XMP causing core temps to rise?

RxDealer

Prominent
Feb 14, 2017
16
0
520
1
Hi,

I have just managed to finally get a stable overclock on my i7 7700k @ 4.9GHz with 1.32v with default memory settings. The next step I took was enabling XMP, but this caused my core temps to rise about 5-15c. What could be causing this, and how do i remedy it? Should I manually set my ram or leave XMP on and decrease the VCCIO and System agent voltage? My temps without XMP on are around 75-80 when stress testing and when turned on are around 85-95!!!!

Setup
I7-7700k
Asus strix z270e mobo
g.skill trident z rgb 3200 14-14-14-34 ram
 

RxDealer

Prominent
Feb 14, 2017
16
0
520
1
I actually fixed the problem finally after hours and hours of trial and error. What I ended up doing is resetting the bios to the default optimized settings (not cmos clear). Then I manually oc'ed my cpu. Stable. Then I manually played around with my ram. Started at its recommended overclocked factory speed and timing which was too hot. So I had to decease the MHz on the ram and temps went down. I slowly started increasing ram MHz and temps gradually got hotter, but when I loosened the timings my temps got stable. So I was able to achieve the same frequency XMP wanted to set. I just had to do it manually.
 

TJ Hooker

Illustrious
Herald
Did you double check what your CPU frequency is after applying XMP? I've had it happen where enabling XMP increased my BCLK rather than just the memory multiplier, which would raise your CPU frequency as well and possibly raise temps.
 

RxDealer

Prominent
Feb 14, 2017
16
0
520
1
So I tried reducing viccio and system agent down to 1.0 and still getting high temps. Will try some other tweaks and report back.
 

RxDealer

Prominent
Feb 14, 2017
16
0
520
1
So after messing with alot of settings, the only thing I could get to work was changing the DRAM voltage from the XMP setting of 1.35 to auto and downclocking my memory from its XMP rating of 3200 to around 2500mhz. This seems drastic to me. Has this ever been heard of before or am I missing something here? Also, would it be better from a performance perscpective to downclock my cpu from 4.9 GHz to achieve reasonable temps with a higher memory clock or the converse of leaving my memory downclocked?
 

RxDealer

Prominent
Feb 14, 2017
16
0
520
1
Yeah that might just be the result I have to live with, but I'm super disappointed about it. Spent extra money on ram for what? Maybe I can take advantage of it later down the road. Just wish there was a better answer.
 

RxDealer

Prominent
Feb 14, 2017
16
0
520
1
Well would reducing the timings reduce heat at all? I retried the 2500 today at stock timings and temps went too high again. Fml I'm really at a loss.
 

koolman007

Distinguished
Jun 25, 2007
74
0
18,640
1


I am with you on this m8. I have similar results as urs 70C on a non-XMP OC with Noctua D15 but as soon as i turn on XMP temps go wild Since my 2500k @ 4.5 temp never went beyond 60C, this is a completed shocker to me, which also begs the question why i spend so much money on higher clock memory ?
Guess I will be just using stock clocks till my new upgrade.
 

RxDealer

Prominent
Feb 14, 2017
16
0
520
1
I actually fixed the problem finally after hours and hours of trial and error. What I ended up doing is resetting the bios to the default optimized settings (not cmos clear). Then I manually oc'ed my cpu. Stable. Then I manually played around with my ram. Started at its recommended overclocked factory speed and timing which was too hot. So I had to decease the MHz on the ram and temps went down. I slowly started increasing ram MHz and temps gradually got hotter, but when I loosened the timings my temps got stable. So I was able to achieve the same frequency XMP wanted to set. I just had to do it manually.
 

RxDealer

Prominent
Feb 14, 2017
16
0
520
1
Really happy with the result. Downloaded battlefield 1 and I'm running 1440p@ 100fps with ultra settings with gpu temp<70 and CPU temps around 65-70.
 

ASK THE COMMUNITY