Undesired Power Throttles

I have described the situation on the AW17 R5 forum, and will also try my luck here.

In the past my CPU, on two same laptops, would run hot until it burned itself, and then thermal throttles would kick in to salvage the situation. The CPU would reach and sometimes exceed 100C before the thermal throttles. Then, depending on load, it would sometimes oscillate back and forth between 80C and 95C-100C. This is a very common scenario and it happens to a lot of (unlucky) people.

Last few days I have noticed a completely new behaviour. Now when you shock the CPU it reaches 100C just like it did before, and thermal throttles cut in. After a few seconds (2-3) then a power throttle appears and the thermal throttle disappears. For the duration of the stress test, this power throttle remains pretty much constant, or it comes and it goes, but no more thermal throttles. The temperature stays religiously below 94C, and as soon as it hits 94C the power throttle re-appears or, if already on, the power throttle reduces more power. It manages to keep the CPU *just* below 94C perfectly.

Finally to get the thermal throttles to re-appear you can go to idle and wait a bit, as if letting the system reset itself. Or you can trick it by loading only 2-3 cores out of the 6. Now the temps rise again, but only on a few cores, and the power throttle cannot manage to do anything anymore, and now we get our thermal throttles back mixed with power throttles.

I do not think the above is "standard" behaviour because I have not chosen anywhere this specific 94C trip temperature. Thermal throttles generally appear at over 98C, so there are at least another 4-6C the CPU could go up to until the thermal throttles appeared - edit: and then it would be thermal throttles not power throttles.

So who or what has implemented a 94C thermal limit implemented by a variable power throttle (40W, 50W, 60W, whatever it takes to keep it below 94C)?

doofus99 Notebook Deity

ColaMonster Notebook Consultant