> The M4 Pro continues to be manufactured using a 3-nm process and on the old M3 Pro (27-28 watts), we measured a lower consumption than on the M2 Pro models (~36 watts), despite its improved performance. In contrast [...] the new M4 Pro can consume up to 46 watts, settling at around 40 watts during the further course—so at its peak, it consumes 60% more.
Assuming they refer to the full chips rather than the binned ones, each generation of pro chips has the following number of p and e cores:
| Model | # p-cores | # e-cores |
|--------|-----------|-----------|
| M2 Pro | 8 | 4 |
| M3 Pro | 6 | 6 |
| M4 Pro | 10 | 4 |
Thus m3 pro has more e-cores and less p-cores than m2 pro thus the big increase in efficiency, while the m4 pro has more p-cores than m2 pro thus the increase. It is all about tradeoffs and, honestly, the result is pretty much expected when you count the cores. I assume there is some improvement per generation, but if the number of cores is not constant, the latter is gonna drive most of the variance generation to generation.
That’s comparing a M4 Pro (middle level) to a M3 Air. The Air is a lower power machine with the low spec processor.
There is no M4 Pro Air. They have to be using a MacBook Pro. That likely has a bigger display, a display capable of getting way brighter, showing more colors, better speakers, all sorts of other stuff.
That’s not a very valid comparison.
If anything, the fact that the M4 Pro gets so close to the M3 is impressive.
The M3 was on a process that was known to run hot. I strongly suspect that every M4 chip is more efficient than the equivalent M3 chip.
I am not sure that is true, does it not use more peak power but get more work done leading to less energy overall for say exporting a hundred photos because it finishes quicker?
https://www.notebookcheck.net/Apple-M4-Pro-analysis-Extremel...