Overheating G3258

[Big_Cazza]

I have been stuffing around with my g3258 and i am having a bit of bad luck with temperatures, using a 360mm aio unit temps were at 75c under prime95. the cpu is clocked at 4.5ghz and uses 1.33v, any ideas why it is overheating? i plan to delid it soon.

[TrigrH]

 

It's not.

[Big_Cazza]

i feel like it should run cooler than this tho, considering it was at 80c with a hyper 212x

[TrigrH]

6 minutes ago, Big_Cazza said:

i feel like it should run cooler than this tho, considering it was at 80c with a hyper 212x

under prime95 at 1.33v? i doubt that

 

btw if you don't quote me I wont see it.

[Froody129]

Those reasonably good temps actually. You are running a high voltage so it can be expected

[JoostinOnline]

75C on Prime95 is a really good temperature.  Don't delid your CPU for that.  Maybe if it was reaching 90C+, but not 75C.  Hell, 80C was great.  Prime95 doesn't represent realistic temperatures.  It's just to see if your PC will crash.

[Big_Cazza]

37 minutes ago, JoostinOnline said:

75C on Prime95 is a really good temperature.  Don't delid your CPU for that.  Maybe if it was reaching 90C+, but not 75C.  Hell, 80C was great.  Prime95 doesn't represent realistic temperatures.  It's just to see if your PC will crash.

its not really that good considering that my pump is at 100 percent and the fans are also close to it, i would like to bring down temps because i know it will do 4.7ghz at 1.42 volts and ive seen lots of haswell chips do that voltage without getting too hot

[JoostinOnline]

2 minutes ago, Big_Cazza said:

its not really that good considering that my pump is at 100 percent and the fans are also close to it, i would like to bring down temps because i know it will do 4.7ghz at 1.42 volts and ive seen lots of haswell chips do that voltage without getting too hot

A Haswell CPU might be able to do 1.42V if you were using liquid nitrogen or something, and even then it would cause serious damage to the CPU.  Going above 1.3V on Haswell is a bad idea. 

 

I don't think you understand how stress tests work.  The fans are supposed to be 100% during those.

[Big_Cazza]

Just now, JoostinOnline said:

A Haswell CPU might be able to do 1.42V if you were using liquid nitrogen or something, and even then it would cause serious damage to the CPU.  Going above 1.3V on Haswell is a bad idea. 

 

I don't think you understand how stress tests work.

in games it hovers around mid 50s is that good?

[JoostinOnline]

2 minutes ago, Big_Cazza said:

in games it hovers around mid 50s is that good?

Yes.  That's 40C less than when the chip starts thermal throttling to lower temps.  Mine is around 60-65C when playing games.

[Big_Cazza]

Just now, JoostinOnline said:

Yes.  That's 40C less than when the chip starts thermal throttling to lower temps.  Mine is around 60-65C when playing games.

so realistically as long as it runs under 100 in stress tests its fine for general use, so i could probably put it back to the 4.6ghz oc that runs 1.37 volts

 

[JoostinOnline]

Just now, Big_Cazza said:

so realistically as long as it runs under 100 in stress tests its fine for general use, so i could probably put it back to the 4.6ghz oc that runs 1.37 volts

 

It should run under 90C at stress tests for stability.  Usually people push to see what they can get at ~85C.

[Big_Cazza]

Just now, JoostinOnline said:

It should run under 90C at stress tests for stability.  Usually people push to see what they can get at ~85C.

ok, should i try and push for just core clock or a bit of cache as well

 

[JoostinOnline]

Just now, Big_Cazza said:

ok, should i try and push for just core clock or a bit of cache as well

 

I don't recommend increasing the voltage.  You're already above recommended levels.

[Big_Cazza]

Just now, JoostinOnline said:

I don't recommend increasing the voltage.  You're already above recommended levels.

ok fair enough, i think my chip lost the sillicon lottery, im probably gonna just get a 4690k when i can afford but i live in australia so everything is overpriced, like an i5 brand new goes for like 370 bucks

 

[ApolloX75]

2 minutes ago, JoostinOnline said:

I don't recommend increasing the voltage.  You're already above recommended levels.

Haswell can take a lot more than you think as long as temps are kept under control. And by under control I mean well under 85C in a heavy stress test over a good length of time.

 

My 4670k for example ran 4.7 at 1.45v for months on end with no degradation, but my water loop kept it well below 70C even running prime 95.

[Big_Cazza]

2 minutes ago, ApolloX75 said:

Haswell can take a lot more than you think as long as temps are kept under control. And by under control I mean well under 85C in a heavy stress test over a good length of time.

 

My 4670k for example ran 4.7 at 1.45v for months on end with no degradation, but my water loop kept it well below 70C even running prime 95.

considering that its like a 100 dollar chip i want to just see what it can do, i might delid and push it hell far after i have a 4690k ready to swap in

[JoostinOnline]

2 minutes ago, ApolloX75 said:

Haswell can take a lot more than you think as long as temps are kept under control. And by under control I mean well under 85C in a heavy stress test over a good length of time.

 

My 4670k for example ran 4.7 at 1.45v for months on end with no degradation, but my water loop kept it well below 70C even running prime 95.

How do you know there was no degradation?

 

Also, it really the luck of the draw.  Not everyone is as lucky as you.  Even the idle temps start to spike at 1.3V on my 4790K.

[Big_Cazza]

2 minutes ago, JoostinOnline said:

How do you know there was no degradation?

 

Also, it really the luck of the draw.  Not everyone is as lucky as you.  Even the idle temps start to spike at 1.3V on my 4790K.

what sort of frequency did you get at 1.3? i heard that haswell refresh tends to be slightly better with overclocking than the original haswell cpus

[ApolloX75]

6 minutes ago, JoostinOnline said:

How do you know there was no degradation?

 

Also, it really the luck of the draw.  Not everyone is as lucky as you.  Even the idle temps start to spike at 1.3V on my 4790K.

Guess so. My 4790k can hit 5.0 at 1.35v but temps outpace my current cooling setup. 

 

Soon. Soon I will have that fixed.

[Big_Cazza]

Just now, ApolloX75 said:

Guess so. My 4790k can hit 5.0 at 1.35v but temps outpace my current cooling setup. 

 

Soon. Soon I will have that fixed.

5ghz thats a lucky chip

[JoostinOnline]

2 minutes ago, ApolloX75 said:

Guess so. My 4790k can hit 5.0 at 1.35v but temps outpace my current cooling setup. 

 

Soon. Soon I will have that fixed.

Damn, you really won the silicon lottery.  All I can get is 4.6GHz at 1.3V without crashing.

[Big_Cazza]

2 minutes ago, JoostinOnline said:

Damn, you really won the silicon lottery.  All I can get is 4.6GHz at 1.3V without crashing.

if i put 1.42 volts in my cpu and temps stay under 85 how long do you think the chip will last?

[JoostinOnline]

Just now, Big_Cazza said:

if i put 1.42 volts in my cpu and temps stay under 85 how long do you think the chip will last?

I don't know.  I can't imagine your temps will stay under 85C at that voltage though, unless you have a badass custom cooling loop.  You should bump up voltage by 0.01 at a time.

[Big_Cazza]

Just now, JoostinOnline said:

I don't know.  I can't imagine your temps will stay under 85C at that voltage though, unless you have a badass custom cooling loop.  You should bump up voltage by 0.01 at a time.

i dont have a custom loop but i have a deepcool genome rog edition that has a built in 360mm aio, surely its worth a try, if i can at least squeeze 4.7ghz out of it