Normal for prime95 to overheat a cpu?

[SteinerW]

For started my 4790k is cooled by a custom loop with a 360 rad. I use an EK nickle plated block.

 

I was playing with my cpu today and decided to see what the sensor test in RealTemp did. I started prime95 and my system shuts off after 2 seconds from thermal protection.  I was running 4.6 at 1.29 volts.

I found this kind of odd/worrisome, so I set everything to stock speeds. Everything still got to 100 degrees in prime95, but the system stayed running.

 

The temps in some other programs at stock speeds are

 

65 while rendering a video

73 in furmark's cpu burner

69 in aida 64

65 in 3dmark firestrike's physics test

 

 

Is this a normal thing to see or is my cpu/motherboard off in some way?

 

[Principis]

10 minutes ago, 3DGamerOnSteam said:

For started my 4790k is cooled by a custom loop with a 360 rad. I use an EK nickle plated block.

 

I was playing with my cpu today and decided to see what the sensor test in RealTemp did. I started prime95 and my system shuts off after 2 seconds from thermal protection.  I was running 4.6 at 1.29 volts.

I found this kind of odd/worrisome, so I set everything to stock speeds. Everything still got to 100 degrees in prime95, but the system stayed running.

 

The temps in some other programs at stock speeds are

 

65 while rendering a video

73 in furmark's cpu burner

69 in aida 64

65 in 3dmark firestrike's physics test

 

 

Is this a normal thing to see or is my cpu/motherboard off in some way?

 

I don't know but I know that my 6700k gets REALLY hot  with prime95 that's why many people are warning to not use it.

 

Edit: just tested: after a few seconds 99C on 6700k 4.7 GHz 1.355V in bios (when idle 1.375V)

[DocSwag]

Prime95 WAYYY overstresses your CPU, it makes it way hotter than you'll ever see, even from other CPU benchmarks or stress tests. Something like Aida64 or Asus Realbench will make sure your CPU is stable without putting it on fire. Ever since one of the updates Prime 95 has been a bit of a overdone stress test.

[aisle9]

10 minutes ago, 3DGamerOnSteam said:

For started my 4790k is cooled by a custom loop with a 360 rad. I use an EK nickle plated block.

 

I was playing with my cpu today and decided to see what the sensor test in RealTemp did. I started prime95 and my system shuts off after 2 seconds from thermal protection.  I was running 4.6 at 1.29 volts.

I found this kind of odd/worrisome, so I set everything to stock speeds. Everything still got to 100 degrees in prime95, but the system stayed running.

 

The temps in some other programs at stock speeds are

 

65 while rendering a video

73 in furmark's cpu burner

69 in aida 64

65 in 3dmark firestrike's physics test

 

 

Is this a normal thing to see or is my cpu/motherboard off in some way?

 

Those numbers seem odd. My 4790K, for example, is at 4.7GHz, 1.265v with a Kraken X61 cooling it, and it rarely breaks 75C on Prime95 v26.6. The rendering, Furmark, AIDA64 and Firestrike numbers don't seem too far off, but Prime95 seems very high, even given your increased voltage. What version are you using?

 

It does look to be within silicon lottery range, though.

[beach_boy98]

1 minute ago, aisle9 said:

-SNIP-

Prime95 is actually really bad for haswell based CPUs. I'd avoid it.

[Imakuni]

19 minutes ago, 3DGamerOnSteam said:

I was running 4.6 at 1.29 volts.

But was it REALLY running at that? You could just have set it to 1.29, but in reallity be runnning much higher if LLC wasn't manually set.

 

At any rate, though, 1.25v used to be the max recommended voltage for Haswell CPUs. You'll see people nowadays recommending more because they got used to it due to Skylake's higher voltages, but before that, 1.25 was what was considered the "limit'. If you take into account how much you've gone over that, it kinda makes sense that you are getting the temps you do.

 

Assuming it's running at the actual voltage and/or a voltage drop is not possible, you really only have 2 options: A- Lower your OC so that you can lower voltage as well; B- Resort to delid methods to get better temps.

 

3 minutes ago, aisle9 said:

Those numbers seem odd. My 4790K, for example, is at 4.7GHz, 1.265v with a Kraken X61 cooling it, and it rarely breaks 75C on Prime95 v26.6.

Ofc you'd have cool number by using the potato version of P95... use latest and you'll see temps skyrocket (see comment above about voltage limits).

 

Just now, beach_boy98 said:

Prime95 is actually really bad for haswell based CPUs. I'd avoid it.

Define "bad".

[beach_boy98]

3 minutes ago, Imakuni said:

-SNIP-

http://www.overclockers.com/forums/showthread.php/751197-Prime-95-v28-5-Haswell-CPU-s-and-you-is-it-bad-for-the-CPU-s-health

[aisle9]

6 minutes ago, beach_boy98 said:

Prime95 is actually really bad for haswell based CPUs. I'd avoid it.

26.6 is fine for Haswell. It's later versions that bone it like a vibrator covered in hot sauce.

5 minutes ago, Imakuni said:

Ofc you'd have cool number by using the potato version of P95... use latest and you'll see temps skyrocket (see comment above about voltage limits).

...that's kind of why I use 26.6. I don't like the idea of my $300 CPU being turned into bacon pancakes.

 

Don't get me wrong, OCCT and RealBench are my preferred tests, but nothing tells you what your CPU can handle faster than a couple hours under P95 load.

[Imakuni]

3 minutes ago, beach_boy98 said:

http://www.overclockers.com/forums/showthread.php/751197-Prime-95-v28-5-Haswell-CPU-s-and-you-is-it-bad-for-the-CPU-s-health

I asked for your definition of "bad", not an explanation as to why it's "bad" (whatever that might mean...).

 

4 minutes ago, aisle9 said:

...that's kind of why I use 26.6. I don't like the idea of my $300 CPU being turned into bacon pancakes.

Can we both agree that it's only going to burn if you set the voltage too high, in which case it would be the user's fault, not the program?

[aisle9]

Just now, Imakuni said:

Can we both agree that it's only going to burn if you set the voltage too high, in which case it would be the user's fault, not the program?

Completely. Most of the problems with P95 stem from people running later versions of the program at 4.6GHz+ on adaptive voltage with a POS 212 EVO cooling their 4790K.

[SteinerW]

17 minutes ago, Imakuni said:

But was it REALLY running at that? You could just have set it to 1.29, but in reallity be runnning much higher if LLC wasn't manually set.

 

At any rate, though, 1.25v used to be the max recommended voltage for Haswell CPUs. You'll see people nowadays recommending more because they got used to it due to Skylake's higher voltages, but before that, 1.25 was what was considered the "limit'. If you take into account how much you've gone over that, it kinda makes sense that you are getting the temps you do.

 

Assuming it's running at the actual voltage and/or a voltage drop is not possible, you really only have 2 options: A- Lower your OC so that you can lower voltage as well; B- Resort to delid methods to get better temps.

 

Ofc you'd have cool number by using the potato version of P95... use latest and you'll see temps skyrocket (see comment above about voltage limits).

 

Define "bad".

I had i manually set to 1.29 volts. this shutdown from thermal protection

 

The rest of my temps are with all stock setting.

[Imakuni]

3 minutes ago, 3DGamerOnSteam said:

I had i manually set to 1.29 volts. this shutdown from thermal protection

I know that. What I want to know is if the voltage you add was the actual voltage being applied; for instance, I dial 1.275v on my, but get 1.296 as the actual supplyed votlage. Once, I put LLC at the wrong level and it shot up to 1.35 or something, 80mv more than what I told it do. If that's happening to you, it could explain the problem.

[airdeano]

also, you can dial the synthetic load through the custom menu:

you can change the thread use to 8 since you are using the i7 CPU.