macbook pro 15" 2011 throttles CPU when plugged to external screen

[grrrz]

so to expand what's in the title; It become really apparent because of the heat wave ; I mostly use my computer plugged to a external monitor with an HDMI adapter; and there is a staggering difference in temperature/thermals management/noise between standalone usage and when a screen is plugged (I don't have a VGA or display port monitor for comparison but I should also try this).
basically in standalone the fan kicks in lightly (4000 RPM) when working on a standard project (ableton live); and mostly stay quiet (2000 RPM) for web browsing/videos/etc. The CPU temps is around 50°C idle and doesn't go over 75. GPU temp is around 60°C. When the screen is plugged; not only does the gpu runs hotter (which is expected); but the CPU takes a least a 10°C hike while the fan go full gear as soon I have two tabs open.
When I try to do anything a bit CPU intensive (30/40% overall with the turbo) the CPU starts to severely throttle (I mean to 1,3 Ghz or so when the base clock is 2.2 Ghz after a few minutes). I've given up on working this way.
Doing the exact same task without the screen gets me a constant 2.8 Ghz turbo and about 80°C temperatures.
I don't understand how driving an external screen causes such differences (I get that the cooling system is shared; but this is a lot).
On a specially hot day, I also got the CPU temp limited to 70°C; and it would throttle to a crawl beyond this point (instead of going hotter); just using firefox was laggy as hell.
also, there's no additional cpu usage when I plug the screen, so it's not that. it just gets hotter. I know a dusting is overdue but still, that's pretty mysterious to me. These thermals and noise problems are driving me crazy on those machines. (and apparently the newer models have problems too)

 

[jerryk]

External monitor runs GPU. So now system has to dissipate heat from CPU and GPU. The system was already already dealing with the CPU heat prior to cranking up the GPU. Now you added the heat from the a really hot GPU driving the external display. Once it gets too hot, the system has to slows the clock speed to reduce the heat from the CPU and get the thermals back under control. And remember this system was plagued by GPU issues because of excess heat.

You might try redoing the thermal paste and getting rid of any dust. Also, is the screen open or closed? You might also try to disable the GPU. Lot of posts on this. But I not sure if that kills the external monitor

 

[grrrz]

Yeah that makes sense.
The problem is the system sometimes defines "too hot" as the cpu being 70°C and start throttling from this point. With no screen attached it merilly go up to 95°C with turbo before even going back to base clock speed (and stay there for good).
the screen is open. I could try clamshell (or something with a specialized software to desactivate the internal screen). I don'think you can use any external monitor without using the dGPU (tried this with gfxcardstatus and you get nothing when selecting iGPU only). I'll do some dusting with canned air but for thermal paste I better find a professional to do it, it's a way too delicate operation to risk it.

 

[jerryk]

Yeah that makes sense.
The problem is the system sometimes defines "too hot" as the cpu being 70°C and start throttling from this point. With no screen attached it merilly go up to 95°C with turbo before even going back to base clock speed (and stay there for good).
the screen is open. I could try clamshell (or something with a specialized software to desactivate the internal screen). I don'think you can use any external monitor without using the dGPU (tried this with gfxcardstatus and you get nothing when selecting iGPU only). I'll do some dusting with canned air but for thermal paste I better find a professional to do it, it's a way too delicate operation to risk it.

The CPU is getting to 70C, but with the dGPU running the total (CPU + dGPU) thermal load is much higher, so they need to throttle the CPU sooner.

 

[grrrz]

The CPU is getting to 70C, but with the dGPU running the total (CPU + dGPU) thermal load is much higher, so they need to throttle the CPU sooner.

this is still weird; the CPU alone can go to 100°C before safely throttling; wether the dGPU is in use or not; 70°C seems ultra conservative, mind you it happens usually during really hot days. Maybe the factor in the temperature of the heatsink and trigger a more agressive throttling when it gets too hot?

 

[jerryk]

this is still weird; the CPU alone can go to 100°C before safely throttling; wether the dGPU is in use or not; 70°C seems ultra conservative, mind you it happens usually during really hot days. Maybe the factor in the temperature of the heatsink and trigger a more agressive throttling when it gets too hot?

It does make sense that the ambient temperature would be a factor.

But a GPU can generate more heat than the CPU. Look at the massive heat sinks and multiple fans on any GPU card. My deskside workstation has 3 fans and a 7 inch heatsinks on each of the GPU cards. Much more cooling than that required by the 8 core i7-9900K CPU.