6700K hitting 95C immediately

A

alex_di

n00b
Joined
Mar 15, 2016
Messages
40
I'm looking for direction on how best to deal with a cooling problem with a new system.

I had a 3770K with a 212 Evo cooler with two 120mm fans. Overclocked, it wouldn't break 80C in any test. Room is 72F.

I upgraded to a 6700K and kept the same cooler. After the initial install with a mild (4.4 GHz, 1.25 vcore) overclock, it would slowly rise to 87-88C after a few minutes Prime95. Tearing it down, it appeared I'd used way too much paste; my half-pea had coated both surfaces and seeped out the sides.

I cleaned all that off and installed the 212 again with about 1/4 the paste. Prime95 immediately pushed it to 90C+. The paste had spread to an area the size of the dime. Most of the chip was untouched.

I tried again, this time dabbing paste on the four corners and the center. Same result in Prime 95. Worse, even. I stopped the test immediately as two cores passed 95C. Tearing it down showed five even blobs, though about 20% of the CPU surface remained untouched.

All of these temperatures suck. I've gone over both the CPU and heatsink with a straight-edge and a flashlight and they *seem* to be flat.

I see these options:

* Assume the CPU is warped. Exchange it.
* Assume the heatsink is warped. Buy sandpaper and attempt to smooth it out, or buy a new heatsink.
* Assume they're both warped and do both of the above.

Which of these is most sensible to start? Assuming I don't want to lap the CPU, is lapping the heatsink likely to fix this problem? Or is there an approach or technique here I'm missing?
 
I would stick to the middle blob technique, but just use a very small amount.

Lapping the heatsink will not hurt anything. If it's not warped, it will be flatter. If it is warped, then it'll be better than it was or even fixed.

When you check for flatness with a razor, do you make sure to shift the position of the razor and rotate the CPU? It may only be a very slight dip, and not necessarily in the center. You can also check the heatsink.
 
First of all things, and I always have to make people remember this, DON'T use Prime95 with anything newer than Ivy Bridge (So, Haswell, Haswell refresh = devil's canyon and Skylake), if you want to use prime95 then only use the version 27.7 if not trust me you can even damage your chip. ok now, second, what motherboard are you using? with what kind of settings and more important, are you using Manual Vcore? Auto? Adaptive? offset?. Be always sure that the Vcore remains that way during any stress test and re-do all the test with prime 27.7 to see if the overheating issue are still present.

On the other hand, Put a small pressure with your fingers in the cooler against the CPU (push just slightly the CPU against the mobo) during the Stress test, see how temps behave and check if any difference in temp occur, why this? well few things, not every motherboard PCB is the same Thickness, some manufacturers to cheap out use thinner PCB than most others, and also skylake use a thinner PCB so when you combine a thinner mobo PCB with a thin CPU PCB you may be able to found in a situation where the cooler doesn't make proper contact causing issues with temperatures, so this fast test can help to confirm or dismiss this issue, I know from own experience that this can happen with Cooler master Hyper 212 coolers, as they are typically kinda loose mounted.

Contrary is the case where the Cooler can crate such huge pressure against the CPU that it can literally warp or bend the CPU PCB causing also insufficient contact, so in this case loosening a bit the cooler help to decrease the temperatures.

Also test everything at stock settings and then overclocked to make complete comparisons between stock and overclocked temps.
 
Araxie said:
First of all things, and I always have to make people remember this, DON'T use Prime95 with anything newer than Ivy Bridge (So, Haswell, Haswell refresh = devil's canyon and Skylake), if you want to use prime95 then only use the version 27.7 if not trust me you can even damage your chip. ok now, second, what motherboard are you using? with what kind of settings and more important, are you using Manual Vcore? Auto? Adaptive? offset?. Be always sure that the Vcore remains that way during any stress test and re-do all the test with prime 27.7 to see if the overheating issue are still present.

On the other hand, Put a small pressure with your fingers in the cooler against the CPU (push just slightly the CPU against the mobo) during the Stress test, see how temps behave and check if any difference in temp occur, why this? well few things, not every motherboard PCB is the same Thickness, some manufacturers to cheap out use thinner PCB than most others, and also skylake use a thinner PCB so when you combine a thinner mobo PCB with a thin CPU PCB you may be able to found in a situation where the cooler doesn't make proper contact causing issues with temperatures, so this fast test can help to confirm or dismiss this issue, I know from own experience that this can happen with Cooler master Hyper 212 coolers, as they are typically kinda loose mounted.

Contrary is the case where the Cooler can crate such huge pressure against the CPU that it can literally warp or bend the CPU PCB causing also insufficient contact, so in this case loosening a bit the cooler help to decrease the temperatures.

Also test everything at stock settings and then overclocked to make complete comparisons between stock and overclocked temps.
Click to expand...
That's an interesting point about the heatsink bending the IHS. I wonder if the heatsink is being supported properly given that he would have had to move to a new motherboard. I'm assuming that the PC is standing up. It might be worth testing the CPU with the PC laid on its side so that the heatsink is standing upright.
 
Good call on the razor, I was only using a metal ruler. Both surfaces are concave. The heatsink has one corner that warps concave just a bit; it took a few tries to catch this. With the CPU OTOH, the concavity is consistent across the surface and was immediately evident.

Which is annoying. I'm starting to think I should bring a razor to Microcenter and test the exchange in the store. Unless they're all likely to be terrible. I recall I had to exchange my first 3770K for exactly this problem. It's not the lapping part that concerns me, it's voided the warranty on a CPU I just bought, but I'll do it if that's what takes to get the temperatures out of the stratosphere.

Araxie, I had the voltage set to standard vcore + 0.5 offset. No other changes. The board is a Gigabyte GA-Z170MX-Gaming 5.

Shameless, the system is actually flat now. The case hasn't arrived yet, I have the board set up on an antistatic pad.
 
Last edited:
I feel you on the warranty, I wouldn't lap the CPU either. Couldn't hurt to take a razor with you to microcenter. Not sure if they'd let you check them like that, but if you're returning one for not being flat, it's a fair request.
 
alex_di said:
Araxie, I had the voltage set to standard vcore + 0.5 offset. No other changes. The board is a Gigabyte GA-Z170MX-Gaming 5.
Click to expand...

Careful with the antistatic bag. It can produce more damage than what can protect. I've seen some dead GPUS and motherboards by being placed over antistatic bag.

Now, fast test. Run a prime95 27.7 test with manually fixed vcore and check Temps.

Shameless Liar said:
Not sure if they'd let you check them like that, but if you're returning one for not being flat, it's a fair request.
Click to expand...

It count as defective unit, so totally fair request, non-flat IHS is a common cause of return if the buyer state clearly how it affect the out of the box function without mention any overclock. =)
 
Araxie, I've actually got it on an antistatic mat.

I checked out two replacement CPUs at Microcenter. Both had, to a greater or lesser extent, the same concavity. I bought some plate glass and sanding paper, so the first thing to try is cleaning up the stock base of the heatsink. Then crossing my fingers that's good enough to keep the temperatures down. :/
 
Alright. Heatsink is lapped. 320/400/1000 grit. Not a mirror shine, but smooth to the touch and flat according to my razor.

Mounted on a new 6700K with a small divot of Artic Silver, all cores idle around 20C. Prime95 immediately pushes them to 90C. Temps rise to 93C, then the CPU makes a high-pitched noise and appears (at least according to the varying frequencies in CPU ID) to throttle. Temps drop to 78C. If I stop all the workers and restart, the cycle starts over.

This is really disappointing. The only improvement is that all cores are consistently hot. The previous 6700K had a single core that was 5-7C above the rest.

Incidentally, for all of this testing, the heatsink is barely warm to the touch.
 
Have you tried Intel XTU software instead of Prime95 for stability testing? I'm just curious what temps you might get with that compared to 95.
 
Dude, nothing is perfectly flat. This is why I always roll my eyes at the "pea in the center" purists.

That only works if everything is flat or nearly flat. But every review I've seen the CPUs are slightly concave, so there's a chance you'll never find your dreamboat.

I'd suggest you try applying it to the entire chip surface manually before you start destroying things :D Just a thin layer.
 
I'll try both suggestions. I'm done exchanging things, I didn't see anything that suggested QC variance. They were all bad.
 
Araxie said:
First of all things, and I always have to make people remember this, DON'T use Prime95 with anything newer than Ivy Bridge (So, Haswell, Haswell refresh = devil's canyon and Skylake), if you want to use prime95 then only use the version 27.7 if not trust me you can even damage your chip. ok now, second, what motherboard are you using? with what kind of settings and more important, are you using Manual Vcore? Auto? Adaptive? offset?. Be always sure that the Vcore remains that way during any stress test and re-do all the test with prime 27.7 to see if the overheating issue are still present.
Click to expand...
How is it going to damage the CPU? Are you claiming the Intel thermal protections don't work?

The "magic" IE: most recent version of prime95 that doesn't use AVX/AVX2 instructions is v26.6. 27.7 has AVX instructions, which presumable is what you're trying to avoid.

However, if you're going to use the system with a workload that uses AVX/AVX2 instructions you should be checking for stability and thermal throttling with a utility that uses them. If your workload doesn't use AVX instructions then don't use a utility that uses them to check for thermal throttling or system stability.
 
There's the ROT (that's pretty dang reliable) that for every 10 C higher operating temperature, you're cutting your electronic's effective lifetime in half. Obviously, the most important part of that equation is idle temps, but still, running high temps, even when self-protective mechanisms are in place, is less that ideal. Cool temps are a good thing, even if the lifetimes of a working chip is quite long!

Alex--how many thousandths are we talking here? I suppose you're putting a razor across the diagonals and backlighting it.
 
Derfnofred said:
There's the ROT (that's pretty dang reliable) that for every 10 C higher operating temperature, you're cutting your electronic's effective lifetime in half. Obviously, the most important part of that equation is idle temps, but still, running high temps, even when self-protective mechanisms are in place, is less that ideal. Cool temps are a good thing, even if the lifetimes of a working chip is quite long!
Click to expand...
Well, I wasn't saying that running high temps are desirable or ideal. However, he makes it sound like running a few hour test is going to render the CPU useless / dead in short order. That's certainly not the case.
 
if you are noticing that the cpu is concave, putting a little thicker paste in the center should help fill the gap and improve temps. am I wrong in that thought? Even with AS5 I don't think there is any worry of shorting things out, no where for it to go/ get into. I don't worry about it with my AMD chip...
 
Araxie said:
First of all things, and I always have to make people remember this, DON'T use Prime95 with anything newer than Ivy Bridge (So, Haswell, Haswell refresh = devil's canyon and Skylake), if you want to use prime95 then only use the version 27.7 if not trust me you can even damage your chip.
Click to expand...


I am also curious, why we should not use Prime95? So far it has been excellent overclocking stability testing program. Not only does it stress it enough to see general stability its also useful in detecting small miscalculations which may not cause hard crashes but will make the software behave wierdly.

*edit* Quick googling revealed that there is a bug in Skylake's design that causes it to freeze on certain calculations but that had nothing to do with overclocking or overheating and was fixed with mobo BIOS updates earlier this year. No warnings about Prime95 actually breaking the CPU at all.
 
MaZa said:
I am also curious, why we should not use Prime95? So far it has been excellent overclocking stability testing program. Not only does it stress it enough to see general stability its also useful in detecting small miscalculations which may not cause hard crashes but will make the software behave wierdly.
Click to expand...
He's scared of the AVX/AVX2/FMA2 instructions the new versions of Prime95 use. He doesn't want a system that's actually 100% reliable rock solid while overclocked. He just wants one that doesn't flagrantly crash while running a game or two.

On the other hand, I don't really get what Intel has going on. The same exact cooler on a 115W TDP v1 E5 Xeon (Sandy Bridge-EP) can keep it much cooler under load than a newer generation chip like a Haswell with a lower TDP (even when overclocked), by like 20C. Does soldering of the heatspreader to the die and the larger die size of the older chips really help that much in getting the heat out of the CPU? I guess it must. Seems like Intel is going the wrong way.
 
Check temps at stock BIOS settings, see if things look normal before chasing down this warped CPU theory. Look at the RPM of the fan while doing that, and you should also try adjusting the CPU fan speed in the BIOS. By default, the fan won't be spinning at max RPM, you can manually set the fan to spin at 100% all the time and this will keep temps down.

I think a lot of people are going off the deep end here with the warped CPU ideas. And Prime 95 is great at checking temps and stability, the AVX instruction cause the CPU to bump the voltage slightly which makes it a better stress test.
 
Last edited:
Stereodude said:
On the other hand, I don't really get what Intel has going on. The same exact cooler on a 115W TDP v1 E5 Xeon (Sandy Bridge-EP) can keep it much cooler under load than a newer generation chip like a Haswell with a lower TDP (even when overclocked), by like 20C. Does soldering of the heatspreader to the die and the larger die size of the older chips really help that much in getting the heat out of the CPU? I guess it must. Seems like Intel is going the wrong way.
Click to expand...

Despite the lower TDP, it is much harder to pull the heat out of a much smaller die area effectively...It makes it doubly worse since the Xeon you are referring to has no IGP taking up a big chunk of the die..

AS for the solder/non solder issue, I am firm believer that it helps a TON...I have de-lidded a ton of Intel CPUs and the insane temperature drop speaks for itself..I wish they would just go back to soldering them, but the bean counters wanna save $.02 per unit.
 
An update on this-- I ultimately returned the 6700K in favor of a 5820K. Entirely new platform, this time with a water-cooling kit. Using more heatsink paste definitely helped, but I was still in the high 80s. I haven't assembled the new system yet. I imagine the larger die will help offset the additional heat of the extra cores.
 
What cooler are you planning to use on the 5820? Hopefully not the 212?
 
Ocellaris said:
What cooler are you planning to use on the 5820? Hopefully not the 212?
Click to expand...
alex_di said:
An update on this-- I ultimately returned the 6700K in favor of a 5820K. Entirely new platform, this time with a water-cooling kit. Using more heatsink paste definitely helped, but I was still in the high 80s. I haven't assembled the new system yet. I imagine the larger die will help offset the additional heat of the extra cores.
Click to expand...
 
As of now, my plan is to use a Scythe Mugen 2 Rev. B with two fans. It's an 850g heatsink I've had on an overclocked 875K for years with a single 600 RPM fan. That chip never breaks 65C. The 212 Evo will be repurposed to an X3450 Xeon rig.

Liquid cooling may be in the cards if the Scythe isn't good enough. I'm not sure with what yet. Acoustics is the driver; whatever I buy, I don't want to hear even with the chip overclocked. The kit I originally bought (Zalman 320) doesn't seem like it'll hold up when the chip is stressed.
 
alex_di said:
I'm looking for direction on how best to deal with a cooling problem with a new system.

I had a 3770K with a 212 Evo cooler with two 120mm fans. Overclocked, it wouldn't break 80C in any test. Room is 72F.

I upgraded to a 6700K and kept the same cooler. After the initial install with a mild (4.4 GHz, 1.25 vcore) overclock, it would slowly rise to 87-88C after a few minutes Prime95. Tearing it down, it appeared I'd used way too much paste; my half-pea had coated both surfaces and seeped out the sides.

I cleaned all that off and installed the 212 again with about 1/4 the paste. Prime95 immediately pushed it to 90C+. The paste had spread to an area the size of the dime. Most of the chip was untouched.

I tried again, this time dabbing paste on the four corners and the center. Same result in Prime 95. Worse, even. I stopped the test immediately as two cores passed 95C. Tearing it down showed five even blobs, though about 20% of the CPU surface remained untouched.

All of these temperatures suck. I've gone over both the CPU and heatsink with a straight-edge and a flashlight and they *seem* to be flat.

I see these options:

* Assume the CPU is warped. Exchange it.
* Assume the heatsink is warped. Buy sandpaper and attempt to smooth it out, or buy a new heatsink.
* Assume they're both warped and do both of the above.

Which of these is most sensible to start? Assuming I don't want to lap the CPU, is lapping the heatsink likely to fix this problem? Or is there an approach or technique here I'm missing?
Click to expand...

There is one thing you didn't consider. Some heatsinks and cpu blocks have a mounting mechanism in a way that they control the exact distance from the pcb to the block/heatsink. For instance, Swiftech uses it. What this means is that if you lap anything... you will change such distance and, thus, lower the performance of your cooling system.

In any case, you are using a mediocre cooler for a very small chip. What did you expect? I'm also very surprised that you can return things just for the sake of it...

Araxie said:
First of all things, and I always have to make people remember this, DON'T use Prime95 with anything newer than Ivy Bridge (So, Haswell, Haswell refresh = devil's canyon and Skylake), if you want to use prime95 then only use the version 27.7 if not trust me you can even damage your chip. ok now, second, what motherboard are you using? with what kind of settings and more important, are you using Manual Vcore? Auto? Adaptive? offset?. Be always sure that the Vcore remains that way during any stress test and re-do all the test with prime 27.7 to see if the overheating issue are still present.

On the other hand, Put a small pressure with your fingers in the cooler against the CPU (push just slightly the CPU against the mobo) during the Stress test, see how temps behave and check if any difference in temp occur, why this? well few things, not every motherboard PCB is the same Thickness, some manufacturers to cheap out use thinner PCB than most others, and also skylake use a thinner PCB so when you combine a thinner mobo PCB with a thin CPU PCB you may be able to found in a situation where the cooler doesn't make proper contact causing issues with temperatures, so this fast test can help to confirm or dismiss this issue, I know from own experience that this can happen with Cooler master Hyper 212 coolers, as they are typically kinda loose mounted.

Contrary is the case where the Cooler can crate such huge pressure against the CPU that it can literally warp or bend the CPU PCB causing also insufficient contact, so in this case loosening a bit the cooler help to decrease the temperatures.

Also test everything at stock settings and then overclocked to make complete comparisons between stock and overclocked temps.
Click to expand...

You don't have a clue.

a) You can't damage a cpu by running software, since nowadays they all have thermal throttling in place. Heck its been like that for ages.

b) Pcb width is not important, because the distance from the pcb to the IHS is STANDARD as in, ITS ALWAYS THE SAME (so long as you are using the same socket, that is), or retention systems such as those in use by Swiftech wouldn't exist.

d) Putting pressure with your fingers? You have to be kidding me. Yes, lets put the fingers on top of a motherboard of a running system and lets try to apply pressure on a system that is already pressured to hell. For starters you will not do anything regarding adding pressure on top of the cpu (but you will add stress on the motherboard itself), and secondly you might short something and fry the whole thing. Heck, you might even zap yourself while you are at it.

Please, stop misinforming people, as you have absolutely no clue at all. You are also a safety-hazard.

alex_di said:
Good call on the razor, I was only using a metal ruler. Both surfaces are concave. The heatsink has one corner that warps concave just a bit; it took a few tries to catch this. With the CPU OTOH, the concavity is consistent across the surface and was immediately evident.

Which is annoying. I'm starting to think I should bring a razor to Microcenter and test the exchange in the store. Unless they're all likely to be terrible. I recall I had to exchange my first 3770K for exactly this problem. It's not the lapping part that concerns me, it's voided the warranty on a CPU I just bought, but I'll do it if that's what takes to get the temperatures out of the stratosphere.

Araxie, I had the voltage set to standard vcore + 0.5 offset. No other changes. The board is a Gigabyte GA-Z170MX-Gaming 5.

Shameless, the system is actually flat now. The case hasn't arrived yet, I have the board set up on an antistatic pad.
Click to expand...

Some heatsinks / cpu blocks are concave on purpose. Just saying...
 
I've never heard of an intentionally concave heatsink. Convex, yes, but concave wouldn't make sense because the majority of the heat is coming from the middle of the heat spreader.

I'm looking forward to giving the Scythe a go (assuming I can still fit my RAM under it). Had to buy an adapter screw set for it. I can't imagine many of these are running on the 2011 platform.
 
prava said:
You don't have a clue.

a) You can't damage a cpu by running software, since nowadays they all have thermal throttling in place. Heck its been like that for ages.

b) Pcb width is not important, because the distance from the pcb to the IHS is STANDARD as in, ITS ALWAYS THE SAME (so long as you are using the same socket, that is), or retention systems such as those in use by Swiftech wouldn't exist.

d) Putting pressure with your fingers? You have to be kidding me. Yes, lets put the fingers on top of a motherboard of a running system and lets try to apply pressure on a system that is already pressured to hell. For starters you will not do anything regarding adding pressure on top of the cpu (but you will add stress on the motherboard itself), and secondly you might short something and fry the whole thing. Heck, you might even zap yourself while you are at it.

Please, stop misinforming people, as you have absolutely no clue at all. You are also a safety-hazard.


.
Click to expand...

hahaha you boy, you have no freaking idea of what are you talking about right?. just spiting shit like if you know what shit do you think you know.

first of all, just google it a bit and enlight yourself, Prime95 can damage a CPU newer than haswell that's the main reason intel refreshed the 4770K to 4790K with a more robust power delivery circuitry, better VRM that reacted better to sustained overload.. prime95 with newer AVX2 instruction do abuse of those instructions that cause insanely fast chip degradation, this same apply to Broadwell and Skylake, heck even ASUS have a power guide for overclocking haswell-E where they clearly do not recommend the use of Prime95 due the fast damage it can cause due to couple of reasons, the first one related to huge power consumption of those chips overclocked, a 5960X overclocked to just 4.2ghz will pull over 400W which can cause both Motherboard damage and CPU damage. so boy, the one who do not have clue is you..

Second, the PCB of Skylake is thinner than any other used by intel, so most coolers received new mounting system to prevent issue, of wait let me post some reading stuff for you because you need to learn a bit to avoid a bigger embarrassment in the future.

Investigating Reports Of Intel Skylake CPUs Damaged By CPU Coolers (Update 4)

PSA: Thinner Skylake CPU Can Get Damaged from Cooler Mounting Pressure UPDATE: 12/7/2015 - Modders-Inc


oh wait couple of pictures:

S-Thinkness.jpeg


PCW-Thick.jpg


intel-2015-skylake-delid-03.jpg


Curved_CPU_bottom-pcgh.JPG


IMG_8244.JPG




the third point, damn, im not going to even respond to your blatantly ignorant commentary... "you might even zap yourself while you are at it" hahahahaha. you are a joke, clearly.. just be applying A BIT (yes, tiny amount of force) to a CPU cooler against the CPU to check if the pressure is correct? hahahahahahahaha a freaking damn joke..
 
Last edited:
Araxie said:
first of all, just google it a bit and enlight yourself, Prime95 can damage a CPU newer than haswell that's the main reason intel refreshed the 4770K to 4790K with a more robust power delivery circuitry, better VRM that reacted better to sustained overload.. prime95 with newer AVX2 instruction do abuse of those instructions that cause insanely fast chip degradation, this same apply to Broadwell and Skylake, heck even ASUS have a power guide for overclocking haswell-E where they clearly do not recommend the use of Prime95 due the fast damage it can cause due to couple of reasons, the first one related to huge power consumption of those chips overclocked, a 5960X overclocked to just 4.2ghz will pull over 400W which can cause both Motherboard damage and CPU damage. so boy, the one who do not have clue is you..
Click to expand...

I did google and found... nothing. Lots of people asking if X might damage Y cpu, but no thorough testing proving anything. How many cpus were sold in that generation? Yet I couldn't find a single proper report of the issue. Also... I like how you take numbers out of context. The only link I have found of a 5960X showing power consumption figures comes from an Anandtech's review. Other reviews do not show such a huge power consumption, not even with higher overclocks:

Core i7 5960X - 5930K and 5820K processor review

Power, In Depth: Eight and Six Cores at 4.5 GHz - Intel Core i7-5960X, -5930K And -5820K CPU Review: Haswell-E Rises

Power, In Depth: CPU Health at 4.8 GHz - Intel Core i7-5960X, -5930K And -5820K CPU Review: Haswell-E Rises

You will have to find more sources if you want anybody to believe your claim that prime95 is hazardous for cpu's. So far I haven't seen anything that might indicate that.

Second, the PCB of Skylake is thinner than any other used by intel, so most coolers received new mounting system to prevent issue, of wait let me post some reading stuff for you because you need to learn a bit to avoid a bigger embarrassment in the future.

Investigating Reports Of Intel Skylake CPUs Damaged By CPU Coolers (Update 4)

PSA: Thinner Skylake CPU Can Get Damaged from Cooler Mounting Pressure UPDATE: 12/7/2015 - Modders-Inc


oh wait couple of pictures:

S-Thinkness.jpeg


PCW-Thick.jpg


intel-2015-skylake-delid-03.jpg


Curved_CPU_bottom-pcgh.JPG


IMG_8244.JPG
Click to expand...

Again, you should check your sources more thoroughly.

1) Incidents of bent cpu's are not very common, but they do have something in common: they were found bent after being transported and having on them a heavy tower heatsink.

2) Intel acknowledges that the cpu pcb is indeed thinner... but the mechanical pressure it can sustain is, effectively, the same. The problem, then, isn't in the pcb.



the third point, damn, im not going to even respond to your blatantly ignorant commentary... "you might even zap yourself while you are at it" hahahahaha. you are a joke, clearly.. just be applying A BIT (yes, tiny amount of force) to a CPU cooler against the CPU to check if the pressure is correct? hahahahahahahaha a freaking damn joke..
Click to expand...

Clearly introducing your hand in a running computer and trying to randomly "add pressure" isn't potentionally harmless. Right. I bet you get the boxes on top of the closet by using a chair with wheels, too. Also, you seem to have special powers if by touching the cpu you can understand if the pressure is right or not. Mmmm.
 
alex_di said:
I've never heard of an intentionally concave heatsink. Convex, yes, but concave wouldn't make sense because the majority of the heat is coming from the middle of the heat spreader.

I'm looking forward to giving the Scythe a go (assuming I can still fit my RAM under it). Had to buy an adapter screw set for it. I can't imagine many of these are running on the 2011 platform.
Click to expand...

You are right. I mixed them up. Heatsinks have to be either flat or convex.
 
Most likely you have a defective chip. Mine is currently at 33C and i'm using it with a noctua cooler.
 
Try with stock Intel heatsink and no over clock. See how the temps are. If they are high for everything stock, RMA with Intel.
 
Armenius said:
Haswell-E and Skylake K- processors do not come with an OEM HSF.
Click to expand...

Oh you're right; Intel started cheaping out. Maybe Intel will let you RMA it with the claim that temps are too high. Otherwise, you'll have to purchase an "Intel approved" cooler, test it to prove temp issues and then RMA.
 
At the moment, I have the 5820K at 4.4, all cores, adaptive voltage with max set to 1.31V. At 1.3V, I could complete benchmarks at 4.7 and it would lock at 4.8. This seems about typical.

Temperatures with the Scythe and a single slow 120mm push fan start at mid-60s under load, eventually rising to 80C or so, which triggers a fan speed increase that holds that temperature. Stable in regular use, though I haven't stressed it with a full load for more than ten minutes at a time yet.

Oddly, it benchmarks better with all cores locked together than it does letting the first two or three go to 4.6 GHz.
 
Last edited:
Nice overclock on your 5820K. I can't seem to get past 4.5GHz on mine without RealBench crashing and that's at 1.375Vcore.
 
You must log in or register to reply here.
Back
Top