Overclock running hot?

The_Staplergun

Estimable
Jan 30, 2017
1,395
0
2,960
I slapped a h115i on my 7700. Went into the bios and got it at 5.0 without a hitch at 1.375. Havent played with the voltage to lower it yet, that's next.

My concern is I'm using CPU z to stress test the cpu and it peaks at 75c. Is that normal or safe?

Also, should I make the cache ratio the same as the cpu ratio or is it fine. Cache ratio is at x42 and cpu ratio at x50 with BCLK 100.

Idle temp ~30 or less.
 
Solution
Watercooling varies a lot. Lower end AiO coolers are worse than inexpensive tower coolers, while custom loops ($300+) can significantly better. Each chip will have different thermal characteristics, too. Some CPUs will naturally run 80+ even with good cooling at stock, while others might run 20 or 30c lower than that. 75c sounds pretty good to me, considering a very large overclock.

Degradation is hard to quantify, especially since no 14nm CPU is more than 2 years old. Lower temperatures make more voltage less destructive, but there have been recorded cases of chips spontaneously dying very early on with voltages in excess of 1.4v.

The_Staplergun

Estimable
Jan 30, 2017
1,395
0
2,960
Ok. You say safe, but what's a reasonable temperature I should be at with a 5ghz oc and a water cooler during a cpu stress test? Is it normal for it to sit at 75c?
Also, what's the average temperature where cpu degradation (life span shortening) comes into play?
 
Watercooling varies a lot. Lower end AiO coolers are worse than inexpensive tower coolers, while custom loops ($300+) can significantly better. Each chip will have different thermal characteristics, too. Some CPUs will naturally run 80+ even with good cooling at stock, while others might run 20 or 30c lower than that. 75c sounds pretty good to me, considering a very large overclock.

Degradation is hard to quantify, especially since no 14nm CPU is more than 2 years old. Lower temperatures make more voltage less destructive, but there have been recorded cases of chips spontaneously dying very early on with voltages in excess of 1.4v.
 
Solution

The_Staplergun

Estimable
Jan 30, 2017
1,395
0
2,960
I'm considering playing with a 5.1 oc. Probably just gonna stay with a 5.0.
That was a great answer.

The other question I had is should I run my cache multiplier at my cpu multiplier, or is it fine being at 42 and 50 respectively?
 
A higher cache multiplier will give you more performance, but not as much as core clockspeed. Pretty sure I saw an article a while back that showed 100mhz of cache clock being worth somthing like 25-50mhz core clock.

In other words, if you have to sacrifice CPU clocks to get the cache higher, it's not worth it.
 

The_Staplergun

Estimable
Jan 30, 2017
1,395
0
2,960
So I just put my BCLK Speed to 102. (Which is FSB core clock speed I believe). With a 50 multiplier that puts me at 5.1ghz with a core power setting of 1.375 volts (runs at 1.376). My cpu still doesn't go over 75c on a stress test and it doesn't error. Awesome.
 

The_Staplergun

Estimable
Jan 30, 2017
1,395
0
2,960
Well. I just got a bsod that was more like a green with purple edges. Unexpected kernel mode trap. Looks like I'm not going to be using the FSB to overclock.

Is the error listed here caused from overclock? I was idling the computer when that error hit.