No it's not harder unless the CPU can blow up within 60 sec, you can let the CPU increase temps and deal with the heat after it clocks down again.
How much thermal mass do you think a CPU
has?
And it's non-trivial to transfer 350 W of heat into a heatsink. I'm reminded of Toms' testing of Skylake 18-core HEDT CPU, where they found the bottleneck to keeping it cool was actually in the CPU package, itself.
Here's a really solid writeup of just how challenging it is to cool a CPU, when you get in the neighborhood of 300 W. It's not pretty.
https://www.tomshardware.com/reviews/-intel-skylake-x-overclocking-thermal-issues,5117.html
Here's what happens at just 300 W:
"Now, can we hit the 300W mark without our system dying a fiery death using our trusty Alphacool Eiszeit 2000 Chiller? To make a long story short: yes, we can...for about 10 minutes. After that, the voltage converters hit 105°C and the CPU’s frequency drops to 1.2 GHz with a power consumption of 70W. That’s enough for the components to recuperate quickly, after which the whole sequence starts over, resulting in an endless loop.
So close, but no cigar. Then again, who else out there uses an Alphacool Eiszeit 2000 Chiller? Most folks won't be able to push their system to the brink of death like this without delidding their processor first. Even with our high-end cooling solution, the Tpackage measurement exceeds 100°C, while the cores run at 94°C and up. This can’t be called sufficient cooling by any stretch of the imagination."
You might think: "okay, so they managed 10 minutes", but:
- that was only 300 W
- the CPU die & heat spreader are quite large (18-core @ 14 nm)
- the chiller they used costs about $1k
This does not bode well for 350 W turbo. Good luck to all who try, but I expect most will hit a thermal ceiling before they get
near Tau.