tipoo

Scientific and high performance compute goes hand in hand with CUDA. They won, and AMD picks up the stragglers by offering cheaper solutions if you can run OpenCL. 

But if the Mac Pro wants to be worthy of the name and attract the right crowd, it really does need Nvidia. These interviews are disheartening to hear. I guess it'll be a Final Cut Pro X box and not very interesting for the rest of us. 

mcdave said:

I thought the T2 was meant to do some of the encoding but I’m seeing little evidence.  Surely this should have been supported by now, or was it just in the Mini?

The 2013 MP hedged it’s bets on multi-GPGPU which took far longer to materialise, but has.  Apple could at least have given it’s customers a nod by releasing Vega20(?) GPU cards.

Yeah I found it interesting that they specifically mentioned that for the Mini, while the Pro had been out for a year before. Not sure about if it's used on the Macbook Pros either. 

Also Snazzy Labs found it wasn't all that faster than Quicksync, so I wonder what the point was. 

As we all expected. Such dominance almost gets boring! 

At the same time, the Snapdragon 855 does present a decent boost if you live in the Android world, from 2500 to 3400 single core in a single generation is a good jump. 

I wonder how things would look if Apple did an even beefier single Prime core. Funny enough, any of their large cores are still larger than the Prime core in the Snapdragon though. 

I wonder if a refresh with the Radeon VII/M150 is coming soonish. The VII is an odd product for almost everyone else, but so long as Apple is sticking with AMD it's ripe for the iMac Pro. Hopefully the M150 version, while the silicon is the same, the VII caps FP64 (double precision) performance at 1/8th. 

The P/E multiple is looking awfully low especially after that 15% flash sale. Been loading up. I very much doubt this is the peak of Apples money making power, and they'll be around for decades, so it seems like an amazing buy-in opportunity.