If you'll base your judgement on geekbench 6 scores, you'll be disappointed in m3 ultra.
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.
M3 Max Chip Around as Fast as M2 Ultra in Early Benchmark Results
- Thread starter MacRumors
- Start date
- Sort by reaction score
GEEK benchmark score.
Current entry level Mac/PC's are more than sufficient to do 90% of what you need. The remaining 10% may take a little longer. Grab a cup of coffee, enjoy life, look beyond your screen. Spend your money on life.
Carpe diem.
Not the m3 max's score is the problem, but the m2 ultra's. Compare that to m2 max, you really believe Apple engineers did such a bad job that double core count only yields 40% performance gains? No, simply that's beyond where geekbench can scale well with cores. That also means comparison with i7 and i9 based on geekbench is also flawed. But that's probably why Apple pushes it, they'll just have to come up with an excuse when releasing m3 ultra.
I doubt the Max will increase P-core count by 50% with every release. People seem to forget this. And the m3 ultra will in all probability have 24 p-cores.
They carefully dodged comparing M3 Pro CPU to M2 Pro. They only showed comparison to M1 Pro as %20. Given that M2 Pro is around %20 faster than M1 Pro, M3 Pro should be around the same as M2 Pro CPU wise (doing so with less performance cores though).
Last edited:
Same speed with a scary price tag, should the marketing slogan have been.
To be fair, with buying a computer the various components are set up to take advantage of eachother. Buy a new CPU, your choices are limited to what the motherboard supports. Replace the motherboard, you often have to replace the RAM and maybe the graphics card. Then there’s the risk of the new components blowing the power budget. Even when I was building desktop PCs I never actually did a major upgrade on one, it was always more suitable / convenient to buy a whole new box.
With laptops and all-in-ones that’s even more the case. I think Apple has just taken what was already the case and taken the engineering a step further. But I’m pleased to see them supporting the right-to-repair bill in California, it’s just a step in the right direction.
The whole min-spec approach I think is more problematic. Without upgradeable RAM and SSDs you run the risk that in a few years there will be a glut of M1 machines with 8/256 setups on the second hand market which won’t be able to run the latest AI applications, even though the processor is fast enough.
No. Geekbench 6 is specifically designed to make multi-core results more realistic.
Geekbench 5 gives a gain of about 90% (28,000 on the Ultra vs. 14,500 on the Max), and Geekbench 6 was deliberately changed because this was deemed unrepresentative of the real world. In reality, cores give you diminishing returns — even if your code parallelizes well, it also needs a steady stream of data that partitions well.
Hmm… A Mac Studio Max available in optional iPad Air Blue, Midnight or Space Black?
I'm well aware of Amdahl's law, but testing high-end processors with mundane tasks is not real-world, very few people buy these to render html faster. The result that for basic tasks any modern cpu is enough is a too obvious conclusion to develop a benchmark for it. And it misleads people into comments like "M3 Ultra will beat 7795wx".
I have a fully specced-out M2 Ultra Studio on order. Do you think it's wise to cancel and wait? I just don't know how long it will take for an M3 Ultra Studio to arrive...
For context, it's replacing a 2019 Mac Pro and is used for casual 3D work in Blender, Motion etc.
For context, it's replacing a 2019 Mac Pro and is used for casual 3D work in Blender, Motion etc.
I would take the extra time especially when you consider your work casual.
And here I am still amazed at how fast my base model M1 MacBook Pro is!
Thinking about m3 Pro: it’s more like a scaled up m3 than it was before on the prior versions. Less Performance cores, good (calculated and scalable) results around GB 17k (11500x1,5) -> this may be the door opener for a 15inch MBAir with M3pro because of less thermal constraints.
Last edited:
Meh it's understandable M2U buyers will have some regret now, they were aiming at ultimate performance and got beaten by a laptop in less than a year. Can wave that better RAM speed flag of course, but it's unclear how it translates to actual performance.
Please show us where Apple has ever “pushed” geekbench, because I have never seen them cite it even once.
Every release we have this comment....every release "real world performance" is as stated, or better.
Last edited:
That’s kind of stupid that a laptop can match the power of a Mac Pro.
Yes, these changes do punish chip with more cores, so a 96-core one won't do so hot. But ultimately, the goal is to answer "will this CPU be faster", and even if you do buy a Threadripper, most of your tasks simply won't make good use of that many cores. They just won't.
So you'll find people who get a Threadripper and M3 Ultra and will say "well, I just ran Cinebench, and it's way faster on the Threadripper". And other people who get both and say "I did video editing / software development / some science analysis, and they were about equal, and the M3 Ultra drank less juice doing so".
And ultimately, even people who buy a Threadripper will spend a lot of their time rendering HTML faster.
Yea, it’s wise to wait. I have a full specked M2 Ultra here and will probably have to replace it again in march or June the latest now
its not that its faster
problem this is so fast it renders M2 Mac Studio/Pro useless
they gonna upgrade them to M3 Max/Ultra very soon (early spring would be my bet)