Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
M2 Max 38-Core Benchmarks
- Thread starter Adult80HD
- Start date
- Sort by reaction score
In future, could you expense the 8TB instead, please? We want you to have the very best, ok? 😄
Go to the System Preferences -> Battery and change the power settings for both "On Battery" and "On Power Adapter" to "High Power." When I changed it my CPU went up to 3.68 GHz. Messing around with this, it's only the "On Battery" setting that seems to have an impact on what GeekBench is reporting and it does affect the results.
Does that mean I should start calling my M1 Ultra a 68 core?
Or better yet, a 90 core as that includes the neural engine?
This is fun, let’s totally normalize gross core count to describe processors.
Or better yet, a 90 core as that includes the neural engine?
This is fun, let’s totally normalize gross core count to describe processors.
So I think that I've found a bug in Ventura 13.2 with power management. I'm doing testing on some other applications, and in Adobe Lightroom, when I try to do an HDR Panorama merge, it's taking forever on both my old M1 Max and the new M2 Max, much slower than it did when I last did this testing on MacOS 12.3. It's ONLY happening on the MacBooks, none of my Mac Studios having this issue, and they are all on 13.2 as well.
If I change the power setting to "Automatic" or "Low Power" it seems to "stick" the processor in a low-power mode that's slower, if if I change it back. It's not noticeable in most application usage, but it will show up as a slower CPU frequency in GeekBench and lower scores, too. I think it's the problem with Lightroom as well, as the CPUs never spike very hard, nor does the GPU, whereas in the past when I did this they would run at full capacity, just like they do on the Mac Studio.
If I change the power setting to "Automatic" or "Low Power" it seems to "stick" the processor in a low-power mode that's slower, if if I change it back. It's not noticeable in most application usage, but it will show up as a slower CPU frequency in GeekBench and lower scores, too. I think it's the problem with Lightroom as well, as the CPUs never spike very hard, nor does the GPU, whereas in the past when I did this they would run at full capacity, just like they do on the Mac Studio.
What on earth are you going on about? You DO realize there is a version of the M2 Max with 38 GPU cores, don't you?
Can't find the setting you are referring to
Attachments
I do. There are also 48 or 64 gpu core versions of the M1 Ultra but it is still commonly referred to by the number of CPU cores it has (20) both commonly and on the Geekbench website.
Last edited:
That's the highest single core score I have seen for an M2 Max so far!
Has nobody got the 38 core GPU variant on a 14" model yet? THAT's the one I am keen to see benchmarked.
It definitely does, the 16" 30 core is still uncertain, I will be rather annoyed if that is clocked down for single core.
These are the best scores iv had so far on my 16" with 30 cores... finally over the 15000 but still running at 3.49 ghz.
The sample points are still a bit lacking but right now it does indeed seem to be looking that only the 38core max might have the full single core speed available, at least multicore scores seem to be similar enough, but there's enough single core tasks still that max possible is quite desirable, even if it is only a couple of percent extra.
I'm waiting for proper thermal comparisons between them too to know whether I'll stick with the 30 core or want to bump upto 38 in end.
My 14 inch 38 core is in Tennessee right now so I hope you get it within a few days and I will run test to see what my score is and report back here for you unless someone else beats me to it. 😉
same happened with the M1 Max 14". draw less power and its understandable...you still get more power than the M pro with the max even in the 14" but its more portable for people who needs it
What's everyones Cinebench scores? Now we know its possible apples under clocked the 30 core I'm curious to see what the 3.49 vs 3.68ghz scores.
Again here's the 30 core with a score of 14737
Again here's the 30 core with a score of 14737
It will be bizarre if the 16” 38 core is clocked the highest being the one with the highest power requirements.
Judging by the geekbench multi core scores I think both probably clock down to 3.3ish for heavily multithreaded loads.