Ryzen 5800x 87c in Cinebench...

[Gabor Vajek]

Op has tried different coolers and re-seating? more case fans? Maybe not enough air flow.

none of these aio-s are thermally limited by their radiator or fluid temperatures...under absolute full p95 load 50% fanspeed makes no difference to 100%, the radiator and the tubes are both warmish to the touch, not even approaching hot. these aios are all limited by pumpspeed. adding more fans woudnt make any differendce henceforth. I had push-pull on my x62 under which i remounted the coldplate on the cpu like 4times, then moved onto the arctic freezer 280, 2 fans and seeing the exact same temps. some chips just loose the silicon lotto is all there is to this.

 

[DiceAir]

What the op must try an dthat's also a thing I did was pressing on the pump just slightly and then turning the screws slightly to make sure there is proper contact. Actually 86C in cinebench is not bad depending on ambient temp. For me I'm from south africa and it was raining last night and today it's a bit cooler and before it was raining temps never saw above 83-85C with now that it's a bit cooler I see 80C. I have a 3d printer so if I can make a bracket to fit in the 5.25" bay then I can have another cougar vortex pwm fan that I took from the back and have even lower temps on the cpu as it will be closer to the front fan of the cpu so will have much fresher air.

What I can suggest is do a warranty claim on the cpu

I wish I bought the new fractal design celsius+ s36 cuase it has a better pump so I think that will also help temps out a bit. I think rather than spending large amount of money why not returning the chip and get the 5900x

 

[Acolyte]

Just as a comparison point for some...

5800X @ CO +150Mhz (Peak single core bursts 5000Mhz), with a Negative 10 offset on all cores. Motherboard max unhinged PPT settings e,t,c. R20 Draw 150w.

Custom loop, EK Block, 2x360 Rad. Conductonaught liquid metal.

R20 Score - 6250, max temp 72 Degrees over 10 runs (room ambient 22). Sustained 4,675 - 4,700Mhz on all cores.

This looks to be a smidgen better than most at 80+ degrees but does show how the PBO algorithm will scale better performance with lower temperature. It's still hot though and ultimately as said above its probably a silicon level bottleneck with the way the PBO2 runs and smashes large voltages quickly though the chip at light loads for high clock speed single threaded reliability.