Does this 4790k need delidding?

[MstrRoshi]

Hello all,

I'm helping a friend out with his computer build. He's just put a lot of money into the build and is really frustrated with some of the issues he's running into.
He plays Fallout4 and does like to stream to his youtube channel occasionally. During firefights or using VATS the game will lag and sometimes crash. It is more common to happen after a few hours of gaming. In other games, its is sometimes not responsive for no reason. Example: Magic Duels - really shouldn't be a graphics or cpu intensive game, but at times the game 'hangs' and wont let him click on anything, when it gets going, he's missed the chance to play as its a timed game. A lot of little annoyances that add up to me thinking something is wrong...

When he first had these issues, I thought I could fix things with a small OC. His chip has a base 4.0ghz that boosts to 4.4 - and I was able to get him stable @ 4.4 on a 24/7 OC but I don't recall the specific settings or temps he was getting.

This Solved most of his issues but at times the computer would still crash after long gaming hauls. 3-4 Hours, then crash, and more crashes happening more frequently after that.

I came to the conclusion that it's a heat issue. And since he was jamming his $1000 machine into a tiny $30 mid tower atx case... I didn't think too much of it. I just advised him to upgrade his case before doing anything drastic - like upgrading to a 7700k for a tiny 8~10% performance bump...

So he upgraded the case to a bequiet Dark Base Pro 900 - Which is pretty incredible. After getting everything installed - also restored stock bios settings to see what the temps were - I did a p95 small fft test to see where the max temps landed, and I was shocked to see the core temps hit 97C almost instantly... sure enough it hit the wall and throttled in under 2 minutes. Again this is all at stock setting (Optimal Defaults - haven't cleared cmos). So ya crazy temps, and I know something wacky is up.

First thought was too much thermal paste. So we re seated the cpu cooler and sure enough he had a ton of extra paste there. cleaned all that up, pea method to reapply and thought we would be good to go... But same story. Almost instantly up to 97C again, I didn't even wait for it to throttle, I just killed the test.

Auto voltage settings were giving it ~1.3v under load, and that supported the boost to 4.4.
I disabled the turbo boost, and left the voltages on auto to see what they would be set to after a reboot. Sure enough vcore dropped to 1.1 - I ran another small fft and was able to get through 10 minutes, with max temp @ 93C.

Now... this was ok for my friend, because he knows he's never going to be pushing the temps that hot in gaming. But this is in no way acceptable by my standards lol.

These are his computer specs - updated.

CPU: i7-4790k @ 4.0
Mobo: Gigabyte Z97X-UD3H-CF Rev 6.0
Cooler: Hyper 212 Evo
RAM: Crucial Vengeance 2x 8gb - XMP 1600mhz 9-9-9-24
GPU: XFX R9 390x 8gb
Case: bequiet DarkBase Pro 900 - 3 intake/2 exhaust fans
PSU: Corsair HX1000i
Storage: Sandisk 480gb SSD & 500gb HDD 7200 RPM
OS: Win7 Ultimate 64 bit

I used CoreTemp to see if maybe it was just one core that was sky rocketing... unfortunately no. There's only a ~10C degree difference between the hottest/coldest core under load. but its still hovering at ~93C under stock settings. I know these chips tend to run really hot, but this can't be normal..

I've read some other threads about delidding, and watched a few tutorials. I've never done it personally, but it seems simple enough. Coming back to the thread title... Do you guys think this can be fixed by delidding, or would you suggest trying to RMA? (He's had the chip for a long time, so i don't know if it'd be covered). do you guys have any recommendations for TIM? Methods of delidding? Oh specifically the "glue" used to put the lid back on... haven't seen a lot of info on that yet.

Also open to any other suggestions on how to get these temps down. I could under volt the thing, but with temps this drastic, I don't expect it to make much of a difference.

Thanks for all the help

PS> I know this is a stupid question, but I just want to clarify. I've seen folks with sigs that have "i7-7700k @ 4.5 Delidded" when i first saw that, I thought they were running the chips without the lid entirely... so CPU DIE with TIM directly touching the CPU cooler - this is not the case correct? Delidding simply means you've opened up the cores for maintenance of some kind - reapplying TIM and cleaning up any gunk that may be preventing the IHS/TIM from touching the lid, right? Just want to make sure I understand this correctly. Thanks!

 

[||Console||]

After deliding most ppl dont bother reattaching the IHS they use use the bracket that keeps the cpu in place to keep the ihs there.
Most will recommend Liquid metal for inbetween the DIE and the IHS . and then pick your fav Tim for between the IHS and the HSF .

93 deg @ 1.1v seems very high .
Make sure the rest of the voltages @ stock as well . CPUvrin if set to high could cause higher temps .
Does the MB have the newest bios ?
What are the room temps ?
That system should do 4.5ghz pretty easy if not 4.7-4.9
how fresh is the windows install ?
what are you using to read the temps ? Coretemp ? or somthign else .
Are temps any better with HT disabled ?
Is there anything running in the background using alot of resoruces ? (FAH)

 

[Nebulous]

Here's a really good explanation of what delidding means. It's NOT for the faint of heart.

Everything else console has you covered.

 

[MstrRoshi]

Here's a really good explanation of what delidding means. It's NOT for the faint of heart.

That's a fantastic read! Thank you!
It's exactly what I gathered from the videos I watched of people delidding, but the images in that article make everything crystal clear. Also, none of the videos I watched had mounted the HSF directly to the DIE, which was what confused me about the term DeLid haha. Makes sense that you can do it either way - depending on what kind of cooler you have obviously.

Looks like a fun project! I have a dead 8350 that i will get some practice on

 

[MstrRoshi]

After deliding most ppl dont bother reattaching the IHS they use use the bracket that keeps the cpu in place to keep the ihs there.
Most will recommend Liquid metal for inbetween the DIE and the IHS . and then pick your fav Tim for between the IHS and the HSF .

93 deg @ 1.1v seems very high .
Make sure the rest of the voltages @ stock as well . CPUvrin if set to high could cause higher temps .
Does the MB have the newest bios ?
What are the room temps ?
That system should do 4.5ghz pretty easy if not 4.7-4.9
how fresh is the windows install ?
what are you using to read the temps ? Coretemp ? or somthign else .
Are temps any better with HT disabled ?
Is there anything running in the background using alot of resoruces ? (FAH)

Thanks for the clarification - 1 video I watched did use the socket as you mentioned. I just thought he was making due, but after reading the article Nebulous shared it makes much more sense now.

Voltages - As of now everything is set to auto. It is possible that under load the auto settings are spiking the voltages, but like you said - seems very high. I'll go through and manually set the voltages to see if that changes anything under load.
Mobo Bios is currently F9 on the Z97X-UD3H-CF Rev 6.0
Room Temps are sitting at 77F / 25C - uncomfortable for us, but shouldn't be too much
Windows install was on 4/7/17 - at least thats the oldest date in the event log. Seems about right, we just bumped him back to Win7 from Win10 to see if the fewer background processes had any effect on his games lagging out. It did not..
For my initial test I was using SpeedFan. When the temps shot up, I added CoreTemp to confirm what SpeedFan was reading and it was accurate.

Is HT the turbo boost that bumps it from 4.0 to 4.4 under heavy work loads? Cuz disabling that did help a little.

Or do you mean try disabling the Hyper Threading and only run 4 cores/4 threads? - This I have not tried. Didn't even consider it an option. If I do drop it down to 4 threads and it makes a significant difference, what would that be indicating?

As far as background processes / resources - we ran the tests right after a fresh reboot. I always employ selective start ups through msconfig to keep those background process down, so nothing should have been running. that said, i did not have task manager or performance monitor running to confirm.

Really appreciate your help!

 

[MstrRoshi]

So after setting all voltage values to normal, rebooted and checked the bios to see if the values changed from auto - they did not.
Next i manually set the voltages to their stock/normal values. Re tested:



CoreTemp Zoom - this is after 2 min

 

[MstrRoshi]

Are temps any better with HT disabled ?

Yes they are. This is after 35 min of small fft

still a bit hotter than id expect/like

 

[trents]

1.3v vcore at 4.4ghz seems high to me. When you took the vcore off of auto did you try getting by with less. If you could be stable at 1.25 that would have a very favorable impact on temps.

 

[SPL Tech]

Here's a really good explanation of what delidding means. It's NOT for the faint of heart.

Everything else console has you covered.

It's also not half as complicated as people make it sound. You can do it easily in about 10 minutes. I use a razor blade, not the overpriced, unnecessary deliding tool, and I have zero issues getting the job done. If you use the deliding tool, then the risk really is quite low as long as you have half a brain and dont do something stupid.


Anyway, deliding shouldent even be on the table here. Clearly there is a problem with the processor or heatsink mount. There is no way he should be getting 100c at STOCK clocks and voltages. I am guessing it's a problem with the cooler mount. Check to make sure the cooler is truly mounted properly and there is zero gap between the cooler and CPU. Try using the stock Intel cooler to see if that fixes anything. If that doesent work, I'd RMA the processor as something is clearly wrong. There is no reason why the temps should be that high at stock and I would not delid a processor that might have a more complicated issue wrong with it.

 

[MstrRoshi]

1.3v vcore at 4.4ghz seems high to me. When you took the vcore off of auto did you try getting by with less. If you could be stable at 1.25 that would have a very favorable impact on temps.

You are right, but its not totally unheard of. Even at that voltage it shouldn't have been hitting 100C and throttling. When I disabled the turbo boost, the auto settings changed the voltage down to 1.1v but as the screen shot shows it still came in at 93C after a 10 min small fft.

Anyway, deliding shouldent even be on the table here. Clearly there is a problem with the processor or heatsink mount. There is no way he should be getting 100c at STOCK clocks and voltages. I am guessing it's a problem with the cooler mount. Check to make sure the cooler is truly mounted properly and there is zero gap between the cooler and CPU. Try using the stock Intel cooler to see if that fixes anything. If that doesent work, I'd RMA the processor as something is clearly wrong. There is no reason why the temps should be that high at stock and I would not delid a processor that might have a more complicated issue wrong with it.

I think you are spot on here. When he first got the Hyper 212 Evo he did the initial installation of the back plate, when we reseated the cooler just recently, we did not take the back plate off, and I think the problem lies there, or with the mounting screws/bolts. It didn't seem like all the screws tightened down completely into the bolts, or the bolts weren't hitting the backplate, but it aslo was quite snug, and didn't rock or wiggle at all when seeing if it was fastned down. Granted, the mobo was still in the case so I couldn't really look at it from the side to see if it was flush. Did not go as smoothly as the videos did thats for sure. Tonight after work I'll pop the mobo out and remount that backplate, and the cooler.

If that doesn't work, we'll try the stock cooler and see what that gets us. that should determine if its the CPU or the cooler. Obviously stock cooler with normal temps means we just screwed up mounting the Evo or it is defective, and if we still get crazy high temps with the stock set up, he should try to RMA. Is there any time limit on RMA process?

 

[trents]

If the coolant has leaked out of the Evo's pipes because of damage or factory defect then that could cause it to be ineffective.

 

[MstrRoshi]

If the coolant has leaked out of the Evo's pipes because of damage or factory defect then that could cause it to be ineffective.

I'll make sure to check the pipes for any cracks. It's fairly new tho, so I'm not expecting to find any.

 

[wingman99]

Something is not right I do well with hyper 212 at 4.6GHz. Check your mounting.

 

[Mjolnir]

Cooler Masters mounting can be usually only installed one way that matches with the intel backplate. I personally and several others I've known who are also experienced builders have still clumsily installed the CM backplate wrong so the contact was not perfect and temps skyrocketed. This may not be your issue, but just confirm it all lines up at the back so you can see the 2 screws from the intel backplate. I also find the 212 Evo mounting in general to be finicky, but maybe that's just me..

It really sounds like a mount or otherwise cooler issue. The 4790k gets hot, but not that hot on a reasonable cooler like the 212 evo.

 

[Tech Tweaker]

These are the kind of temps I saw when I installed a stock Intel E97378-001 heatsink/fan on my 4790K with no thermal paste (don't do that) with it running at low fan speeds (BIOS fan speed control is dumb, doesn't increase speed with temps for reasons unknown.)

I can run my 4790K at 1.32v for 4.7GHz (temps at 79°C max), 1.18v for 4.4GHz in my case (temps at 68°C max) (Your results may vary from my own.). I wouldn't expect your friend's 4790K to need 1.3v for 4.4GHz unless it is an unusually bad overclocker (possible, but probably not likely), really if it needs 1.3v for less than 4.5-4.6GHz that would seem strange to me. My processor is strange though, with the voltage on Auto with stock frequency it defaults to 1.28v (which there is no way it needs at 4GHz base or 4.4GHz boost), so mine needs to be under-volted at stock to keep temps in check and power consumption closer to normal for one of these processors.

I think your friend (and you) are dealing with a bad cooler or a bad mount in this case. Even if the heatsink fan were unplugged it shouldn't hit 90+°C as fast as it sounds like it is.

I'm not certain how reliable Core Temp is for Intel processors (I've always used it more for AMD processors), I use Real Temp 3.7 for Intel processors typically.

 

[EarthDog]

Coretemp is extremely reliable for intel. In fact, the coolest hit me up a couple weeks ago for testing on SL-X.

 

[MstrRoshi]

When i talked to him about re mounting the cooler he did not want to deal with it. Hes determined to buy a threadripper and doesnt care if this chip fries.
Im thinking ill just fix it for him while hes at work and not tell him about it. Not worth the debate. Ill update after i can get it re mounted. I'm certain that it is a mounting issue

 

[MstrRoshi]

After remounting the cooler again, i am seeing better temps...ish. With HT disabled i get this after 15 min small fft.


i have HT enabled now and will post results. Doesnt seem to be much change tho. Still up to 90 after 5 min

- - - Updated - - -

 

[knoober]

What about the fan? Not sure of your fan curve, but I see it isn't moving any faster between one shot and the next. Im pretty sure my fans are running 2x the rpm for my phenom II @ 4GHZ. Not sure if there is a big difference between fan models/cpu type but I think those numbers are a bit low for adequate cooling. Food for thought I guess

 

[EarthDog]

Mastr - If you hit prtscn button, then paste in paint and save, you can attach a screenSHOT which is a lot eaiser to read. Or use windows snipping tool.