My guess is the pros are going to be the tock to the consumer tick in the Apple silicon upgrade cycle. It takes months before pros are comfortable upgrading MacOS to begin with, and it will probably be a year or two before they're comfortable that pro software vendors have flushed out their bugs on the new architecture.
Basically I'm guessing that no one who wants more than 16GB on their professional machine was going to upgrade on this cycle anyway. We'll see
Anyone remember calling a taxi dispatch over and over? I used to have to call both taxi dispatches in my area (there were only two!) just to guarantee I'd have a cab get me to the airport in time. And you had to pay cash.
I think we can all agree that taxis sucked before Uber/Lyft forced them to update. But that doesn't mean we can't have a better taxi experience using the tech stack Uber/Lyft created.
I don't like arguing semantics but I don't really think any of what Geekbench does is a "micro" benchmark. At least for me that typically refers to running a small snippet of code, like calculating a dot product or something. Geekbench tests whole program performance.
It's not aida64 but it is a pretty decent metric, and consistent.
There are many reasons. The latency of getting data back and forth to the GPU is a pretty high threshold to cross before you even see benefits, and many tasks are still CPU bound because they have data dependencies and logic that benefit from good branch prediction and deep pipelines.
Many high compute tasks are CPU bound. GPUs are only good for lots of dumb math that doesn't change a lot. Turns out that only applies to a small set of problems, so you need to put in lots of effort to turn your problem into lots of dumb math instead of a little bit of smart math and justify the penalty for leaving L1.
Geekbench isn't a micro benchmark, it's a comprehensive test of the system using a variety of programs and workloads and aggregates the results. It's not a single program that one can play games with compiler optimizations.
To use startup lingo, they were a unicorn turning into a zombie because they never hired people who knew how to run their organizations as a cockroach. A lot of the people that I know who went there were basically fresh out of college or academia, because there was no one else in Florida and ML operated in stealth for way too long. That's not a bad thing (who among us hasn't been a fresh, doe eyed engineer at a startup?), but when you throw a couple billion dollars at them before they've learned how to build anything... might not work out.
I don't want to doxx myself so I'm intentionally leaving out details, and you can take this comment as rumor and baseless.
But based on the people I know (personally) who work there - there wasn't a lot of talent being thrown at the product. I think their organization is incapable of bringing a product to market, even if that market existed.
they allegedly had an insane demo for investors only, before around 2016 (at least that's when I heard of it) you needed to sign an NDA to even get in the room to see it.
The consumer product is quite lacking compared to what people said was a demonstration of game changing technology.