Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

1096bimu

macrumors 6502
Original poster
I bought a refurbished M4 max studio 48GB because.... well it was available and I have a fetish for big chips and the number 48. I tried to justify it to myself with local LLMs but having tried them now, they're not useful to me at all. I also realize the M5 generation is on the horizon and it could depreciate a couple hundred dollars very soon.

However I've no intention to take advantage of the return policy because this machine rocks. I'm no stranger to Apple silicon but I've only ever had the small ones, and the crippled pro chips a few times, never the full biggest max chip. It is still unbelievable to me that this thing is just as efficient as the small devices, 0 watts at idle on the desktop!!! HOW DID THEY DO IT?

My PC which I have tried everything to make as efficient as possible, idles at 80W. Devices like the Steam Deck idles at 5W, and the Intel 258V current king of PC efficiency, still needs 3W at idle, and neither of these are anywhere near as powerful as the M4 Max. It's not just for idling, when you do something like using the browser, the power use is still extremely low, which should be obvious since the M4 Max MacBook Pro doesn't really have a battery life disadvantage, but again, seeing this happen still feels surreal to me! The power numbers are insane even when running LLMs, as low as 20w only! while delivering similar performance to 2x3090 or even 4090 which would no doubt be around 300w per card.

I also didn't expect this thing to actually feel faster than the base M4 Mac mini because theoretically same cores, and most everyday tasks should be single threaded but actually no, it boots much faster, and a lot of fairly mundane apps really do somehow run faster, indicating they've been optimized for multi-thread performance, which I totally didn't expect.

When you look at hardware benchmarks the PC guys always like to focus on constant non-stop ideal work conditions like cinebench and they'd be like oh look the M4 Max is 100w max versus 160w on the 9950x, similar efficiency. They're not wrong, the Zen5 core is decently efficient at max load especially if you back down the clocks a bit, but nobody cares about how the 9950x is 40w minimum load, where as the M4 max is hard to ever get past 40w on any real life work load.
 
bought a refurbished M4 max studio 48GB because.
Congrats, I feel the Studio is one of the best desktops, great machine very fast, extremely quiet .

but nobody cares about how the 9950x is 40w minimum load, where as the M4 max is hard to ever get past 40w on any real life work load.
Most Yters, influencers, hobbyists don't care care about low wattage, or they care about is performance and they're not wrong. Its largely only Apple who was trying to promote effeciency per watt which makes sense for laptops, but it largely falls short on desktops, where performance is more important then efficiency imo
 
  • Like
Reactions: edubfromktown
Congrats and +1 for the Studios being one of the best desktops.

A 16 GB Mini refurb was my first M1 and I still run two 2011 Mini's 24x7 hehe. The M1 Mini did fine though when the Max came out I decided to get one and sell the Mini (which netted me more than I paid for it).

Only thing that surprised me going from Mini M1 to Max M1 was power consumption. My UPS that did fine with my 4K display, the Mini and a few peripherals could not deal with even short power blips when I swapped in the Studio.

The Studio M4 Max is more power thrifty at idle than the M1 Mini which is amazing imo

ModelIdleFull Load
Mac Mini M1 (16GB)6.8W 39W
Mac Studio M1 Max (32GB)11W115W
Mac Studio M4 Max (36GB)6W145W
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.