Waiting for the M1 16" MacBook Pro Thread

[Krevnik]

I am really curious how Apple will solve this technical design problem. Maybe I'm overthinking it, and Apple can just double the CPU and GPU cores and surpass current MBP16 CPU/dGPU performance?

The die size increase is probably not as big as you think. At least not at this scale based on the die shot from AnandTech: https://www.anandtech.com/show/16226/apple-silicon-m1-a14-deep-dive

Let’s say Apple goes 8P/4E/16GPU. That’s roughly 50% larger die by area. They could go up to 8P/4E/24GPU while keeping things less than double the area. Such a chip would still give reasonable yields.

From my perspective, it’s when trying to replace the Mac Pro that things get more difficult. Trying to beat both a 28-core Xeon, and a Vega II on a single die is going to be super demanding in terms of manufacturing yields compared to the M1. If demand for such parts is low enough, and people are willing to pay a lot for it, then it’s still possible to do it on a single die. But this is where I think you’ll start seeing more creative solutions from Apple as a way to manage die size (and the costs associated with them). But I’d be surprised if Apple breaks up the die on a hypothetical M1X.

 

[jdb8167]

The die size increase is probably not as big as you think. At least not at this scale based on the die shot from AnandTech: https://www.anandtech.com/show/16226/apple-silicon-m1-a14-deep-dive

Let’s say Apple goes 8P/4E/16GPU. That’s roughly 50% larger die by area. They could go up to 8P/4E/24GPU while keeping things less than double the area. Such a chip would still give reasonable yields.

From my perspective, it’s when trying to replace the Mac Pro that things get more difficult. Trying to beat both a 28-core Xeon, and a Vega II on a single die is going to be super demanding in terms of manufacturing yields compared to the M1. If demand for such parts is low enough, and people are willing to pay a lot for it, then it’s still possible to do it on a single die. But this is where I think you’ll start seeing more creative solutions from Apple as a way to manage die size (and the costs associated with them). But I’d be surprised if Apple breaks up the die on a hypothetical M1X.

I'm not convinced that Apple will abandon discrete GPUs for the Mac Pro replacement. I just don't think they can match the performance you get with multiple GPUs in the MPX modules. I think they will have to offer MPX GPU modules to even come close to the performance of the current Mac Pro. I can't believe the they created the MPX module for a single generation of Mac Pro. It isn't like they didn't know that the Apple Silicon transition was upcoming when they designed the 2019 Mac Pro.

 

[Krevnik]

I'm not convinced that Apple will abandon discrete GPUs for the Mac Pro replacement. I just don't think they can match the performance you get with multiple GPUs in the MPX modules. I think they will have to offer MPX GPU modules to even come close to the performance of the current Mac Pro. I can't believe the they created the MPX module for a single generation of Mac Pro. It isn't like they didn't know that the Apple Silicon transition was upcoming when they designed the 2019 Mac Pro.

I agree on principle, but Apple did think that the 2013 Mac Pro was gonna be a great long-term move. So who knows what Apple is thinking here. They may actually try to build enormous dies, and give it a good go. But yeah, I doubt that it makes sense to try to merge enough GPU compute to compete with dual Vega II Duos onto the SoC.

My thinking is more that if they can embed in enough GPU power to make an MPX module optional for folks who need more CPU/RAM/Storage than GPU compute, than rumors like the half-sized Mac Pro enclosure start making a lot of sense.

 

[Argon_]

The die size increase is probably not as big as you think. At least not at this scale based on the die shot from AnandTech: https://www.anandtech.com/show/16226/apple-silicon-m1-a14-deep-dive

Let’s say Apple goes 8P/4E/16GPU. That’s roughly 50% larger die by area. They could go up to 8P/4E/24GPU while keeping things less than double the area. Such a chip would still give reasonable yields.

From my perspective, it’s when trying to replace the Mac Pro that things get more difficult. Trying to beat both a 28-core Xeon, and a Vega II on a single die is going to be super demanding in terms of manufacturing yields compared to the M1. If demand for such parts is low enough, and people are willing to pay a lot for it, then it’s still possible to do it on a single die. But this is where I think you’ll start seeing more creative solutions from Apple as a way to manage die size (and the costs associated with them). But I’d be surprised if Apple breaks up the die on a hypothetical M1X.

Maybe there'll be two sizes of Mac Pro. The full tower and mini.

As for the Mac Pro SoC, I think they'll try and make it powerful enough that most pros won't need the dGPU.

As for the massive die, we'll probably see aggressive binning. 32P 4E 64GPU, with variants all the way down to 24P 4E 50GPU, or something like that. The top bin will be expensive, but a whole lot less than a 28 core Xeon.