Motherboard: Gigabyte Aorus Elite
CPU: i7-9700k (all cores set to 4.9GHz)
PSU: Chieftec 700w gps-700a8 (> 6-7 y. About to replace with Seasonic 650W FOCUS Gold NEW FOCUS GX-650 SSR-650FX or similar)
Case: AZZA Toledo 301
RAM: Corsair Vengeance LPX
CPU cooler: Noctua NH-D15S (I was weighting pros and cons for the liquid AIO and went with the air. Probably should have chosen the liquid.)
Testing,tweaking.monitoring tools: Intel XTU, Prime95, AIDA, HWinfo
At first applied ZM-STG2 thermal grease, but CPU was hot (but overall case cooling was also bad).
Than applied Noctua's thermal paste with pea drop and vertical line. Than took of the tower to make sure it's spread well and put it back on.
I'll skip case coolers manipulations. The coldest setup I've groped consists in upper front/rear 120mm Noctua (intake/exhaust), top 2 Noctua 90mm (from old cpu cooler).
(Also I had Noctua 200mm on the side, but it's useless - whether it's on or off, max of 2 degrees difference is observed in heavy tests. Probably because of a huge cpu tower. Removed it altogether. My graphics card is perfectly capable of cooling itself anyway.)
On idle temps are 25-30, regular moderate load temps around 60C (video games).
In XTU set all cores at 4.9 GHz, VCore 1.2V. Adaptive offset -0.11V. (No BIOS tweaking).
Prime95, small FFT, no AVX for 20 minutes.
HWinfo reported max Vcore (Max) 1.284V, Core Max (Max) 85C, Max TDP 173W.
Than ran AIDA stress test for an hour resulting in similar temps and voltages.
Rejoiced on the results and launched Fallout 4. After 2 minutes computer shut down.
Started computer, launched Fallout 4 - crashed.
So I thought -0.11 offset was too much. Set to adaptive offset -0.05V.
Prime 95, small fft, no AVX run for 30 minutes.
Hottest core 94, max cpu temp 83, on average 1.303V but spiked to VCore 1.33V, max TDP 204W.
Ran OCCT to load PSU with 2K resolution (no AVX) and in 30 minutes some cores started to throttle.
BTW, it seems that cpu cores throttle after some time (30 mins or more), meaning that heat accumulates with time, i.e. cooling system doesn't manage to dissipate all the heat.
Of course synthetic tests is not the usual usage, but from what I see in reviews and other forums, much better results can be achieved.
Maybe I just lost the silicon lottery?
Or reviewers of cooling systems and cpus are bluntly lying (by hiding some subtle, but important info), like running tests out of the case?
So I'm not sure in which direction I should proceed now and if I should. Even out of sport overclocking interest.
I'll try to reapply grease with the finger maybe, so that resulting layer will be as thin as possible?
Intel recommends 1.37V for this chip at 4.9GHz, so I'm already below that voltage. As at 1.37 cpu will throttle in few minutes.
CPU: i7-9700k (all cores set to 4.9GHz)
PSU: Chieftec 700w gps-700a8 (> 6-7 y. About to replace with Seasonic 650W FOCUS Gold NEW FOCUS GX-650 SSR-650FX or similar)
Case: AZZA Toledo 301
RAM: Corsair Vengeance LPX
CPU cooler: Noctua NH-D15S (I was weighting pros and cons for the liquid AIO and went with the air. Probably should have chosen the liquid.)
Testing,tweaking.monitoring tools: Intel XTU, Prime95, AIDA, HWinfo
At first applied ZM-STG2 thermal grease, but CPU was hot (but overall case cooling was also bad).
Than applied Noctua's thermal paste with pea drop and vertical line. Than took of the tower to make sure it's spread well and put it back on.
I'll skip case coolers manipulations. The coldest setup I've groped consists in upper front/rear 120mm Noctua (intake/exhaust), top 2 Noctua 90mm (from old cpu cooler).
(Also I had Noctua 200mm on the side, but it's useless - whether it's on or off, max of 2 degrees difference is observed in heavy tests. Probably because of a huge cpu tower. Removed it altogether. My graphics card is perfectly capable of cooling itself anyway.)
On idle temps are 25-30, regular moderate load temps around 60C (video games).
In XTU set all cores at 4.9 GHz, VCore 1.2V. Adaptive offset -0.11V. (No BIOS tweaking).
Prime95, small FFT, no AVX for 20 minutes.
HWinfo reported max Vcore (Max) 1.284V, Core Max (Max) 85C, Max TDP 173W.
Than ran AIDA stress test for an hour resulting in similar temps and voltages.
Rejoiced on the results and launched Fallout 4. After 2 minutes computer shut down.
Started computer, launched Fallout 4 - crashed.
So I thought -0.11 offset was too much. Set to adaptive offset -0.05V.
Prime 95, small fft, no AVX run for 30 minutes.
Hottest core 94, max cpu temp 83, on average 1.303V but spiked to VCore 1.33V, max TDP 204W.
Ran OCCT to load PSU with 2K resolution (no AVX) and in 30 minutes some cores started to throttle.
BTW, it seems that cpu cores throttle after some time (30 mins or more), meaning that heat accumulates with time, i.e. cooling system doesn't manage to dissipate all the heat.
Of course synthetic tests is not the usual usage, but from what I see in reviews and other forums, much better results can be achieved.
Maybe I just lost the silicon lottery?
Or reviewers of cooling systems and cpus are bluntly lying (by hiding some subtle, but important info), like running tests out of the case?
So I'm not sure in which direction I should proceed now and if I should. Even out of sport overclocking interest.
I'll try to reapply grease with the finger maybe, so that resulting layer will be as thin as possible?
Intel recommends 1.37V for this chip at 4.9GHz, so I'm already below that voltage. As at 1.37 cpu will throttle in few minutes.
Last edited: