Mac Studio M1 Max 24gpu vs 32gpu

[MayaUser]

Except that doesn't exist in any form, not even as a rumour.

for now, and since the OP is not around here anymore, he already bought it, or he doesnt need it ASAP
So M2 Max will be the way to go if he waits couple of months

 

[seek3r]

It scales linearly. 32 GPU cores are one third faster than 24 cores.

If in doubt, always chose the cheaper option!

Vector processing, the core thing a GPU does, can easily scale exponentially depending on workload FWIW

 

[mr_roboto]

Vector processing, the core thing a GPU does, can easily scale exponentially depending on workload FWIW

That's complete nonsense. The total compute throughput of a GPU is roughly:

(number of cores) * (vector width) * (clock frequency)

If you add more cores and rerun this calculation, you get linear scaling, not exponential. The only workload-to-workload variance is that some tasks may scale sub-linearly, because perfect scaling with core count is sometimes hard.

Vector processing is not superlinear magic. If you think it is, you probably don't understand what it actually does.

 

[siddhartha]

You can ALWAYS justify waiting on the next upgrade, but you will miss out on what's available now. Of course you should research if a release is imminent, but otherwise, get the tool for the job and move on

for now, and since the OP is not around here anymore, he already bought it, or he doesnt need it ASAP
So M2 Max will be the way to go if he waits couple of months

 

[er-minio]

for now, and since the OP is not around here anymore, he already bought it, or he doesnt need it ASAP
So M2 Max will be the way to go if he waits couple of months

Already bought it. Very happy with it.

 

[pshufd]

Already bought it. Very happy with it.

There would be far less agonizing over this decision if Apple resolved the sound issue (with a statement) and also made something in-between the mini and the Studio. I was very happy with the Studio after I bought one but it took a long time to make the decision after trying to make the mini do what I wanted it to.

 

[foliovision]

Scores for base Mac Studio M1 Max (24-core GPU).
OpenCL (49401)
Metal (60096)

This is an old thread, but that info is wrong. In case anyone is also looking for an older or second machine, here's the current scores.

Base score for Metal for an M1 Max on Geekbench 6 Metal GPU test is about 104,000 (24 GPU cores). 32 GPU cores is about 123,000. Metal is the important GPU framework for macOS.

Here's a more detailed list I've put together covering the main professional machines (outside of ultra) for Geekbench 6 GPU:

M1 Max 24 GPU ≈ 104000
M1 Max 32 GPU ≈ 123000
M2 Pro 16 GPU ≈ 75000
M2 Pro 19 GPU ≈ 85000
M2 Max 30 GPU ≈ 129000
M2 Max 38 GPU ≈ 145000

 

[foliovision]

Here's a more comprehensive chart covering M1, M2, M3, M4 Max and Ultra's in high and low configuration. Took over an hour of legwork (AI gives wrong numbers at every turn, each device had to be attacked separately and then grouped in a larger chart). Hope it's useful to someone else. If any of the numbers are wrong, let me know and I'll update the post.

Chip	GPU Cores	Typical GB6 Metal Score	Score Range	Device(s)
M1 Max	24	104,000	100,000–107,000	Mac Studio, 16" MBP
M1 Max	32	120,000	118,000–122,000	Mac Studio, 16" MBP
M2 Max	30	132,000	130,000–134,000	Mac Studio, 16" MBP
M3 Max	30	130,000	128,000–133,000	16" MBP
M2 Max	38	148,000	144,000–153,000	Mac Studio, 16" MBP
M3 Max	40	161,000	153,000–166,000	Mac Studio, 16" MBP
M4 Max	32	160,000	155,000–163,000	16" MBP
M1 Ultra	48	155,000	150,000–158,000	Mac Studio
M1 Ultra	64	161,685	160,000–166,000	Mac Studio
M4 Max	40	180,243	180,000–194,000	16" MBP, Mac Studio
M2 Ultra	60	186,000	180,000–189,000	Mac Studio
M2 Ultra	76	209,240	209,000–222,000	Mac Studio
M3 Ultra	80	250,000	228,000–260,000	Mac Studio

 

[neomorpheus]

Thanks for that info @foliovision

Would it be possible to add M4 and M4 Pro (both 16 and 20 GPU cores)?

 

[foliovision]

Hi Neomorpheus. I ended up with a M2 Ultra 60GPU. The chart is focused on Max and higher.

Feel free to go and grab the numbers from Geekbench for more M4's, and publish them below this chart. It would make sense to have a separate chart for basic and Pro GPU across all the Mx variants.

 

[neomorpheus]

Hi Neomorpheus. I ended up with a M2 Ultra 60GPU. The chart is focused on Max and higher.

Feel free to go and grab the numbers from Geekbench for more M4's, and publish them below this chart. It would make sense to have a separate chart for basic and Pro GPU across all the Mx variants.

Thanks for the follow up.

I actually tried that but man, that geekbench browser is a mess.