Apple today announced the new Mac Studio, the most powerful Mac ever made, featuring M4 Max and the new M3 Ultra chip.
It would be bad if it was the least powerful Mac ever, It makes mW laugh they need to tell people it is the most powerful everytime
Did they remove the stupid limit of 2 VMs per machine or is it still an expensive toy?
Is the limit 2 VMs or two macOS VMs? I thought it was technically a “licensing” restriction.
Two Mac OS VMs enforced at Kernel level, which wasn’t a thing on Intel Macs.
If I understand correctly, it is two MacOS VMs.
EDIT:
Holy shit.
M3 Ultra chip with 32-core CPU, 80‑core GPU, 32-core Neural Engine, 512 RAM
Still a ton of money, but I’m salivating. M4 is only Max at this point, but now I’m dreaming about what that might become.
I wonder if this will be the game plan going forward, with the Ultra chip lagging by a year. Seems a likely cadence.
M3 Ultra chip with 32-core CPU, 80‑core GPU, 32-core Neural Engine, 512 RAM
That RAM is nice, but core count doesn’t say much at this point: there are different cores with different architectures, multithreading, pipelining, caches, speeds, etc.
I’d rather see a TOPS comparison:
- M3: claims 18 TOPS
- M4: up to 38 TOPS
- nVidia H100: up to 3900 TOPS/TFLOPS (INT8/FP8)
Meta is claiming to have 350,000 H100s, to put things into perspective.
I mean, sure, but largely GPU-based TOPS isn’t that good a comparison with a CPU+GPU mixture. Most tasks can’t be parallelized that well, so comparing TOPS between an APU and a TPU/GPU is not apples to apples (heh).
Agreed, but my point is that stating “x-core CPU, y-core GPU, z-core NPU”, is basically non-information.
- CPUs run general logical processing
- GPUs run integer/float matrices
- NPUs run minimal effort matrices for inference
I’d like to see the TOPS for each of those, instead of a “core count” that tells me nothing about actual performance. Even the TOPS are orientative… but would be a good start.
Agreed! I’m just not sure TOPS is the right metric for a CPU, due to how different the CPU data pipeline is than a GPU. Bubbly/clear instruction streams are one thing, but the majority type of instruction in a calculation also effects how many instructions can be run on each clock cycle pretty significantly, whereas in matrix-optimized silicon its a lot more fair to generalize over a bulk workload.
Generally, I think its fundamentally challenging to generate a generally applicable single number to represent CPU performance across different workloads.
I think the next ultras won’t use the fusion thing and will be just a larger die. (the m4 max does not seem to have the connector) so it might take them a bit to sort it out.
Here’s the link to the actual press release: https://www.apple.com/newsroom/2025/03/apple-unveils-new-mac-studio-the-most-powerful-mac-ever/
how much it costs with 512Gb of video ram?
A single 128GB mainboard is $2000, so four of them is $8000. Not the same as a machine with 512GB of unified memory.
For the M4 Mac Studio, it costs $4800 to upgrade from 32GB to 128GB. I believe everything else “base model” is $2000, for a total of $6800, but this is also a full working computer and not just a mainboard.
“Most powerful mac ever” is a weird headline. Although at least it is reusable every year or so.
