Weird air cooled CPU temps *need help*

[hs0003]

In the last month or so, I've bluescreened more often than I'd expect, and I've been unable to narrow down the reason. 

So yesterday I checked my CPU temps under load and saw something really strange. (It's a 9700k running an easy 5.1 Ghz overclock with a Thermalright Macho Rev.B tower and some silver based thermal paste)

 

In short, the temp would chill at the 45-55c area while idling or at low load, but when I run kombustors cpu burner software, the temps would spike from that area to 100c within 1-2 seconds, maintain that temp for the duration of the burn and then go back down to the regular temps after 3-4 seconds. The image below shows the usage and temp spikes.

 

From what I understand about heat generation and heat dissipation, the heat curves should be a lot more gradual than shown in the picture. 

 

Does anyone have any idea as to what could be wrong?

ps. I took the tower off and reapplied an appropriate amount of paste, and I'm getting the same results.

 

pps. I know the idle/low load temps aren't good either.

[GoldenLag]

Voltage?

[hs0003]

6 minutes ago, GoldenLag said:

Voltage?

CPU-Z reports 1.472 at idle/low load and 1.344 at high load. The watt is about 50-60 at low and about 185 at 100%

[GoldenLag]

1 hour ago, hs0003 said:

CPU-Z reports 1.472 at idle/low load and 1.344 at high load. The watt is about 50-60 at low and about 185 at 100%

And temps in Cinebench run?

 

Also do you have an AVX offset in place?

[hs0003]

14 minutes ago, GoldenLag said:

And temps in Cinebench run?

 

Also do you have an AVX offset in place?

This is the result of my cinebench run. I don't recall fiddling with AVX offsets

[GoldenLag]

1 hour ago, hs0003 said:

This is the result of my cinebench run. I don't recall fiddling with AVX offsets

I cant actually see anything in this picture, but i assume same jump to 100*C. 

 

What is a likely culprit is thermal density and a lack of AVX offset.

 

I would suggest putting the AVX offset at -2 or -200mhz.  That might help. 

[hs0003]

1 minute ago, GoldenLag said:

I cant actually see anything in this picture, but i assume same jump to 100*C. 

 

What is a likely culprit is thermal density and a lack of AVX offset.

 

I would suggest putting the AVX offset at -2 or -200mhz.  That might help. 

If you click it a few times, you'll get the full resolution image.

 

Even if I reduce the mhz by 200 in AVX calculation tasks, it shouldn't be able to explain the extremely fast and almost instantiations  heat increase and decrease of about 50c, it is after all implying that 300-400 joules of extra heat energy is capable is capable of heating the chip up by 50 degrees which would give the chip a heat capacity of roughly 8 J/K.

[GoldenLag]

40 minutes ago, hs0003 said:

If you click it a few times, you'll get the full resolution image.

 

Even if I reduce the mhz by 200 in AVX calculation tasks, it shouldn't be able to explain the extremely fast and almost instantiations  heat increase and decrease of about 50c, it is after all implying that 300-400 joules of extra heat energy is capable is capable of heating the chip up by 50 degrees which would give the chip a heat capacity of roughly 8 J/K.

The leaves bad mounting which we have eliminated. 

 

Cracked solder (unlikely)

 

Thermal density and the inability to conduct heat fast enough. 

 

How is the chip in stock operation?

[hs0003]

2 minutes ago, GoldenLag said:

The leaves bad mounting which we have eliminated. 

 

Cracked solder (unlikely)

 

Thermal density and the inability to conduct heat fast enough. 

 

How is the chip in stock operation?

I haven't had it down to it's base 4.7 ghz since a few days after I bought it back in late 2018, and back then I just followed OC guides till I got a stable OC that could handle a 100% load and then I turned it down a nudge more.

[GoldenLag]

2 hours ago, hs0003 said:

I haven't had it down to it's base 4.7 ghz since a few days after I bought it back in late 2018, and back then I just followed OC guides till I got a stable OC that could handle a 100% load and then I turned it down a nudge more.

So how is the chip running stock?

 

Because if it gets the same temps then something is very wrong.

[hs0003]

1 hour ago, GoldenLag said:

So how is the chip running stock?

 

Because if it gets the same temps then something is very wrong.

So I jumped back into overclocking and the bios. The default acted basically as expected, but I also found a noob friendly setting to cap the CPU temp at lower than the max safety limit.

 

So now I'm busy trying to see how low I can push the volts before the system complains.

[GoldenLag]

8 minutes ago, hs0003 said:

So I jumped back into overclocking and the bios. The default acted basically as expected, but I also found a noob friendly setting to cap the CPU temp at lower than the max safety limit.

 

So now I'm busy trying to see how low I can push the volts before the system complains.

So...... Again. What is the temp behavior at stock settings?

[hs0003]

20 minutes ago, GoldenLag said:

So...... Again. What is the temp behavior at stock settings?

Sorry, 40-45 ish at idle, a spike up to the high 60's and then slowly climbing to the 75-80c if memory serves.

[GoldenLag]

3 minutes ago, hs0003 said:

Sorry, 40-45 ish at idle, a spike up to the high 60's and then slowly climbing to the 75-80c if memory serves.

No wonder it reaches 100*C then. Does opening sidepanel in the case help woth temps? 9700k isnt exactly a cool CPU

[hs0003]

28 minutes ago, GoldenLag said:

No wonder it reaches 100*C then. Does opening sidepanel in the case help woth temps? 9700k isnt exactly a cool CPU

Sorry, 70-80 at full load

[WoodenMarker]

15 hours ago, hs0003 said:

Even if I reduce the mhz by 200 in AVX calculation tasks, it shouldn't be able to explain the extremely fast and almost instantiations  heat increase and decrease of about 50c, it is after all implying that 300-400 joules of extra heat energy is capable is capable of heating the chip up by 50 degrees which would give the chip a heat capacity of roughly 8 J/K.

A 50c reading doesn't meant the entire chip is 50c. The temps spiking under load is pretty normal unless the cooling solution has a large heat buffer. 

Are clock speeds dropping when it hits 100c? It should throttle when temps are too high and heavy throttling can be what's causing the later drop in temps. 

Did you continually check cinebench scores as you were testing for stability and making sure that the scores didn't stop dropping at some point?