|
|
|
|
|
|
by satellitemx
266 days ago
|
|
|
Cinebench 2024 takes considerably longer to finish and would saturate the whatever possible heat headroom the device has. It can be treated as an accurate reading. Chart below is the aggregated result from CPU-monkey, Geekerwan's chip analysis, devices of my own and various other reports. Apple M1 series 3.2GHz 5W: ~115
Apple M2 series 3.5GHz 5W: ~120
Apple M3 series 4GHz 7.5W: ~140
Apple M4 series 4.4GHz 7.5W: ~170 Snapdragon X1E 4.3GHz 10-20W: ~140
Snapdragon X2E 5GHz >20W: ~160 AMD 9950X 5.7GHz (Desktop Best): ~140
AMD AI 9 HX 375 5.1GHz (Laptop Best): ~125 Intel Ultra 9 285K 5.7GHz (Desktop Best): ~145
Intel Ultra 9 285HX 5.4GHz (Laptop Best): ~135
Intel Ultra 9 288V 5.1GHz (Low Power Laptop Best): ~125 Apple M5 may be the first ever CPU to near ~200pts on Cinebench single-core while still maintaining less than 10W of core power draw. Competitors lose on all fronts by about 2 or even 3 generations at their respective device class. |
|
|
For example, the M4 does get around 170 single, but the Snapdragon X2E gets just under 2000 for multi, over double what the M4 scores. If your application is relevant to Cinebench, the X2E is a better CPU for that. To match the X2E you need to go up to the 16 cores M4 Max.
The 16 cores variant of M4 Max is only available on a 16inch MBP that starts at 4.8K€; it's not clear how much the X2E laptops will cost but I would bet a lot of money that it's going to be much less than that...
As for the desktop's parts, the only Apple product that beats the typical X86 CPUs in multi, is actually the M3 Ultra which is pretty bad deal because it doesn't scale very well in other ways (GPU). Otherwise, Intel (i9s/Core Ultra 9) and AMD (Ryzen 9) still hold the crowns in pure multicore score.
The score of an M4 Max 16 cores, actually puts you down in Core Ultra 7 265K territory. You can put together a system based around that CPU and a 5070Ti GPU (that raw bench around the same as the M4 Max 40 cores variant but will actually perform much better for most things) for a full 1200€ less than a Mac Studio equipped like that (it even has the luxury of double the storage). If you don't need dedicated GPU power or could do with a low-end GPU the savings would be between 2000-1700€ (the 5070Ti is an 800€ part).
Let's be real, the Apple Silicon SoCs are very power efficient but they are definitely not performance maxing and they are even less money efficient. It is very suspicious arguing about top performance while ignoring multicore.
Now here is another fact: the M4 Max 16 cores can draw more power than the 140W power adapter included with the 16-inch MPB. It has a battery capacity of 100Wh. If you run the things at full tilt or near that, it will actually have a runtime of less than an hour. It's actually funny because the Apple afficionados keep singing the praise of Apple Silicon and many have been burned by that unexpected fact. It's easy to shit on the high-power gamer type laptop that can't run well on battery but that's actually true as well if you use the full power of an Apple Silicon laptop. You might get like half an hour more runtime but that's basically irrelevant.
The reality is that everyone singing that Apple Silicon efficiency praise don't have truly demanding workloads otherwise the battery life wouldn't be a meaningful difference.
High performance laptops don't make a lot of sense whether they are Apples or other brands. They are mostly convenience products (for docking/undocking) or status symbols.