261–270 of 672 posts
If I remember correctly, the maximum configuration was something like $35k back in the day. I wonder what those people feel like now. On the other hand, if they have $35k to burn, probably they don't even think about it.
If you spend $35k and just idle the machine or just check e-mail you've burnt the money. If it's your work machine and you've got a $100/hr billable rate it's paid for in a little over a month. Three months at a $50/hr rate. If you bought the $35k Mac Pro in 2023 when it was released and have a $50/hr rate it's been paid off for about 30 months. So as of today those owners probably aren't too broken hearted. They'll…
Are you working 996 weeks or something?
At standard 40h work-week the math works out to 8.75 weeks to "pay for itself".
I bet there’s gonna be a banger of a Mac Studio announced in June. Apple really stumbled into making the perfect hardware for home inference machines. Does any hardware company come close to Apple in terms of unified memory and single machines for high throughput inference workloads? Or even any DIY build? When it comes to the previous “pro workloads,” like video rendering or software compilation, you’ve always been…
I don't think Apple just stumbled into it, and while I totally agree that Apple is killing it with their unified memory, I think we're going to see a pivot from NVidia and AMD. The biggest reason, I think, is: OpenAI has committed to enormous amount capex it simply cannot afford. It does not have the lead it once did, and most end-users simply do not care. There are no network effects. Anthropic at this point has com…
None of the things people care about really get much out of "unified memory". GPUs need a lot of memory bandwidth, but CPUs generally don't and it's rare to find something which is memory bandwidth bound on a CPU that doesn't run better on a GPU to begin with. Not having to copy data between the CPU and GPU is nice on paper but again there isn't much in the way of workloads where that was a significant bottleneck.
The "weird" thing Apple is doing is using normal DDR5 with a wider-than-normal memory bus to feed their GPUs instead of using GDDR or HBM. The disadvantage of this is that it has less memory bandwidth than GDDR for the same width of the memory bus. The advantage is that normal RAM costs less than GDDR. Combined with the discrete GPU market using "amount of VRAM" as the big feature for market segmentation, a Mac with >32GB of "VRAM" ended up being interesting even if it only had half as much memory bandwidth, because it still had more than a typical PC iGPU.
The sad part is that DDR5 is the thing that doesn't need to be soldered, unlike GDDR. But then Apple solders it anyway.
Apple need to sort out their software. Mac OS is a horrible experience.
(but yes, Apple seems happy to ship buggy software these days)
Earlier quoted context omitted.
I'll admit that while I've used the trash can but never taken it apart myself. But I can't imagine it would have been impossible to throw 2x Polaris 10 GPUs on the daughterboards in place of the FirePros.
I think on a technical level you're right, but you need to run two of them and they'd need a custom design like so: https://i.ebayimg.com/images/g/RQIAAOSwxKFoTHe3/s-l1200.jpg For what is essentially a dead-end technology, I'm somewhat doubtful people would have bought it (since the second GPU is going to be idle and add to the cost massively). the CPU being upgraded would have been much easier though I think.
Apple even in 2017 had the money and engineering resources to update or replace their flagship computer - whether with a small update to Skylake & Polaris and/or a return to a cheesegrater design as they did in 2019.
But they chose not to. They let their flagship computer rot for over 2000 days.
Earlier quoted context omitted.
I don't think Apple just stumbled into it, and while I totally agree that Apple is killing it with their unified memory, I think we're going to see a pivot from NVidia and AMD. The biggest reason, I think, is: OpenAI has committed to enormous amount capex it simply cannot afford. It does not have the lead it once did, and most end-users simply do not care. There are no network effects. Anthropic at this point has com…
No one cares about Metal in that space, plus CUDA already has unified memory for a while. https://docs.nvidia.com/cuda/cuda-programming-guide/04-speci... Can we also stop giving Apple some prize for unified memory? It was the way of doing graphics programming on home computers, consoles and arcades, before dedicated 3D cards became a thing on PC and UNIX workstations.
Apple really dropped the ball here. They had every ability to make something competitive with Nvidia for AI training as well as inference, by selling high end multi GPU Mac Pro workstations as well as servers, but for some reason chose not to. They had the infrastructure and custom SoCs and everything. What a waste. It really could have been a bigger market for them than even the iPhone.
Earlier quoted context omitted.
DGX workstations, expensive but allow PCI cards as well. https://marketplace.nvidia.com/en-us/enterprise/personal-ai-...
How much do those workstations cost? All of the different manufacturers links on that page lack pricing info and you have to contact them for pricing.
Apple need to sort out their software. Mac OS is a horrible experience.
Have you seen windows lately? (but yes, Apple seems happy to ship buggy software these days)
Apple's hardware is great, but without choice of software, they need to provide an amazing default option.
Pour one out for John Siracusa