Anyone have benchmarks on the M1 Max 24-core?

[spiritbox]

I purchased the 14" M1 Max: 24 core GPU /32GB RAM. I am curious how this stacks up against the M1 Max 32-Core and M1 Pro 16-Core.
I have been looking for benchmarks for the M1 Max 24-core, but haven’t had much luck.

Anyone find benchmarks online or purchased this model and willing to run them?

 

[fa8362]

I've seen benchmarks numerous times, so it shouldn't be hard to find them. Almost any review has them, and there were chip benchmarks available weeks ago. Don't recall the 24 core though.

 

[fa8362]

I've seen benchmarks numerous times, so it shouldn't be hard to find them. Almost any review has them, and there were chip benchmarks available weeks ago. Don't recall the 24 core though but you can easily guess based on the 16 and 32 results.

 

[nquinn]

I don't think anyone has received the 24 core model yet since it was a built to order option. Most likely people would start receiving them in the next 1-2 weeks I think.

I'm more curious if anyone has run cpu based benchmarks on the 16core/16gpu pro vs 16core/32gpu max to see if there is much difference with the improved memory bandwidth and cache. (since the 24 would see similar cpu, not gpu gains).

 

[Jára Tyky]

The same question for me

 

[nquinn]

The same question for me

I feel like people should already be able to do this now with the 16 core gpu pro vs 32 core max cpus but no one is doing it. I might have to just dig around for some individual pro vs max videos to see if I can compare geekbench/cinebench/xcode results on them.

 

[mciarlo]

Based on benchmarks, it appears the GPU scale is fairly linear with each variation. So 32 core is 2x the 16 core. Source: AnandTech

"The M1->M1 Pro ->M1 Max performance progression is almost exactly 2x at each step"

The real question is what will the real world performance differences be for programs like Figma, Photoshop, Blender, etc. Currently waiting until December for a 24 core, with my local store having 16 cores in stock...

 

[nquinn]

Ya I expect mostly linear gpu scaling. It's more cpu focused apps that I want to see comparisons on that we should be able to get NOW since both the pro and max chips are out.

 

[7even]

Based on benchmarks, it appears the GPU scale is fairly linear with each variation. So 32 core is 2x the 24 core, and the 24 core is 2x the 16. Source: AnandTech

"The M1->M1 Pro ->M1 Max performance progression is almost exactly 2x at each step"

The real question is what will the real world performance differences be for programs like Figma, Photoshop, Blender, etc. Currently waiting until December for a 24 core, with my local store having 16 cores in stock...

FYI I think you mean 32 core the Max they tested is 2x 16 core Pro they tested which is 2x 8 core M1 vanilla they tested.

So we can expect the 24 core to be somewhere halfway between the 16 core and 32 core models. I’m waiting for mine as well, hopefully delivered once they fix the plane issues in China

 

[mciarlo]

FYI I think you mean 32 core the Max they tested is 2x 16 core Pro they tested which is 2x 8 core M1 vanilla they tested.

So we can expect the 24 core to be somewhere halfway between the 16 core and 32 core models. I’m waiting for mine as well, hopefully delivered once they fix the plane issues in China

Ah you're right! Thanks - will update my post.

 

[lankox]

Just received my 24 core in the mail today. Unfortunately I don't have time to set it up until tomorrow. Very strange that there are no 24 core scores posted yet??

 

[spiritbox]

Just received my 24 core in the mail today. Unfortunately I don't have time to set it up until tomorrow. Very strange that there are no 24 core scores posted yet??

Congrats on your new MacBook! Please post benchmarks tomorrow if you are able, would be much appreciated ??

 

[MrTan]

Same here, searching everyday for reviews. Mine is due on Monday, I'd imagine we get some user tests by then.

 

[Hankster]

There are several tests already posted.

 

[Jára Tyky]

There are several tests already posted.

if you post such comment, you should insert link directly. Otherwise your words have no meaning.

 

[spiritbox]

Anyone find benchmarks for the 24 core M1 Max this week?

 

[Jára Tyky]

I think the question is: Does somebody already has it? ?

 

[nquinn]

Ya I think the 24core models will only start getting in people's hands next week. It was a BTO order with longer shipping times (mine was originally nov 5-10th though some got a bit earlier I think)

 

[HylianKnight]

GPU performance scales pretty linearly, so the theoretical performance should be easy to determine.

 

[nquinn]

GPU performance scales pretty linearly, so the theoretical performance should be easy to determine.

For games/etc for sure. More curious how it might help in other productivity apps/etc that are more memory intensive. In general the benchmarks seem to indicate almost no difference. My only use cases that it might help would be some video export/editing (rarely), or maybe if i dabble with some machine learning models I suspect the gpu could help.

 

[lankox]

Just set my 14", 1TB, 32GB, 24 core machine up. Geekbench score = 60798 for Metal. Not bad.

 

[nquinn]

Just set my 14", 1TB, 32GB, 24 core machine up. Geekbench score = 60798 for Metal. Not bad.

Would you mind running some javascript/browser benchmarks as well?

M1 Pro / M1 Max benchmark requests: browser/javascript benchmarks

Could anyone with a 10-core M1 Pro or M1 Max (24 or 32 core) run some of the browserbench tools? I'm curious to see if the Max chip helps improve javascript/web browsing snappiness: https://browserbench.org/ Some existing results: Speedometer 2.0 Browserbench results on a new Macbook Pro...

forums.macrumors.com

 

[lankox]

Speedometer 2.0 Browserbench results = 276 on Safari

 

[lankox]

More Geekbench
Single Core = 1773
Multi Core = 12735

 

[nquinn]

Speedometer 2.0 Browserbench results = 276 on Safari

Got it so basically identical. Might be worth testing the better Jetstream benchmarks on max vs pro as well since that is a more modern benchmark if you feel like it:

JetStream 2.2