Overclock.net banner

1 - 14 of 14 Posts

·
Registered
Joined
·
159 Posts
Discussion Starter #1
Hello, I used to have a 6700K with an AsRock Z170 Extreme 4 motherboard, I delided that cpu and my results were 1.344 vcore @ 4.6 ghz with 62 max temps on the cores. So I called it a "successful delid".

I upgraded recently to an 8700K with an Asus Z370-E Strix board, got the cpu, made sure it works while stock and checked for temps under load with XMP on. 71c max.

I delided that 8700K, put 1.2 volts to test and I reached 4.9 Ghz stable (AIDA64 1 hour). The concerning thing was my temps.

I used a Noctua C14s for both cpu's, and my 8700K reaches 69 degrees max, with only 1.2v, while the 6700K reached 62c with 1.344. Do you think this is normal?


I used the delid kit from RockIT, Thermal Grizzly Condactonaut liquid metal and glue to put the IHS back on. Thermal grease is Arctic MX4 (pea size method, as I used for years)

Multi core enhancement is off, LLC level 6, XMP profile running memory at 3200 as its supposed to, and it's an open test bench. All those variables scare me a bit since I haven't put the system even in a case yet.

Any suggestions? Thanks.
 

·
Overclocker
Joined
·
3,761 Posts
Sounds like you didn't win the silicon lottery.
To get better temps you have to downclock the cpu. IMO.
But have you played games to get an average and not synthetic tests?
 

·
Registered
Joined
·
5,374 Posts
Did you scrape off all the silicone?
Have you reapplied the TIM? You want it very thin.
Did you glue it back down?

It's hard to tell why your temps are where they are without actually seeing settings and I wouldn't recommend posting them all.
Just start by reading Darkwizzies guide for your CPU.
Reset to defaults. Don't enable XMP.
Set cache manually to stock.
Set LLC to it's middle setting.
Disable CPU spread spectrum.
Set your desired core frequency and voltage. Set CPU input to about 0.5v over vcore.
Set manual/static voltage till you find your minimum then do adaptive when you're done stressing.
Then RAM, then cache.

If you've done all this then your temps are probably as good as they'll get.

I stress with realbench using the stress setting and benchmark on a loop.
 

·
Registered
Joined
·
159 Posts
Discussion Starter #4
Sounds like you didn't win the silicon lottery.
To get better temps you have to downclock the cpu. IMO.
But have you played games to get an average and not synthetic tests?
Honestly I wouldn't say that, since I'm keeping it very stable at 4.9 with only 1.2v. I haven't seen that many chips able to do that.


Silicone is completely scraped off. Its all clean under the IHS. Just the smallest amount of glue under the two side bulges to keep it in place.

If you mean the TIM on the IHS, yes as I said, its AC MX4, it's very thin. Contact patch is great.

I'll check darkwissie's guide thanks a lot.

Right now I'm testing it with a Kraken x62 cooler on it to see if my CPU cooler is simply not enough. But I highly doubt it for only 1.2 volts. I'll post results.
 

·
PC Evangelist
Joined
·
46,986 Posts
Its a 6 Core vs 4 Core CPU.
 

·
Registered
Joined
·
159 Posts
Discussion Starter #6
Its a 6 Core vs 4 Core CPU.
Sure, except my temps weren't like the ones I read on the forums lately. That worried me a bit.

Reporting with a kraken x62, max temp is 61c. Also the Noctua cooler has some minor scuffs on the coldplate, I assume that has an effect too.I'll be buying a D15s very soon, as I don't really like the cable mess the AIO coolers make you put up with.

Also I played around with some settings, like disabling SVID and spread spectrum.

Still stable at 4.9 with 1.2 vcore 45x cache and no AVX offset. I consider myself lucky with the chip, what do you guys say?
 

·
Registered
Joined
·
1,586 Posts
Sure, except my temps weren't like the ones I read on the forums lately. That worried me a bit.

Reporting with a kraken x62, max temp is 61c. Also the Noctua cooler has some minor scuffs on the coldplate, I assume that has an effect too.I'll be buying a D15s very soon, as I don't really like the cable mess the AIO coolers make you put up with.

Also I played around with some settings, like disabling SVID and spread spectrum.

Still stable at 4.9 with 1.2 vcore 45x cache and no AVX offset. I consider myself lucky with the chip, what do you guys say?
I think your temps are right where they should be. I'm running delidded at 1.4 vcore for 5.0 46x cache 0 AVX offset. Temps hit the low 80s but it stays stable so far. I'd love to run 1.35 but it won't stay stable. Maybe I should mess with LLC settings (which are currently on auto)?
 

·
Registered
Joined
·
159 Posts
Discussion Starter #8
I think your temps are right where they should be. I'm running delidded at 1.4 vcore for 5.0 46x cache 0 AVX offset. Temps hit the low 80s but it stays stable so far. I'd love to run 1.35 but it won't stay stable. Maybe I should mess with LLC settings (which are currently on auto)?
I can not really put my finger to point what LLC settings you should be running since I don't know your motherboard, but nearly all the motherboards tend to drop the voltage when under load (a phenomena called vdroop). The highest the number of LLC for example level 1 to 7 on my motherboard, means that the motherboard supplies more voltage under load to compensate for the vdroop. I'm using level 6 and it seems to keep the voltage balanced under load (1.2 on the bios, 1.2 under full synthetic load). I suggest you go back and forth starting at the middle, to find your systems sweet spot.

Man those 6 core cpus are hot.
 

·
Registered
Joined
·
1,168 Posts
If that comforts you, I'm 5.0Ghz stable at 1.280v and 0 AVX offset and my temps running Aida 64 are 70~75ºC, also delidded, used Conductonaut between DIE and IHS and the same Arctic MX-4 on top of IHS (waiting for Kryonaut to come to change), I use an Corsair H110i and my temps are like that when I'm at 27~30ºC ambient temperature.
 

·
Registered
Joined
·
3,004 Posts
To much or too little liquid metal can also effect it. For a reference I'm at 1.36-1.38v 5.3ghz and get around 56c max in CB, around 54c average in OCCT large data. All I use is an h100i,
 

·
Registered
Joined
·
1,168 Posts
To much or too little liquid metal can also effect it. For a reference I'm at 1.36-1.38v 5.3ghz and get around 56c max in CB, around 54c average in OCCT large data. All I use is an h100i,
Really??? You live in an iceberg?
 

·
Registered
Joined
·
1,168 Posts
As about the amount of liquid metal, I do this: I put the amount and then I hold the CPU vertically and if the LM doesn't go down (gravity) or creates a ''fat'' in the down side of DIE, I know that this way is not too much LM, so I use to apply and test like that.
 

·
Registered
Joined
·
159 Posts
Discussion Starter #14
As about the amount of liquid metal, I do this: I put the amount and then I hold the CPU vertically and if the LM doesn't go down (gravity) or creates a ''fat'' in the down side of DIE, I know that this way is not too much LM, so I use to apply and test like that.
I also do this! I thought I was alone with that :(
 
1 - 14 of 14 Posts
Top