Strange temperature changes.

[StorAxis]

Hello,

Recently I helped a friend build a gaming computer. All was going well until recently when he noticed that his temperatures were getting uncomfortably high, after no modification had been made to the internal components, after around 2 weeks of its installation.

 

The CPU is an i5-4590, and the Motherboard is a Gigabyte H97-HD3.

Initially, the CPU was idling in the low 30's©, and got up to low 60's© when gaming (What you'd expect)

Now, when running Prime95 SmallFTT, we have gotten the CPU up to 97© after less than a minute of running the test. I imagine it would get higher but I terminated the test to avoid damage to the CPU. I assumed the Thermal paste might be to blame, so I re-applied some aftermarket CoolerMaster Thermal compound and the issue was not resolved, and I tried applying it in different sizes, as well as attempted to re-seat the cooler numerous times.

I also tried an aftermarket cooler, the H80i, and the exact same thing happens. The temperatures skyrocket extremely quickly, often reaching uncomfortable temps in under 5 seconds!

 

I'm at a complete loss, all that he has done on the software side since its installation is change the BIOS version, but I doubt that would be causing this issue?

If anyone has some info on how this could be fixed we'd really appreciate it!

 

[Sousuke]

I am thinking of that issue that popped up a year or so ago when the thermal solution which took heat from the die to the heat spreader on the CPU Itself was bad this could be the issue.

 

Best bet is to RMA the CPU.

[Toxicable]

snip

i'd suggest your try RMA, CPUs can jump 10s of degrees in seconds but not straight to 97, considering that's it's thermal limit (I think ).

[StorAxis]

i'd suggest your try RMA, CPUs can jump 10s of degrees in seconds but not straight to 97, considering that's it's thermal limit (I think ).

Did some research afterwards and apparently the top temp for the CPU is 72C, didn't freeze or anything though which is what I have experienced when other CPU's get too hot.

[Toxicable]

Did some research afterwards and apparently the top temp for the CPU is 72C, didn't freeze or anything though which is what I have experienced when other CPU's get too hot.

Im not sure what you mean by top temp. There are a few different stages of thermal limits on CPUs, there's the point at which it will then throttle, I assumed 97, then the point at which the system will shutdown, 105, and the point where it will cause irreversible damage 110.

THose numbers aren't the ones for your CPU but for the older gen CPUs, either way 72 degrees is far too low for a thermal limit, especially considering that yours is going much higher, if 72 degrees was then it would have just shut down to prevent damage

[StorAxis]

Im not sure what you mean by top temp. There are a few different stages of thermal limits on CPUs, there's the point at which it will then throttle, I assumed 97, then the point at which the system will shutdown, 105, and the point where it will cause irreversible damage 110.

THose numbers aren't the ones for your CPU but for the older gen CPUs, either way 72 degrees is far too low for a thermal limit, especially considering that yours is going much higher, if 72 degrees was then it would have just shut down to prevent damage

Ah yeah, thanks for clearing that up. I was uncertain. I think the CPU would continue to rise in temp if I kept the test going, which I don't think I'm brave enough to do to a friend's computer haha. I might run Unigine Heaven or something, and see if that test gives me more acceptable results for a realish life scenario, the machine will never reach 100% usage in games I assume.

[Toxicable]

Ah yeah, thanks for clearing that up. I was uncertain. I think the CPU would continue to rise in temp if I kept the test going, which I don't think I'm brave enough to do to a friend's computer haha. I might run Unigine Heaven or something, and see if that test gives me more acceptable results for a realish life scenario, the machine will never reach 100% usage in games I assume.

That is true, but im fairly certain there is still something wrong and you're right not to let it go on lol, no way should a CPU run that hot

[Elapo]

Have you tried more than a single monitoring program?

 

Sensor could also be broken, but I'd RMA it nonetheless.

[StorAxis]

Have you tried more than a single monitoring program?

 

Sensor could also be broken, but I'd RMA it nonetheless.

Yeah, have tried SpeedFan, CPU-Z as well as corsair link. You may be right about the sensor, as the air coming off the heatsink doesn't feel any hotter than it did when the temps appeared to be normal. (That's an accurate way to measure, right? )

[Ollof]

You're running haswell in prime95, that is the problem. You shouldn't use p95 as it tries to pump more voltage into the cpu and it messes with haswell. prime95 is a worst case torture test, use AIDA 64 or even the intel tuning utility.

 

Your gaming temps are normal, and 97c wont damage your cpu as the limit is 100c at which point it would throttle then shut down to protect itself.

[Blackie Sheen]

-snip-

 

You're running haswell in prime95, that is the problem. You shouldn't use p95 as it tries to pump more voltage into the cpu and it messes with haswell. prime95 is a worst case torture test, use AIDA 64 or even the intel tuning utility.

 

Your gaming temps are normal, and 97c wont damage your cpu as the limit is 100c at which point it would throttle then shut down to protect itself.

Make sure that if you use prime95 with haswell that the voltage dlivery is set to manual in your bios or someting similar because if not, Prime95 can push you cpu to 1,375 volt which way higher than you would ever want to run the chip and which makes the chip run super hot. You should rather use Intel XTU or Aida 64 has a stability test

[Syfes]

Make sure that if you use prime95 with haswell that the voltage dlivery is set to manual in your bios or someting similar because if not, Prime95 can push you cpu to 1,375 volt which way higher than you would ever want to run the chip and which makes the chip run super hot. You should rather use Intel XTU or Aida 64 has a stability test

 

This. Re-run your stress tests using one of these utilities and see what happens. There's a chance that you'll get much more managable temperatures. Still that chip is running at stock speeds, I can't imagine even Prime95 causing temperatures this high? If you get similar high temperatures using AIDA64 or Intel XTU then I would consider RMA'ing the chip. 

[Queek]

What the above two posts say.

60°C is a very high idle, but the way that you confirmed that there is something wrong was flawed (prime95 on not manual voltage).

Also, the maximum temperature for this chip is 100°C (so you were seeing thermal throttling). The maximum recommended operating temperature is 72°C (ie. When overclocking, you want your chip to be below this temperature most of the time).

[runit3]

The 72C is in reference to Intel's max TDP that they will warranty for normal operating temperature on the stock cooler. P95 does not fall under normal operating conditions, but gaming certainly does. Double check that manual voltage is on and do some more reasonable tests like XTU and AIDA.

[LogicalDrm]

The 72C is in reference to Intel's max TDP that they will warranty for normal operating temperature on the stock cooler. P95 does not fall under normal operating conditions, but gaming certainly does. RMA ASAP.

Gaming temps were at 60C. This is propably just P95 being naughty. But we shall see after Aida, XTU, IBT and maybe OCCT results.

[Jumper118]

Hello,

Recently I helped a friend build a gaming computer. All was going well until recently when he noticed that his temperatures were getting uncomfortably high, after no modification had been made to the internal components, after around 2 weeks of its installation.

 

The CPU is an i5-4590, and the Motherboard is a Gigabyte H97-HD3.

Initially, the CPU was idling in the low 30's©, and got up to low 60's© when gaming (What you'd expect)

Now, when running Prime95 SmallFTT, we have gotten the CPU up to 97© after less than a minute of running the test. I imagine it would get higher but I terminated the test to avoid damage to the CPU. I assumed the Thermal paste might be to blame, so I re-applied some aftermarket CoolerMaster Thermal compound and the issue was not resolved, and I tried applying it in different sizes, as well as attempted to re-seat the cooler numerous times.

I also tried an aftermarket cooler, the H80i, and the exact same thing happens. The temperatures skyrocket extremely quickly, often reaching uncomfortable temps in under 5 seconds!

 

I'm at a complete loss, all that he has done on the software side since its installation is change the BIOS version, but I doubt that would be causing this issue?

If anyone has some info on how this could be fixed we'd really appreciate it!

the board is probably just over compensating with the voltage. 

[runit3]

Gaming temps were at 60C. This is propably just P95 being naughty. But we shall see after Aida, XTU, IBT and maybe OCCT results.

My brain hadn't caught up to my eyes that early >_<

*edited earlier*

[Queek]

My brain hadn't caught up to my eyes that early >_<

*edited earlier*

Yeah I was wondering what you were on about 

[runit3]

Yeah I was wondering what you were on about 

It amazes me how much I can accomplish at 5:30am, and how much I miss. This fell into the latter.

[StorAxis]

Alright -- Sorry for the late response, typical Australian internet went down for 8 hours after I downloaded Aida64.

With Aida64 stress test I reached a maximum CPU temp of 77c, and the average was around ~70c for the majority of the test.

EDIT: Forgot to add that the graph shows that there was no thermal throttling throughout the 2 hours I ran the test for.

[LogicalDrm]

Alright -- Sorry for the late response, typical Australian internet went down for 8 hours after I downloaded Aida64.

With Aida64 stress test I reached a maximum CPU temp of 77c, and the average was around ~70c for the majority of the test.

EDIT: Forgot to add that the graph shows that there was no thermal throttling throughout the 2 hours I ran the test for.

Those would be the normal stress test temps. So your original issue was just P95 being too much for Devils Canyon chip. Doesn't H97 have like auto only with voltage? That would mean that P95 is raising them too much.