tht

blastdoor said:

When the studio was first announced they strongly implied a new Mac Pro would come later. That never happened. If the studio laps the pro with no mention of a future pro, I’ll conclude the pro’s future is in serious doubt.

Gruber should ask about it in his WWDC edition of his podcast.

I'm sad that Apple has shown their hand with the Mac Pro yet, and that it is taking so long. It basically implies the solution they were planning on for the Mac Pro didn't not provide the performance expected for its price target.

The clue to me is that their GPU performance scales very poorly with GPU cores from 16, 32, to 64 cores, and they weren't expecting it. For many of the GB5 GPU compute loads, it looked it was reaching diminishing returns from 32 to 48 cores. Once they saw it not panning out, that meant at least a 1 cycle delay, 12 to 18 mo, and the true solution is 2 cycles away, 24 to 36 mo. So if there is an M2 Ultra, they at best could only do late changes, like cache size changes, to address it. Then, the M3 cycle could institute architectural changes to improve scaling.

The rumor was Apple was going to have 4 M1 Max chips bridged together in someway so that it would appear as one CPU and one GPU. How they did the silicon interposers for 4 M1 Max chips would have been interesting to see.

I'm more curious about the capabilities of the headset operating system rather than its brand name.

There's got to be some kind of guarantee from the OS that it will generate the next frame at some minimum frame rate. If the frame rate slows down, it's going to be puke city. In a way, the requirements to deploy an app on Apple's headset may be more stringent than the Watch. Like the Watch and energy usage, the frame rate on the headset has to be absolutely protected. Wondering how they are going to achieve this.

waveparticle said:

Will Apple have an answer to the nVidia juggernaut?

Obviously no. They are not in the business of providing dGPUs for gaming, workstation, or server markets. They aren't in the business of maximizing GPU performance at all costs, only constrained by residential and office electric circuits. The server GPU requirements? Sheesh, those server boxes with 4 GPUs must need 220V 30A circuits going into them. That's what I use to run my dryer, oven or my car.

The Mac Studio, and Mac Pro if it ever is released, are mid-end media production machines, for Youtubers, or smaller media production uses. They are going to have a minimalistic power consumption for a GPU, like 120 W in the Studio and maybe 250 W in a Mac Pro. If they match the performance of 250 to 350 W dGPUs, it's a big win for what they are trying to do. Sell a small or quiet and cool machine to prosumer content creation users. I think that is their goal, and it's not to enter new markets with these boxes, unless creation tools for the XR goggles gives them a new Mac market.

Mid-range dGPUs from AMD and Nvidia basically start at 250 W, and the high end ones are at 350 to 450 W. My 15yo wants a new dGPU. I am thinking I will turn it into a ducting project for him too.

What’s the distinction between adults and children here, other than convention?

If social media is harmful to children, it will be harmful to adults, no? It’s only by convention that people over 18 are fully responsible for themselves and children are not, and therefore get a special carve out. The harm still applies to people of all ages. 

2morrow said:

I have a question. 

So when Apple or it’s suppliers purchase x gigawatts of renewable energy in China or some other country how do they know that the power is for them and not another big company that is asking for renewable energy. Can they not just apply the same power from renewable to as many companies as they want and not really have all of that renewable power allocated?

Grid power and energy is fed by multiple sources of electricity. Once those individual power sources, like individual potatoes, energize the grid, it is basically mash potatoes to end customers. At the point of usage, this energy cannot be isolated to a specific type of source.

What Apple does, and basically all renewable energy contracts from power providers, is ensure that for whatever energy they or their supplier uses, they put in an equivalent amount into the grid from a renewable power source. The money that pays for renewable energy from Apple and its suppliers goes to renewable power sources. The grid doesn't have infinite capacity, and the grid operator has to maintain a balance plus reserve. If a power generator isn't paid for or is too expensive, the grid will ask them to decrease their output or shutdown. They will shut down themselves if they aren't been paid.

Since Apple is guaranteeing a renewable energy source, it typically means the most expensive power generator will be reduced or shutdown. That's coal. Nuclear is actually the most expensive, but the fleet is typically kept alive for strategic reasons. Once grid batteries, fed by renewables, get to significant capacities, natural gas peaker plants will be next. And in the not so distant future, gas plants will be shutdown too.