

Even idling those temperatures can't seem to agree, in balance power mode it reports 26C (AI Suite), 28C (motheboard cpu temp), 32C (cpu temp).
 
What CPU is it? I'm guessing Haswell?

And yes, the stock sink is bad. Don't stress test with it, it's not made for that. It's the bare minimum to keep your PC running with adequate airflow.

Throttling for haswells happens at 95c, then it reduces clocks and voltage to stay under that. It's based on the "package"-temp, or core temps. Package is the heatspreader I think where the sink sits on, usually same as average temp between cores.
 
haha sorry my mistake, yes it's a 4770k Haswell, pretty sure I put that at first, maybe accidentally erased it.

Well I didn't notice any throttling at 96C (i saw it got to 98C even). but from what you say I suppose it's alright then, just means I have to get my watercooling setup sooner...

How good is the standard thermal paste that Intel applies to the heatsink?

Just as an passing thought, but I hear people shouting delid it everywhere. This voids the warranty right even with a Intel Performance Tuning Protection Plan right?
 
Stock cooler isn't that great, but Prime95 on high-heat on Haswell will hit 100 degrees and start the chip throttling even on a good air cooler. Prime95 is not a typical workload, and doesn't simulate anything you'll do in real life. It's like taking your car to a ten mile dragstrip and then equating that to your normal everyday driving.

Even with watercooling, you'll have to delid and use Liquid Metal to get really good temps out of Haswell under Prime95 loads. There's tons of stuff about it all over the forum if you do a bit of searching.
 
Pretty much par for the course with the 4770K unless you go high end cooling (air or water or better). The intel one while they are never amazing isn't really any worse than an average off the shelf aftermarket cooler - they'll all quickly hit 95C with the 4770K under torture testing.
 
Stock cooler isn't that great, but Prime95 on high-heat on Haswell will hit 100 degrees and start the chip throttling even on a good air cooler. Prime95 is not a typical workload, and doesn't simulate anything you'll do in real life. It's like taking your car to a ten mile dragstrip and then equating that to your normal everyday driving.

Even with watercooling, you'll have to delid and use Liquid Metal to get really good temps out of Haswell under Prime95 loads. There's tons of stuff about it all over the forum if you do a bit of searching.

I understand, however I didn't set it to the high heat setting, just the basic test really. i personally think 40C on idle is way to hot, my laptop with crappy cooler idle's at 32C (high performance setting) and even under prime95 loads my laptop does not go beyond 93C and also doesn't throttle. So I was frankly expecting similar or better results from a much bigger cooler, even though the TDPs are different.

Pretty much par for the course with the 4770K unless you go high end cooling (air or water or better). The intel one while they are never amazing isn't really any worse than an average off the shelf aftermarket cooler - they'll all quickly hit 95C with the 4770K under torture testing.

With or better I suppose you refer to those Tec coolers? I personally think water cooling with a rad space of 7x120 should be more than sufficient for my cpu, the xcrosschill from asus (if i'll even put that in the loop) and my gpu. I can always add another 4x120 rad space, but i think that'd be overkill anyway (I expect at most 400 watt of heat load). but that's for an other time/topic.


It just comes down to me getting worried, I mean my old desktop cpu (and I mean old :P) didn't get remotely close to any of these temps on a stock cooler. I know it's not an indication of real world scenario's, just never thought they'd get that hot really.
 
Stock cooler isn't that great, but Prime95 on high-heat on Haswell will hit 100 degrees and start the chip throttling even on a good air cooler. Prime95 is not a typical workload, and doesn't simulate anything you'll do in real life. It's like taking your car to a ten mile dragstrip and then equating that to your normal everyday driving.

Even with watercooling, you'll have to delid and use Liquid Metal to get really good temps out of Haswell under Prime95 loads. There's tons of stuff about it all over the forum if you do a bit of searching.

This. It's synthetic and won't replicate ANYTHING you'll use the machine for. Not unless your computer gets hacked by Skynet and it proceeds to try and solve the mysteries of the universe.

Just use the machine and check temps with Core Temp every now and then.
 
The stock intel cooler gives me load temps (prime95) never higher than 73degrees if i remember correctly. I'm using a i7 3770 at stock speeds.
Having owned a 3770k and now a 4770k, (both at 4.5ghz). The 4770k is in a different class for heat. My 3770k maxed at 81c in intel burn test on 1.34v. My 4770k in sig hits 94c under p95, (1.265v) which isnt as severe as burn test. Both are known hot running chips but hw is a good deal worse than ib. Best way is to just not use synthetic tests unless you delid and stick some better paste on the die. Liquid pro/ultra, any other paste is a waste of time.
 
OP,

just ditch the stock cooler and grab something big and juicy to satisfy the 4770K badboy. Like everyones mentioned those stress tests are not real world usages. Even 8 pack now mentions quite regularly just use the pc as normal. But if you must stress , grab yourself a decent cooler bro.
 
OP,

just ditch the stock cooler and grab something big and juicy to satisfy the 4770K badboy. Like everyones mentioned those stress tests are not real world usages. Even 8 pack now mentions quite regularly just use the pc as normal. But if you must stress , grab yourself a decent cooler bro.

Not quite sure I'd agree with that advice if I'm reading it right - massively increases the chance of ending up with for instance HDD corruption resulting in an avoidable wipe and reinstall even though torture tests aren't catch alls.
 
I understand, however I didn't set it to the high heat setting, just the basic test really. i personally think 40C on idle is way to hot, my laptop with crappy cooler idle's at 32C (high performance setting) and even under prime95 loads my laptop does not go beyond 93C and also doesn't throttle. So I was frankly expecting similar or better results from a much bigger cooler, even though the TDPs are different.

The problem has already been described a lot in the other Haswell threads. It's a small, dense chip, with poor thermal grease between it and the integral heat sink. So the problem isn't just the type of cooler you use, it's getting the heat from the chip, through the TIM, and onto the IHS where a big heatsink can actually draw the heat away.

This is why so many people are delidding for cool temps at high clocks. No matter how good your heatsink is, it won't work well if you can't get the heat from the chip to the heatsink properly in the first place.
 
Last edited:
The thick epoxy glue that holds the ihs on doesnt help either. Even high end custom water cooling struggles if you get a chip thats been badly glued/pasted.
 
Not quite sure I'd agree with that advice if I'm reading it right - massively increases the chance of ending up with for instance HDD corruption resulting in an avoidable wipe and reinstall even though torture tests aren't catch alls.

Hi

Really my points were to ditch the stock cooler and get something a bit beefier for the 4770K. :p
Though using stress testing is useful when checking the stability of an overclock and voltage settings etc , but its not the end all be all. But if we must, have adequate cooling solutions in place.
 
I have prolimatech megahalems on my 4770k, stress tested it and max temps I got were around 70, with 2 pwm fans / turbo mode ( 800+ rpm ). But you could get a h80 for same money if you like clc.
 
I have prolimatech megahalems on my 4770k, stress tested it and max temps I got were around 70, with 2 pwm fans / turbo mode ( 800+ rpm ). But you could get a h80 for same money if you like clc.

I've read that a H110 can barely handle a Haswell when overclocked, so a H80 is probably a bad buy.
 
Last edited:
I have prolimatech megahalems on my 4770k, stress tested it and max temps I got were around 70, with 2 pwm fans / turbo mode ( 800+ rpm ). But you could get a h80 for same money if you like clc.

Stress tested it using what software/settings/clocks/voltage?
 
Last edited:
Back
Top Bottom