AW 15 R1 throttling (igpu throttling?)

I know about the PCH related issues with throttling. I was getting drops to 800mhz until I did the PCH cooling by thermal pad and copper shim, and I also bridged the copper shim to the gpu heatplate with another thermal pad and some electric tape.

I haven't really noticed until now, but I am still getting throttling down to base clock (2.5ghz) on games like GTAV that use alot of cpu/gpu. The throttling last for ~15 seconds before it goes back to throttle clocks. I redid the PCH cooling by having thermal pad bridge go all the way to bridge to the heatpipes right before the exhaust heatsink. It somewhat improved, but still getting the throttling. Tried the older drivers for intel DTPF and MEI drivers off AW15R1 support driver page, and those improved a little bit, but still throttling.

Since this is AW15R1, I can't read PCH temps dynamically. But when I restart immediately after a long game session and go to Diagnostics, PCH temp is only ~64C. I can't say if this throttling is PCH temp based. I also have the bottom plate open for maximum airflow while testing this. I am also able to undervolt by ~ -85mv and cpu power pkg is only ~41W running GTAV. With an undervolt, it holds throttle clocks for much longer, but still eventually throttles to base clocks. There is no throttling on the 970m. CPU temps and dGPU temps are good.

The weird thing is that throttlestop reports the throttle reason as iGPU power. HWinfo also states its Power limit related to iGPU throttling. I find it weird that its the iGPU as the reason for throttling.

When using the AGA egpu, there is no throttling whatsoever. If I run only heavy cpu benchmarks but with the dGPU on, there is no throttling. Its only when I use the dGPU + CPU intensive applications, there is this base clock throttling. So I am wondering if it is really iGPU throttling or this is actually PCH related.

Do you guys have any insight on this? Really appreciate it.

 

I have a 17r2 and also experienced this I also done the PCM cooling and redone the thermo paste and pads ,
But this is what I get in throttle stop the the GPU limit is red and the cpu down clocks to 2.5ghz but the the dGPU is only at 65c and the cpu is at 78c and total package power peaks at 44w.

My OS is windows10 1903 may update , but was doing this with 1803 .

 

I am in the same situation. dGPU is only 65C ish but the there are those GPU power limit throttles according to throttlestop and hwinfo. In hwinfo, it is seen as DPLT (domain power limit throttling for iGPU). It is not the dGPU or iGPU. The PCH throttle is somehow flagged as iGPU throttling, so thats what throttlestop and hwinfo see it as.

I figured out it was still the PCH temp that is triggering the 2.5ghz throttle. I did the PCH mod with a thermalpad bridge from the copper shim all the way to the heat pipes right before the exhaust heatsink fins. Got rid of 800mhz BDPROCHT throttling, but still get those 2.5ghz throttles in intensive cpu/gpu applications periodically. Since AW15/17R2 can't read PCH temps in windows, its hard to say what the temp is. But I think its hitting that 85C limit which is the first throttle point.

My solution was to use hwinfo to ramp up the fan settings. If I ramp up the fans to aggressive speeds (like 4500-5500 rpm for cpu 75-85C and 4500-5500 rpm for gpu 60-70C), I never get throttling after one hr of GTAV. CPU stays at 3.3ghz on all 4 cores 100% of the time. Since I have the PCH cooling to the GPU fan side, its important to keep the GPU temp down. PCH is ~5watts TDP and its being added to the GPU side of the cooling.

Unfortunately hwinfo EC reading is somehow locking up my system randomly. Still trying to figure out why. Luckily it happens nearly all the time while idling, so I just need to make sure to turn off fan control and exit hwinfo after I'm done.