I can't solve an issue with my TR 1950x. I've got the cpu in a custom loop and I run it in a slight OC state via Ryzen Master, @ 3.8 Ghz and 1.23v. At 100% usage however HWINFO reports, thermal throttling, and the CPU throttles down for a brief moment to like 0.8v and then back again to the 1.23v. However the Tdie temps are under control, I still get the HTC and Prochot EXT to show up.
I attached two snapshots, one in which it is not throttling and chugging along and the other with throttling. On the far right column you can see the average over >12h of constant 100% usage.
Can anyone make any sense of this?
What does Ryzen Master show? Any thing showing in RED at the very top?
This HWINFO thread people are having the exact same issue where it shows the CPU overheating yet the CPU is under its throttling thermal limits: https://www.hwinfo.com/forum/threads/what-does-the-htc-in-thermal-throttling-htc-mean.6073/
Make sure you have the latest BIOS/UEFI and CHIPSET installed for your motherboard. then check HWINFO again.
This Quora User who is an engineer explains what THERMAL THROTTLING (PROCHOT EXT) means:
I found an answer here:
“BD PROCHOT stands for bi-directional prochot. PROCHOT stands for processor hot which is the signal that is activated within the CPU when it reaches approximately 100C to 105C depending on the model number. This signal is what initiates thermal throttling so the CPU can slow down and keep from over heating. Intel included a bi-directional feature so if something else like a GPU is running too hot, it would be able to send a PROCHOT signal directly to the CPU and force it to cool down so the entire laptop cools down. Very few laptops seem to use this type of throttling. This feature was added for the Asus G51. Disabling this will allow your CPU to continue to run at full speed. Disabling this will not prevent your CPU from thermal throttling at its normal Intel set thermal throttle temperature. By default this is locked and you will need to go into the Options window to unlock it so you can toggle it on and off.”
It is a means for some other management controller on the system motherboard to send a PROCHOT signal to the CPU to cause it to throttle to minimum frequency, usually because something else on the system motherboard is going to overheat and the management controller wants to avoid this from happening. The reason why it commands the CPU to throttle, even if the CPU is not overheating, is the CPU is usually a significant if not the largest power component in a computer, so throttling it to minimum frequency will reduce the total power consumption and thermal contribution of the CPU to the system significantly which can help avoid the other component from overheating.
Two common possibilities are the CPU Voltage Regulator and the GPU. Another possibility is the main PSU.
Reading the other answers, I would say, it’s not to protect the CPU from overheating, but to protect some other part of the computer that either cannot self-throttle or even if it did, would not significantly lower its own temperature. Good candidates for these would be the voltage regulators, i.e. the CPU VR cannot self-throttle without crashing the CPU, obviously (it can’t just stop powering the CPU). But the CPU VR temp can be detected and the BD PROCHOT sent to the CPU if the VR is overheating, to ask the CPU to drop performance and thus power, and thus ease off the load on the VR.
The reason I say that is, the CPU itself is designed to self-PROCHOT when it gets too hot, and designed to self-THERMTRIP if it really gets too hot. The PROCHOT threshold on an Intel CPU is typically around 100 C, and the THERMTRIP threshold around 115 C.
The THERMTRIP works by sending a signal to the chipset that controls the power on/off of the computer, whereby the chipset is supposed to respect this signal and power off the CPU, and the rest of the computer, into the off state where only very few devices are running, i.e. usually just the power button monitor.
The built-in PROCHOT and THERMTRIP mechanisms of the CPU are not meant to be overriden by the end-user so I doubt even a BIOS mechanism will be exposed to let you bypass this to essentially let the CPU “fry” itself from an over-temperature.
Instead, if BD PROCHOT is available, it is an intentional designed option for the board system management controller to, in an emergency scenario, get the CPU to lower its power, to avoid some other portion of the board, from going into overtemp.
Well, because this BD PROCHOT is not built-into the CPU, it can be disabled. The CPU will STILL self-throttle (as the built-in CPU PROCHOT/THERMTRIP cannot usually be turned off) and protect itself, but when whatever other component is going to overtemp, well, we’ve turned off that from going to throttle the CPU. One possibility you might want to do this, is if the overtemp warnings from those components is false (i.e. the sensor is broken) and you effectively want to turn that false warning from dropping CPU performance. But then of course you take on the risk that without that protection mechanism you could still be causing those components to run above their safe operating temperature, leading to potential crashes or permanent hardware damage.
Seems to be more of HWINFO bug than anything else.
Don't think it's a HWINFO bug because the CPU actually undervolts itself and underclocks accordingly to something like 600Mhz and 0.8v for a brief moment (2-3 seconds) and then jumps back to the preset voltages and clocks. It does this continuously.
Ryzen master doesn't show anything in red and I attached a screen shot.
Throttlestop doesn't work on this CPU, gives an error on launch, CPU not supported!
I have the latest BIOS firmware from Asus for the ROG Zenith Extreme mobo, bios no 2001.
For what it's worth, I kind of found the culprit, I think....
For future reference or for whom it may help, my system has 4 Titan V GPUs in it, and if I run the system 100% load (CPU + GPU) and I go above 100% TDP on the GPUs, I start to see the throttling in question. If I leave them at 100% or below, the CPU doesn't throttle. Don't know what's the problem exactly, but I suspect it must be voltage related. The GPUs are within good temps as they are watercooled too, so it must be either a residual heat that escapes into the chassis, and triggers something that triggers the PROCHOT EXT (VRMs are good too btw, at around 64C but still good) or it's voltage related and/or mobo design.
Besides a private supply of liquid nitrogen, the bank of video cards explains why you box is so hot
if you need that much compute, use a rack based setup which can handle more cards if needed
It doens't add up. I guess is very little difference between 100% TDP of the and anythign above 100%. So it can't be that much heat. It must be voltage related. Again: The CPU and GPUs are cooled very good by the cooling solution