I have a PC wiht a RTX 3080 GPU. currently use the 16 inch MBP but know will switch to M1. Does it make sense to get an eGPU enclosure to render and FX in FCP with the 3080? Am I going to get way better performance?
You're saying M1 Macs don't support eGPUs because of patents? Where did you get this info?Due to patents owned by NVIDIA, which Apple got around thanks to help from Intel. New Apple made M1 chips dont enjoy those same patent work arounds. Hence M1 Chips made by Apple cant get around eGPU patents NVIDIA holds.
Works arounds are in the works but not yet ready so don’t hold your breath. Or that they will be as good when released.
cool. looking forward to the new modelsYou're saying M1 Macs don't support eGPUs because of patents? Where did you get this info?
What is an eGPU patent? A GPU is just a PCIe card. You connect it, add some drivers, and you're done. M1 Macs have the drivers but they weren't recompiled for ARM.
I don't know if Apple will ever enable eGPU for Apple Silicon. Third parties can't make graphics drivers for Macs.cool. looking forward to the new models
That’s why I’m waiting for M2X with eGPU native support.M1 does not support eGPUs and my bet is that it never will. It does not make any once for Apple to support eGPUs on Apple Silicon as they are not compatible with their programming model.
That’s why I’m waiting for M2X with eGPU native support.
I think if the next higher end chips have a GPU that is the same or a bit better than the 5600m in my 16" then it should be great.I've used an eGPU in my daily setup for 4+ years now (a Radeon RX580 8GB with my 2016 nTB MBP). It's worked very well for my needs and has been really good especially this past year with the whole work from home thing. While I'm a little disappointed that eGPU support will probably not support Apple Silicon, I'm completely fine ditching my eGPU in the future if the graphics performance in Apple Silicon is close to being as good as an internal dGPU.
You wanna provide some examples here? Because it frankly sounds like you're blowing smoke.As I said before, eGPUs are fundamentally incompatible with Apple Silicon programming model as they break important assumptions guaranteed by Apple architecture. It won’t be any different for M2 or whatever revision.
You wanna provide some examples here? Because it frankly sounds like you're blowing smoke.
There's nothing particularly special about an eGPU, it's a discrete GPU that happens to be connected on PCIe lanes that are externalized via Thunderbolt. Apple Silicon still uses PCIe. The only REAL issue here is drivers. If Apple is supporting discrete GPUs in later Apple Silicon Macs (and this is a pretty good bet), the only thing preventing eGPU support at that point is laziness or attempts at control.
Does this unified memory mean Apple GPUs will likely always be part of the SoC? Even the eventual AS Mac Pro?Examples:
Bottomline: if Apple allows third-party GPUs on their platform, they are fragmenting the API and diluting the strong guarantees offered by their own GPUs. Developer targeting Apple Silicon would need to implement multiple GPU algorithms for different GPU families, which will discourage them from optimizing for Apple GPUs specifically.
- Apple GPUs have unified memory, eGPUs by definition do not
- Apple GPUs are TBDR devices, third party GPUs are not
- Apple GPUs offer certain performance guarantees, third party GPUs are not
- Metal is designed for Apple GPUs and full range of Metal features is only available on Apple GPUs, third party GPUs only support a subset of those features
Right now, Apple Silicon offers a unique benefit for GPU developers: a streamlined development environment with a powerful API and unified hardware capabilities. Third party GPUs and eGPUs in particular are not compatible with this environment, ergo, I don't see Apple sabotaging their platform by offering that feature.
Does this unified memory mean Apple GPUs will likely always be part of the SoC? Even the eventual AS Mac Pro?
I have very shallow depth of technical knowledge, but what I do understand is fascinating. So curious to see how Apple will handle their high end machines and their pro customer base.Seeing how much emphasis Apple put on advertising UMA on Apple Silicon I just don’t see them not having it on the Mac Pro as well. Whether it will be one large SoC or multiple chips interconnected by a shared cache/memory die (like AMD does) is a technical question.
P.S. to be 100% clear, this is just my opinion. I might very well be wrong.
My totally-undeducated opinion is that Apple is moving ever-closer to "completely closed systems" on the Mac.
As such, the m-series Macs will become ever-more "limited" in the range of "upgrades" they can support.
As an example of current M1 benchmarks:So I don't see any need for an eGPU. And it's only going to get better with future iterations. So if you have an Intel mini, you definitely need an eGPU. But it's not needed with the M1.
For 3D modelling mayber. What about GPU Rendering?I had a high-end eGPU setup with my previous i7 mini. It worked well, but definitely revved up pretty loudly when I worked on 3D modeling or Final Cut projects. My M1 mini is much faster than my previous setup and has been more than capable with whatever I've thrown at it (and has been completely silent). So I don't see any need for an eGPU. And it's only going to get better with future iterations. So if you have an Intel mini, you definitely need an eGPU. But it's not needed with the M1.
My totally-undeducated opinion is that Apple is moving ever-closer to "completely closed systems" on the Mac.
As such, the m-series Macs will become ever-more "limited" in the range of "upgrades" they can support.