CPU TDP question for the pro's, maybe

[carbide]

I've got some better TIM on my CPU now, temps under control. I've been trying to find my max core temperatures (stock) under full load using OCCT for small ffts and msi kombustor to stress the graphics card out.
The test lasts for about 4 mins or so, temps looking fine - GPU at 68c hottest core at 80c then all of a sudden the CPU multiplier drops down to 8x.

Does exceeding TDP cause this? when I had a really poor TIM application from the manufacturer temps got to TJmax before dropping the multiplier.
Only thing I can think of is that I've swapped the ram for same quantity but slightly faster (1333 to 1600)
But I assume the problem is load related?

 

[ewok93]

Sounds like it could be the temperature, 80c is pretty hot.

 

[carbide]

sorry guys, some background...

This is a laptop with an nvidia 765m chip (hard soldered to the m/board (hm87 i think)) and a 4700MQ quad core i7

I bought it from a specialist pc builder, the laptop is a clevo.

when I first got it out of the packaging, and loaded it up (CPU small ffts, kombustor for the nvidia) three cores logged 'hot' in real temp, so hit TJ max and the throttling occured i.e. multi dropped to 8x and crippled the cpu speed to 800mhz.

Whch was fine, as I realised it was hitting TJmax. After investigation there were spots about the dia. of a pea that were perfectly clean - TIM had never touched, so I've been re pasting with different pastes and techniques to get the temps down.

Now, testing only the CPU it balances the multi at about 27x for four cores at the max sustained TDP of 47w and remains what I consider cool at 75c.

If i then use the graphics burn in test whilst the CPU test is running, the CPU multiplier drops to about 24-25x and the nvidia GPU temp rises from 40c to about 68c

After a few mins the hottest CPU core temp gets into the low 80s then all of a sudden the multi drops to 8x

So, bouncedk do we think that the mosfets on the board are incapable of sustaining the current draw of both the dedicated GPU and CPU?

Two photos below show the heatsink assembly for the laptop - the white strip sits on top of 4 small black chips - are these the mosfet voltage regulators?
https://dl.dropboxusercontent.com/u/44236923/2013-11-06%2007.20.08.jpg

https://dl.dropboxusercontent.com/u/44236923/2013-11-06%2007.20.19.jpg

here is a stress test I was getting before I repasted, I think the throttle was related more to the temp than TDP, but maybe not
https://dl.dropboxusercontent.com/u/44236923/Clevo%20W370ST%20full%20load%20temps.png

Should I replace this sticky white stuff with a strip of copper and thermal paste?

or just appreciate that i'm trying to pull too much power from the board?

I've already used intel XTU tool to drop the voltage 50mV offset from the CPU and cache to get temps down...