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

chfilm

macrumors 68040
Original poster
Nov 15, 2012
3,451
2,133
Berlin
Hey guys,

so I got the binned M4max (32core gpu), 36gb on Black Friday 400€ off, which seems like a good deal!
I' ve been comparing and testing it the whole day vs my maxed out M1max and am wondering if I made the right choice or should return it and get the 700€ more expensive 40core version.

Here's what I found:
Geekbench single core, on multiple occasions never above 3600, which seems low. Any ideas why that might be? Xnip2024-11-23_22-56-20.jpg

I then ran some Lightroom Classic tests. I exported 420 edited Sony Raws to Jpeg. It took 11 mins on the M1max, and 8 on the M4max. That was the first bummer- after watching all the Artisright videos, I expected at least 6 mins or so. What was also odd, was that LR never really used all the CPU cores, in fact they were mostly idle, only the efficiency cores were really busy.

I reran the test with another project and 900 edited photos, here the difference was more striking at 11:40min vs 17 min on the M1, so that gave me some relief.

Creating 1:1 previews, which is a CPU heavy task was also a bit underwhelming:
6:30 on M4 vs
9:30 on M1 for
420 1:1 previews. Given the multicore score which is 80% higher, I expected more of an impact to be honest, and don't quite understand what the problem is.

In after effects the machine is definitely faster, given the CPU.

Davinci Resolve, 4k timeline, a short clip with noise reduction and speed warp (better) applied,
M4max: 5:45min
M1max: 6:36min

The Cpu was completely idle during that test and I didn't know how to push it more, since Resolve always relies so much on the GPU. Open for ideas.

Anyways- so this is the big picture. Overall increase seems to be around 20-30 % at best in real world, which is really below my expectations. Anybody else have similar experiences? Is this to be expected?

Do you think I would feel much better with the 40core GPU? My feeling is that I would still not feel satisfied because I'm used to an m2ultra at the office.. I kinda thought, well the Ultra has 220.000 points in Geekbench gpu, the M4max 40 has 195.000 so it should feel close, but the 165.000 of MY Ultra should've also felt like a good upgrade on paper. Especially All the reviews always blind you with those crazy CPU numbers- but it seems like in all my use cases except in AE, the CPU is mostly useless.

The 40core M4max is 700€ more.. not sure if it's worth it, it seems like It would further reduce the above numbers maybe by 1 min?
Maybe the additional Ram helps further... Or would you guys say, screw it, take the 30% gain at the great price (3300€) and call it a day?
 
Overall increase seems to be around 20-30 % at best in real world, which is really below my expectations.

What were expectations and what were they based on?

That seems like a decent increase but if your not happy with it then send it back.
 
What were expectations and what were they based on?

That seems like a decent increase but if your not happy with it then send it back.
My expectations were based on all the graphs in the artisright tests on yt..

As a matter of fact I re-ran a bunch of those Lightroom tests and the machine has somewhat gotten faster- maybe it was still occupied with background tasks in the beginning. I see now some decent numbers:

Exporting 420 edited pics:
M1 11:22
M4: 7:00

1:1 Previews of those images:
M1: 14:40
M4: 5:40

Exporting 900 edited pics:
M1: 17 min
M4: 9:30

So this looks much better now! Thank god!

I’m still wondering about the low single core geekbench score though..?
 
My expectations were based on all the graphs in the artisright tests on yt..

As a matter of fact I re-ran a bunch of those Lightroom tests and the machine has somewhat gotten faster- maybe it was still occupied with background tasks in the beginning. I see now some decent numbers:

Exporting 420 edited pics:
M1 11:22
M4: 7:00

1:1 Previews of those images:
M1: 14:40
M4: 5:40

Exporting 900 edited pics:
M1: 17 min
M4: 9:30

So this looks much better now! Thank god!

I’m still wondering about the low single core geekbench score though..?
Adding to this: I found that my OWC TB4 hub impacts my external SSDs more than I was aware of, so I attached them directly to the machine now for better accuracy. Maybe Thunderbolt 5 is a greater addition to the MacBook than I thought, if this was an actual bottleneck here!
 
Hi. I also edit my Sony Raws with Lightroom Classic and I have an 16" M3 Pro. I'm sorry to hear that you're underwhelmed. If you're running LrC v14 it's possible it's not yet fully optimized for the M4 and that performance will pick up over time. You could try uninstalling and reinstalling the program and also rebooting your Mac to run only LrC and nothing else. That should give a better picture of max performance. Keeping the Mac plugged in can boost CPU utilization slightly in my experience.

If you're underwhelmed and you're able to return then maybe consider it and you could save some money for now. You could wait until a future generation.

The more expensive M4 Max probably isn't worth it for Lightroom Classic performance. What you got is surely a great machine but Adobe software can underperform and underwhelm.

If I were you I would observe battery life to see if there's a bigger lift there.

Good luck!
 
  • Like
Reactions: chfilm
I wouldn't let benchmarks be your overall decision maker. I think you doing your real world tests is how you should decide. And yes the next CPU/GPU might give you a small bump but is the cost really worth the little extra time?

Unfortunately I'm in the same boat and don't even have a system to do any benchmarking. I'm benefiting off people like you that are providing stats and metrics. One day the math will make sense and I'll pull the trigger. ;)
 
I wouldn't let benchmarks be your overall decision maker. I think you doing your real world tests is how you should decide. And yes the next CPU/GPU might give you a small bump but is the cost really worth the little extra time?

Unfortunately I'm in the same boat and don't even have a system to do any benchmarking. I'm benefiting off people like you that are providing stats and metrics. One day the math will make sense and I'll pull the trigger. ;)
I have another benchmark here, after effects comp final rendering,

M1Max: 7:30 min
M4Max 14 core: 4 min. So that's quite significant. I think the bottleneck here is really the gpu obviously and it shows that the M2Ultra is still the king :)

Of course I would love to have the latest and greatest and get the top m4max to reach mac studio territory. But then again this is my 2nd machine and more of a fallback solution. So I think given the great offer I got it for, I'll keep it.
 
  • Like
Reactions: djinn
A lot of the benchmarks getting published are under "ideal" conditions, which means no other applications running, no login items running, and a boot from a cold start instead of a restart. So those published results can often vary from what users actually see in daily use.

This is also why the "real world" tests such as Lightroom export, FCP rendering, etc. are more realistic tools for comparison between different generations (M3 vs M4) or even variants within the same generation (M4 vs M4 Pro vs M4 Max). But even those tests should be viewed in relative terms rather than a hard target.
 
A lot of the benchmarks getting published are under "ideal" conditions, which means no other applications running, no login items running, and a boot from a cold start instead of a restart. So those published results can often vary from what users actually see in daily use.

This is also why the "real world" tests such as Lightroom export, FCP rendering, etc. are more realistic tools for comparison between different generations (M3 vs M4) or even variants within the same generation (M4 vs M4 Pro vs M4 Max). But even those tests should be viewed in relative terms rather than a hard target.
Yep…. All my slides I present to customers don’t have any benchmarks and percentage points. Feel like it’s marketing in a twisted way. Apples slides that show their performance gains with old tech to new tech are quite entertaining as well. Real world use and tests by users is the way to go, IMHO.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.