I'd say it's not just about getting more out of existing processors, but being able to put in more powerful ones, in particular more powerful mobile GPU's.
Yeah, that's true. There's not that much headroom on the CPU front, but there is for GPUs — if they make the device thicker and give it more battery (or are willing to give up a significant chunk of battery life).
Having said that, do we have any data on what the actual percent increase in CPU and GPU performance would be if we took the top CPU and GPU in the current MBP and moved them to a much more thermally dissipative chassis? Are they merely "a few more percent"? E.g., have you found any benchmarks comparing the top-of-the-line MBP with that of the same GPU &CPU in, say, a generously cooled gaming laptop or mobile workstation? [GPU might be tough -- most higher-end laptops seem to use NVIDIA.] I'm genuinely curious what those numbers would be.
TL;DR:
it depends on your workload.
The closest data I could find, headline-wise, would be Bare Feats's
"Does the 2019 MacBook Pro 15-inch 2.3GHz 8-core exhibit thermal throttling?".
It's not that great an article, though. Does the DaVinci graph show all four runs? Where did each of them start? Why does the text say "the core frequency dropped as low as 2.39GHz, a tad above the base frequency of 2.3GHz" when it appeared to actually drop to roughly 2.0 GHz,
below base? Why doesn't the LuxMark graph show temperature?
I also find their conclusion a little lacking:
It could be argued that as long as you remain above the base frequency (or 2.3GHz in this case), your 2019 MacBook Pro is not experiencing 'detrimental' thermal down throttling.
I'm actually a bit confused because the DaVinci graphs don't seem to support this — clearly, the CPU briefly drops to about 2.0 GHz before it recovers all the way to about 3.9 GHz? And this does seem to correlate with reaching almost 100 degrees C. There's a fair amount of difference between ~3.5 GHz and ~2.1 GHz.
OTOH, that's clearly Turbo Boost. This depends on a lot of variables, really. The 9880H used here actually goes up to 4.8 GHz, but we don't really know if it isn't doing that here because of thermal throttling, or because DaVinci uses too many cores for that to possible (perhaps ~3.6 GHz is the max for whichever many cores DaVinci uses).
The same page shows results in Blender and LuxMark, and here we see a different picture: for Blender, the frequency never reaches 3 GHz, but remains stable at a slight turbo of 2.8 GHz, and an also fairly stable temperate of 90 degrees C. With the LuxMark benchmark, we don't get temperate results (why not?), and it's unclear if temperate is the reason it eventually goes down to ~2.3 GHz.
So the short answer is: this model does appear to sustain its 2.3 GHz base clock across all eight cores… mostly (the exception being whatever the hell is going on in the middle of that DaVinci graph). That Blender graph looks pretty good.
Where the answer gets trickier is if fewer cores than eight are used and/or turbo gets enabled, both of which you'll frequently see in everyday use (very, very little code out there can make good use of four cores, let alone eight).
Of course, if they *really* went with a completely different chassis, they could get a substantial performance increase by going with desktop processors (ask Dell does with its Alienware Area 51 laptops). Granted, that seems highly unlikely, but it is fun to mention.
Yeah, no way.
(He says, wondering what impact Ive leaving will have.)
