CPU temps too high?

[El_Mayo]

If you had a different chip to test with I'd say do that. Otherwise, if it persists, you might want to RMA the CPU.

Yeah it definitely feels like something's up with the sensor, if you have a look at my screenshot above, compared to what it's doing now, also unlike the above post the chip is actually running at 4.4ghz

 

[GorbazTheDragon]

Test with stock settings, if it's unusually high see what you can do with intel for a replacement.

 

[El_Mayo]

Test with stock settings, if it's unusually high see what you can do with intel for a replacement.

This is completely stock settings. I'm going to try uninstalling AI Suite III, that's the only thing that's different between today and the previous screenshot

 

[GorbazTheDragon]

OK, probably either a sensor issue or a crappy chip with a LOT of leakage.

 

[El_Mayo]

OK, probably either a sensor issue or a crappy chip with a LOT of leakage.

how do u mean leakage? even so I don't know why it's only like this now, but was okay two days ago so I'm confused... load temps when gaming have been the same tho, around 60-65, which I guess is the important thing but I don't want the full load to be ridiculously high cos I can't stress test my overclocks

 

[GorbazTheDragon]

All chips have some leakage current through the transistor gates and the insulators, due to the small scale, some chips leak more than others and therefore get hotter under higher loads, could be that your chip has unusally high leakage on certain parts of it, but if your temperatures in games and any of your normal tasks are not unusual, I would not worry too much. If you are going to be doing a lot of video rendering or something like that I guess it might become a problem.

Prime95 represents an absolute worst case scenario temperature wise, and even if you run into throttling while running prime95, in many cases you can be alright under normal circumstances.

 

[El_Mayo]

All chips have some leakage current through the transistor gates and the insulators, due to the small scale, some chips leak more than others and therefore get hotter under higher loads, could be that your chip has unusally high leakage on certain parts of it, but if your temperatures in games and any of your normal tasks are not unusual, I would not worry too much. If you are going to be doing a lot of video rendering or something like that I guess it might become a problem.

Prime95 represents an absolute worst case scenario temperature wise, and even if you run into throttling while running prime95, in many cases you can be alright under normal circumstances.

I will be editing and rendering tho

how do I test it's the sensors before RMAing the chip? after all the chip was hitting 78 degrees in the bios a few days ago... then all of a sudden was fine (34 degrees)

 

[GorbazTheDragon]

Run it through your rendering workload, if that is problematic you will want to dig deeper.

Have you checked your fan speeds? Does it spin up under load, how hot is the top of the fin stack getting? If it's not heating up a lot, you might be looking at a faulty heatsink/heatpipe or bad mount.

4GHz on haswell is a really tall order for the stock heatsink, I was having trouble in P95 with a 3.3GHz i5 4440, so that doesn't really say too much. But if you are hitting 80c idle with the stock HS it does pretty much eliminate any of the faulty heatsink possibilities. In which case it almost certainly is the chip.

 

[El_Mayo]

Run it through your rendering workload, if that is problematic you will want to dig deeper.

Have you checked your fan speeds? Does it spin up under load, how hot is the top of the fin stack getting? If it's not heating up a lot, you might be looking at a faulty heatsink/heatpipe or bad mount.

4GHz on haswell is a really tall order for the stock heatsink, I was having trouble in P95 with a 3.3GHz i5 4440, so that doesn't really say too much. But if you are hitting 80c idle with the stock HS it does pretty much eliminate any of the faulty heatsink possibilities. In which case it almost certainly is the chip.

oh yeah the fan's are alright, they run up to 100% when under load.

Also I'm not using the stock heatsink atm! I was saying how it was reaching 78 degrees in the bios with a stock cooler possibly indicative of a bad sensor.

My idle temps are around 32 degrees atm with my Thermalright but I've done some reading and people are saying not to use Prime95 version 27.9 or newer with Haswell. I'm running AIDA64 stability test at stock settings now and I'm getting more acceptable temps. Having a look at Core Temp right now it's saying the Power drawn is around 100 watts. When I run Prime95 however it gets up to 170W? maybe that's why people on forums say not to use Prime with Haswell?

edit: don't mean to be tedious with these questions, it's just that Aria have a shit policy of charging you up to £20 if they find no fault with your return, as well as the cost of returning the item to you and they won't refund the cost of sending the item to them in the first place if they find no fault

 

[GorbazTheDragon]

At 170w load your temps seem normal. Try re-doing your overclocks with something else for load tests.

What thermal paste are you using?

 

[El_Mayo]

At 170w load your temps seem normal. Try re-doing your overclocks with something else for load tests.

What thermal paste are you using?

Cheap chinese stuff presumably, £1.50 for 30g, everything's been looking good this past week, temps have looked normal, maxing out around 81 under full load with AIDA 64 at 1.24V

 

[Vayra86]

81 degrees at 1.24V is still on the high side, but nothing to worry about. However AIDA is not very intensive, in no way comparable to full throttle Prime95. If you want consistent testing, use consistent testing methods, in other words, go Prime95 again or you don't know anything.

I think you had a pretty bad mount in the first place, or too much paste, or the paste itself is crappy, or a combination of these things (especially when adding in just a pull configuration, which is often less effective than a push on the CPU heatsink and a pull on the rear case fan -> ideal setup for 2 fan cases.

 

[El_Mayo]

81 degrees at 1.24V is still on the high side, but nothing to worry about.

I think you had a pretty bad mount in the first place, or too much paste.

At full load? What kinda temps should I be aiming for? It's more likely it's too much paste, I tried to do a tiny blob

 

[Vayra86]

Full load in AIDA is not the same as full load in OCCT / Prime95 / Intel Burn Test or any other stress application. They are all different and have various testing methods. Prime95 got hotter because of AVX(2) and you always want to use a heavy test like that for stressing. AIDA is nice as long as you're not serious about stability, honestly, or if you only ever test in AIDA and leave some headroom.

1.24V? I would say 75-78 degrees at the very most.

Also, try HWInfo for sensor data, it may provide more insight (and offers literally EVERY single number/sensor)

 

[Aquinus]

I had a Core 2 Duo E6600 with the stock cooler which worked okay for a little while but after re-mounting it 2 times the plastic feet got worn out so it wasn't holding the cooler against the CPU tight enough enabling the CPU to overheat very easily until it got to the point where it wasn't moving enough heat to even get it to boot into Windows.

All in all, replace the cooler first. If the cooler isn't scalding hot to the touch when it claims your CPU is running that hot, then you really need a better CPU cooler. If the heat sink is burning hot, you need more airflow in your case.

All in all: Stop beeting the crap out of your CPU until you get the temperature thing figured out, it should not idle anywhere near 60-80 degrees Celsius, even with a stock cooler. I've seen a 2600k on a stock cooler have incredibly better numbers than this.

 

[Vayra86]

His cooler doesn't have push pins.

People are way too quick at jumping to conclusions. We are at page 2 and OP has yet to complete solid testing rounds with the same application after a good remount...

Test consistently or we can keep going at this for days.

 

[El_Mayo]

Full load in AIDA is not the same as full load in OCCT / Prime95 / Intel Burn Test or any other stress application. They are all different and have various testing methods. Prime95 got hotter because of AVX(2) and you always want to use a heavy test like that for stressing. AIDA is nice as long as you're not serious about stability, honestly, or if you only ever test in AIDA and leave some headroom.

1.24V? I would say 75-78 degrees at the very most.

Also, try HWInfo for sensor data, it may provide more insight (and offers literally EVERY single number/sensor)

When I run Prime95 it hits 95 degrees instantaneously even at stock clocks

I had a Core 2 Duo E6600 with the stock cooler which worked okay for a little while but after re-mounting it 2 times the plastic feet got worn out so it wasn't holding the cooler against the CPU tight enough enabling the CPU to overheat very easily until it got to the point where it wasn't moving enough heat to even get it to boot into Windows.

All in all, replace the cooler first. If the cooler isn't scalding hot to the touch when it claims your CPU is running that hot, then you really need a better CPU cooler. If the heat sink is burning hot, you need more airflow in your case.

All in all: Stop beeting the crap out of your CPU until you get the temperature thing figured out, it should not idle anywhere near 60-80 degrees Celsius, even with a stock cooler. I've seen a 2600k on a stock cooler have incredibly better numbers than this.

I'm not using the stock cooler anymore, and I'm idling around 29-33 degrees atm which is good. It's the LOAD temps that are high

People are way too quick at jumping to conclusions. We are at page 2 and OP has yet to complete solid testing rounds with the same application after a good remount...

Test consistently or we can keep going at this for days.

I can't use Prime 95 as it immediately exceeds 90 degrees, I'd rather not constantly keep using it. Is that unreasonable?

 

[Vayra86]

When I run Prime95 it hits 95 degrees instantaneously even at stock clocks

I'm not using the stock cooler anymore, and I'm idling around 29-33 degrees atm which is good. It's the LOAD temps that are high



I can't use Prime 95 as it immediately exceeds 90 degrees, I'd rather not constantly keep using it. Is that unreasonable?

No of course not, just be careful with jumping to conclusions based on a different test, you saw the difference yourself when using AIDA. A faulty chip is highly unlikely, though it can happen, but really is the last thing to consider. The fact it jumps up immediately but shows a lower increase under AIDA does not point to a faulty sensor, but a bad mount / bad paste / bad airflow or indeed perhaps a faulty cooler. Apply as little paste as humanly possible, the pressure of the cooler will do the work for you.

For reference, a True Spirit 140 should do 85 degrees max up to about 1.27 vCore, and with quality paste may even do better. I would not want to go much higher on that cooler to be safe, 1.296 is pushing it.

Like others have suggested, skip AIDA and use OCCT from now on, it offers AVX and non-AVX testing, u can test one or more cores at a time and it gives you a shitload of graphs to analyze (things like LLC for example, which can drastically affect stability). Combined with HWinfo you've got a pretty solid view on things.

Last question: did you clean the base of the cooler and the IHS after your remount(s)? Paste residue will negatively affect your temps.

 

[El_Mayo]

No of course not, just be careful with jumping to conclusions based on a different test, you saw the difference yourself when using AIDA. A faulty chip is highly unlikely, though it can happen, but really is the last thing to consider. The fact it jumps up immediately but shows a lower increase under AIDA does not point to a faulty sensor, but a bad mount / bad paste / bad airflow or indeed perhaps a faulty cooler.

For reference, a True Spirit 140 should do 85 degrees max up to about 1.27 vCore, and with quality paste may even do better. I would not want to go much higher on that cooler to be safe, 1.296 is pushing it.

Like others have suggested, skip AIDA and use OCCT from now on, it offers AVX and non-AVX testing, u can test one or more cores at a time and it gives you a shitload of graphs to analyze (things like LLC for example, which can drastically affect stability). Combined with HWinfo you've got a pretty solid view on things.

Okay I will get HWinfo and try Oct. There doesn't appear to be anything wrong with my heatsink and I mounted the heatsink pretty tight but I will try a reseat with less paste tonight

 

[Aquinus]

I'm not using the stock cooler anymore, and I'm idling around 29-33 degrees atm which is good. It's the LOAD temps that are high

I didn't catch that. Those are the kinds of idle temps I have. Haswell, like IVB, has paste between the heat spreader and the die itself, so that could be a little bit of what you're seeing. 75-80 isn't unrealistic from everything I've read. I personally don't like bringing my SB-E chip over 70*C but that's a personal preference.

It seems to me that the new cooler did what it was supposed to. High temps could be due to the wrong voltages being bumped for stability. Unfortunately my lack of experience with Haswell CPUs won't be of any help, but this thread might tell you what you want to know moving forward: Intel Haswell Overclocking Clubhouse.

 

[El_Mayo]

I didn't catch that. Those are the kinds of idle temps I have. Haswell, like IVB, has paste between the heat spreader and the die itself, so that could be a little bit of what you're seeing. 75-80 isn't unrealistic from everything I've read. I personally don't like bringing my SB-E chip over 70*C but that's a personal preference.

It seems to me that the new cooler did what it was supposed to. High temps could be due to the wrong voltages being bumped for stability. Unfortunately my lack of experience with Haswell CPUs won't be of any help, but this thread might tell you what you want to know moving forward: Intel Haswell Overclocking Clubhouse.

Yeah the cooler is doing an okay job, it just can't hack the Prime95 load. It seems. Part of me doesn't mind as my gaming load is tolerable, it's just annoying that I can't fully test stability!

reseated the heatsink just now, here's what it looked like.

wiped it all off an reseated with this amount of paste this time

Last question: did you clean the base of the cooler and the IHS after your remount(s)? Paste residue will negatively affect your temps.

forgot to answer, yeah I wipe it all of every time

 

[vega22]

Yeah it definitely feels like something's up with the sensor, if you have a look at my screenshot above, compared to what it's doing now, also unlike the above post the chip is actually running at 4.4ghz

This is completely stock settings. I'm going to try uninstalling AI Suite III, that's the only thing that's different between today and the previous screenshot

that makes no sense, is either 1 or other dude?

Cheap chinese stuff presumably, £1.50 for 30g, everything's been looking good this past week, temps have looked normal, maxing out around 81 under full load with AIDA 64 at 1.24V

sounds like something which would not help in heat transfer

Yeah the cooler is doing an okay job, it just can't hack the Prime95 load. It seems. Part of me doesn't mind as my gaming load is tolerable, it's just annoying that I can't fully test stability!

reseated the heatsink just now, here's what it looked like.

wiped it all off an reseated with this amount of paste this time

forgot to answer, yeah I wipe it all of every time

yea that tim looks like milky crap. get some good tim and try again.

could also just be that you lucked out on the ic loto. better tim under the ihs does not mean better tim application....

 

[El_Mayo]

that makes no sense, is either 1 or other dude?



sounds like something which would not help in heat transfer



yea that tim looks like milky crap. get some good tim and try again.

could also just be that you lucked out on the ic loto. better tim under the ihs does not mean better tim application....

I was speculating whether it was the software that was messing up my temp readings but I've ruled that out now

I don't know what you mean by IC loto. also that's true but I'm at least hoping for slightly better temps seeing as my most recent tim application should be better than my last one which had a load of air bubbles

 

[OneMoar]

I said this at the start replace the cooler that one is a piece of crap
temps are smack where they should be for that cooler

 

[El_Mayo]

I said this at the start replace the cooler that one is a piece of crap
temps are smack where they should be for that cooler

what other cooler would you recommend around the same price tho?