4 ms·

The other way to look at this is that they failed to make the Mac Pro enough better than the Mac Studio or laptops to make it worthwhile. Graphics cards are one

The other way to look at this is that they failed to make the Mac Pro enough better than the Mac Studio or laptops to make it worthwhile. Graphics cards are one of the biggest use cases for PCIe slots and is something that Apple doesn't allow. If you are looking for the highest horsepower options, they are all Windows based. And I say this as someone who prefers macOS and definitely doesn't need this level of hardware.

erichocean 3y ago
I have two Mac Studios. When I need "more GPU horsepower", I spin up a GPU VM for $0.40-$0.80/hour (depending on my needs) and work away. Then I shut it off. I can't be the only one doing ML development this way.
- ClassyJacket 3y ago
  Sure. But there are alot more use cases for GPUs than machine learning, many of which are latency sensitive and must be done locally.
- sarathyweb 3y ago
  Would you mind sharing what cloud provider you use?
  - erichocean 3y ago
    https://www.runpod.io/ https://www.runpod.io/ I do have a Windows machine with an nVidia GPU (RTX 2070 Super), but I don't use it anymore (bought it years ago). ML on the cloud is way more convenient because you can trivially adjust your cost based on what you're doing: Training? spin up something big/expensive. Inference? cheaper (less VRAM) is usually fine. I also like that I can run multiple instances simultaneously, something that would be prohibitively expensive if I had to have multiple machines sitting around waiting for me to use them.
Spooky23 3y ago
That’s the symptom. The problem is just a trade off of the Apple Silicon architecture. The M1/M2 is brilliant for the form factors that consist of about 90% of PC demand. Macs are better as a device than almost any PC on a few axes. But… an architecture using an SoC to scale from an iPhone to a large personal server, which is what a modern era Workstation really is will hit limits. IMO, the failure was they were boxed into dropping MacOS Intel support in the near future, and for whatever reason were unable to ship the Apple Silicon they intended to. So they shipped a compromised Mac Pro that meets the needs of a few key use cases that can sorta work with the limited top line memory.
- fragmede 3y ago
  That implies that the yields were off for the M2 SuperMax or M3 or whatever, so they couldn't make enough of them to sell a product.
  - PedroBatista 3y ago
    Not necessarily, the yields could be OK but the performance bump of the real silicon didn’t materialize.
  - Spooky23 3y ago
    Maybe - they may not have been able to package it. Workloads are usually CPU bound or memory bound. If I need 200 cores to get a TB of memory, that’s a waste of money. Intel segmented their products because they had the similar constraints. There’s a mondo-expensive 8 core 4 GHz+ part, and a part with lots of cores at half the speed. In Intels case, the memory is of course not part of the SoC.
jmkni 3y ago
Really dumb question but has anybody actually tried putting a graphics card into one of these machines? What happens? Nothing?

erichocean 3y ago

I have two Mac Studios. When I need "more GPU horsepower", I spin up a GPU VM for $0.40-$0.80/hour (depending on my needs) and work away. Then I shut it off. I can't be the only one doing ML development this way.

ClassyJacket 3y ago
Sure. But there are alot more use cases for GPUs than machine learning, many of which are latency sensitive and must be done locally.
sarathyweb 3y ago
Would you mind sharing what cloud provider you use?
- erichocean 3y ago
  https://www.runpod.io/ https://www.runpod.io/ I do have a Windows machine with an nVidia GPU (RTX 2070 Super), but I don't use it anymore (bought it years ago). ML on the cloud is way more convenient because you can trivially adjust your cost based on what you're doing: Training? spin up something big/expensive. Inference? cheaper (less VRAM) is usually fine. I also like that I can run multiple instances simultaneously, something that would be prohibitively expensive if I had to have multiple machines sitting around waiting for me to use them.

Spooky23 3y ago

That’s the symptom. The problem is just a trade off of the Apple Silicon architecture. The M1/M2 is brilliant for the form factors that consist of about 90% of PC demand. Macs are better as a device than almost any PC on a few axes. But… an architecture using an SoC to scale from an iPhone to a large personal server, which is what a modern era Workstation really is will hit limits. IMO, the failure was they were boxed into dropping MacOS Intel support in the near future, and for whatever reason were unable to ship the Apple Silicon they intended to. So they shipped a compromised Mac Pro that meets the needs of a few key use cases that can sorta work with the limited top line memory.

fragmede 3y ago
That implies that the yields were off for the M2 SuperMax or M3 or whatever, so they couldn't make enough of them to sell a product.
- PedroBatista 3y ago
  Not necessarily, the yields could be OK but the performance bump of the real silicon didn’t materialize.
- Spooky23 3y ago
  Maybe - they may not have been able to package it. Workloads are usually CPU bound or memory bound. If I need 200 cores to get a TB of memory, that’s a waste of money. Intel segmented their products because they had the similar constraints. There’s a mondo-expensive 8 core 4 GHz+ part, and a part with lots of cores at half the speed. In Intels case, the memory is of course not part of the SoC.

jmkni 3y ago

Really dumb question but has anybody actually tried putting a graphics card into one of these machines? What happens? Nothing?