Question Question re: High temps with new aio

[Tics]

Hey all,
Ive just installed a cooler master ml240r in place of a phanteks tc14pe, not because that giant air cooler wasn't great, it definitely was, but because it lined up with the x16 slot in my mobo, and my new gpu has a huge backplate and it wouldn't fit into the x16 slot with that cooler installed, so I had to have it in the x8 wired slot which was causing random bottlenecks, such that my fps in games like fortnite would drop from 200 plus to like 5 fps. That is still happening to some extent in the x16 slot so I'm now maybe think the 3770k is bottlenecking it but idk.
But anyway, long story short, with my old air cooler I never had more than an 11 deg celcius difference between core 0 and 4. (I have an i7 3770k, p8z77m-pro, 16gb Corsair LP ram, rtx 2060) Now with this new cooler running prime95 small fft's overclocked to the same 4.5ghz @ 1.36 vcore I have a 22 deg difference between core 0 and 4.
I don't normally focus too much on individual core temps but that just seems odd to me.
The chip has been delidded, liquid metalled and re-lidded a while back and it has ran fine with the phanteks cooler but yeah just seems a little hot for my liking now.
Any ideas what might be causing that to occur?
Cheers for any help
Julian
P.s. let me know if I can provide any more info

Edit: my temps on idle currently are better; 4 deg difference, core 0 @ 33 deg and core 3 @ 29 deg

 

[Roland Of Gilead]

The fact that the difference in core temps (22c difference between certain cores) makes me think the thermal paste hasn't been applied evenly, or the new cooler hasn't been mounted flush with the CPU. Normally with Intel pics there is a bit of a difference between core temps, but normally only 3-6 degrees. Not 22!

The CPU at 4.5 seems to have quite high vcore. 1.36. A lot of folk OC that chip use much lower Vcore, typically about 1.3v. Have you tested for stability on the OC? It's possible getting the vcore lower by stress testing properly to discover your lowest stable vcore, will help some.

 

[Tics]

Yeah I've done plenty of stress testing using prime95 small fft's to produce maximum heat for said purpose. Ive ran it continuously for over 24 hours straight and not had any WHEA errors thrown up, no bsods or anything like that.. vcore is high I know, but ultimately I'm going to be upgrading when Intel finally bring out 10nm so hopefully I won't have the i7 past the end of the year (presuming they don't delay it again in which case I might make the switch to ryzen), but thats why I was happy with a slightly higher vcore as chip longevity at this point is not my highest priority, just maximum performance.
Also fyi I cannot get 4.6ghz with this chip stable no matter how high my voltage is so I'm pretty sure 4.5 is my limit. Bit unlucky with the chip lottery I think which is probably why I've needed a relatively high vcore to get things stable at 4.5ghz.
Ive just dropped the oc down to 4.4, with a lower voltage offset and that has closed the gap slightly but not by much. It has also naturally dropped my max temps a bit.
I also want to mention that at idle once the computer has been running for a while the difference between core temps is only about 5 degrees (core 0 is currently 39 and 3 is 34 and that's 5 seconds or so after stopping prime95 small fft's test which was running for an hour and 6 mins this time) according to hwinfo64 but under a heavy avx based workload it stretches out to that 20ish deg figure.
I've since reseated the block and re-applied thermal paste and it hasn't changed anything too drastically.
Also sorry I wasn't totally clear in my first post but I had been running the exact same overclock and voltage for 6 months or so with the phanteks air cooler.
All I've done recently (prior to reseating the block) is removed the phanteks cooler, cleaned the CPU lid of the old paste, re-applied new paste (that came with the CoolerMaster ml240r) and installed it as per the instructions.

After reseating the block and getting booted back up I found my core 3 to be pretty much the same, however core 0 is still running a fair bit hotter as previously stated.
It is a different thermal paste (to what I used with the phanteks cooler), and I'm kinda wondering if that could make that much of a difference but I'd be surprised if a thermal paste had such variance between brands.. I was very careful to apply the thermal paste the second time as well but it not made a huge difference..
I just don't get why changing the cooler from air to water has affected core zero so much but not the other three cores.
Going to buy some thermal grisley tomorrow, hopefully that helps and if not I'll probably ignore it as I don't work with avx workloads normally anyway so its not like my temps will be anywhere near that under a normal workload. When gaming under full load it sits at 70 on core zero and 60 on core 3 which I can deal with, I guess i just don't like knowing why it's misbehaving lol

Edit: I should also probably mention that I still have speed step on and turbo boost on and as such I'm using the voltage offset function rather than a straight up manual voltage.. like whist I don't necessarily care about potentially burning the chip out and obvs want the best performance I can currently get from this chip I also don't want to just waste electricity when the extra power isn't needed..
I hope the answer isn't something stuffed up with the delid lol as I don't have my der8auer die mate anymore and I'd really rather not have to buy another one. Lol with Intel using solder again I thought my delidding days were over 😭😂

 

[Roland Of Gilead]

I guess it's possible the change in cooler could have somehow disturbed the de-lid. Did you do the cooler swap while the CPU was hot, as in just shut down? Or was it cold?

 

[Tics]

Yeah it had only been shut down maybe 5 mins prior to me removing the old cooler but id say it was probably open to the air (with no cooler) for 15 - 20 mins or so while I stuffed around. It didn't feel hot to the touch, I'd only had the pc on for about 10 mins before the new cooler was delivered.
And I didn't release the CPU clamp at all during the process so I wouldn't think it could have moved too much as if it did I would presume there would be other symptoms given how delicate lga pins can be.. but then again I'm sure stranger things have happened!
Cheers for your help so far btw I appreciate you taking the time

 

[Roland Of Gilead]

Not at all. I'm glad to bounce it around. Being honest, I've never delided a CPU (I don't see the value, plus am currently running AMD which don't have that problem , or requirement as such )

I'm really just trying to think about it logically. The thermal grizzly might help. It will possibly reduce CPU temps by 5-10c instead of the other one. But i don't see how that might fix that huge variance between cores that sill happen. Like you said, if you've done everything by book, then the lowest common denominator might be the de-lid. Maybe the replacement TIM after delid, however thin it is, might have degraded, specially if you've been trying to push an OC? It's hard to know without actually taking a look. But as you said, your temps are within reason, and with your usage, maybe it's something you can put up with, or put down to an anomaly. Still though, I wanna know the answer lol I wonder if after deliding, it's a good idea to 'top it up'. If ya get me. Like some after care and TLC. Maybe do the process again for longevity. But, if your der8auer die mate is gonesky, then maybe just monitor the system regularly, and see how it goes for now.

 

[Tics]

Me too I really want to know as well lol it'll drive me nuts!.. I'll let you know how I go once I try some conductonaut, which will probably not be until tomorrow at this point. Also what I'll try and do to hopefully narrow down the problem a little bit is put the phanteks back on and see if that brings the cores back to within that 8-10 Degs. If it does I'm thinking I might try and RMA the cooler and get it replaced but if not I've just found a business here in Melbourne that will do the delid for me for cheaper than the die mate, and he is happy for me to watch him do it so that's an option if the cores are still flying apart in thermal outputs even with the phanteks on