[hylaride]

My understanding is that Apple has been seeing market share issues at the low end, especially in education. Since everybody has a phone, the "casual" computer market is full of Chromebooks at cheap laptops. Laptops are a tool (again?) instead of a necessity.

[vovavili]

Would be foolish of them not to take advantage of Microsoft having self-sabotage as its favorite pastime.

[bigyabai]

Apple has taken advantage of that for the past 2 decades, to the tune of a minority share in the PC market.

Meanwhile Nvidia is happily cresting what, 5 trillion in valuation? It's a weird time to be an Apple shareholder.

[seanmcdirmid]

Apple M series are competitive in inference at least. I wish Apple would just aim their chip people at NVIDIA in everything else. They are probably the only ones that have the talent, resources, and capital to do that.

[robertjpayne]

I'm quite happy Apple stays focused on their products. They enter a market when they can own it end-to-end -- it makes no sense for them to all of a sudden become an AI chip house or AI server house.

[seanmcdirmid]

There is a lot of money in AI chips, and Apple could definitely get a fairly large slice of that business if they put the work in (well, if they are putting the work in now, depending on what Baltra is really about).

[bigyabai]

They're honestly not competitive for inference, it's why datacenters largely ignore Apple Silicon. Even the M5 Max is still bottlenecked for dense models due to the relatively weak GPU and paltry ~500-600gb/s of GPU memory bandwidth. For reference, the RTX 5080 (a consumer GPU) has 1tb of VRAM bandwidth and runs circles around the M5 Max in GPU compute benchmarks: https://browser.geekbench.com/opencl-benchmarks

Even for home inference, it's hard to recommend a dedicated Mac over a cheap Nvidia server box.

> They are probably the only ones that have the talent, resources, and capital to do that.

Apple invented OpenCL. The problem was their reluctance to work with the rest of the industry, and once CUDA took over it was too late for them to even try.

[seanmcdirmid]

> For reference, the RTX 5080 (a consumer GPU) has 1tb of VRAM bandwidth and runs circles around the M5 Max in GPU compute benchmarks: https://browser.geekbench.com/opencl-benchmarks

NVIDIA hampers their GPUs with un-unified graphics memory, while the M series can use everything the computer has (well, you need to save 4GB or so). It also works on airplanes and in hotel rooms, a cheap NVIDIA server box with 64GB of RAM (what my M3 Max laptop has)....how cheap is that?

[andriy_koval]

I think un-unified memory issue is solved by software layer in datacenter setting: model is distributed across multiple GPUs in the same server, or across multiple servers if model is extra large.

[PaulHoule]

... and a huge amount of help from Dell, Lenovo and the like.

[forsalebypwner]

yeah, that's a good point...

man, my opinion of Lenovo has tanked in the last decade and I'm only just realizing it now. I always thought of Dell as kinda shitty, but Lenovo had a great thing going that just kinda atrophied in some cases, and actively got worse in others

[varispeed]

Chromebook should be classified as torture device.

[actionfromafar]

Enhanced interrogation computing device. We just need your personal data!