Re:Apple Silicon Exec Explains Mac Mini AI Demand and On-Dev...
by Aperocky on Friday July 10, 2026 @02:10AM
(#48856332)
You're describing the mac mini/studio with some facelift.
[ Parent
| on HN ]
Re:Apple Silicon Exec Explains Mac Mini AI Demand and On-Dev...
by huragok on Friday July 10, 2026 @02:52AM
(#48856587)
Absolutely, but not under the control of Apple.
Re:Apple Silicon Exec Explains Mac Mini AI Demand and On-Dev...
by boredatoms on Friday July 10, 2026 @02:21AM
(#48856410)
Yeah but like running linux hopefully
Re:Apple Silicon Exec Explains Mac Mini AI Demand and On-Dev...
by throw1234567891 on Friday July 10, 2026 @06:03AM
(#48857935)
so you have invent unified memory for linux first because that’s the limitation today
Re:Apple Silicon Exec Explains Mac Mini AI Demand and On-Dev...
by curt15 on Friday July 10, 2026 @07:38AM
(#48858554)
Or just buy a MI300A server like https://www.servethehome.com/gigabyte-g383-r80-aap1-amd-inst...
Re:Apple Silicon Exec Explains Mac Mini AI Demand and On-Dev...
by throw1234567891 on Friday July 10, 2026 @07:45AM
(#48858619)
"Just". And then GPUs, and RAM? And cooling? Will you really appreciate it when sitting right next to it?
Re:Apple Silicon Exec Explains Mac Mini AI Demand and On-Dev...
by vardump on Friday July 10, 2026 @07:00AM
(#48858289)
Raspberry Pi and other SBCs, Android phones and practically all of the embedded devices with a display and microprocessor.
All have unified memory. Linux runs just fine on all of those.
Re:Apple Silicon Exec Explains Mac Mini AI Demand and On-Dev...
by throw1234567891 on Friday July 10, 2026 @07:01AM
(#48858304)
Re:Apple Silicon Exec Explains Mac Mini AI Demand and On-Dev...
by vardump on Friday July 10, 2026 @07:02AM
(#48858306)
Ah ok. I replied to ~45 minutes stale page.
Re:Apple Silicon Exec Explains Mac Mini AI Demand and On-Dev...
by Alpha3031 on Friday July 10, 2026 @06:16AM
(#48857999)
Fairly sure most iGPUs these days are zero-copy and can dynamically allocate memory so what does "unified memory" mean to you exactly? A wider bus would be nice but it's not exactly a groundbreaking new invention.
Re:Apple Silicon Exec Explains Mac Mini AI Demand and On-Dev...
by throw1234567891 on Friday July 10, 2026 @06:28AM
(#48858095)
I was actually pretty far off:
> Unified memory in Linux creates a single address space accessible to both the CPU and GPU, eliminating the need to manually copy data between system RAM and video memory. It is enabled via NVIDIA's CUDA, AMD's ROCm/HIP, or generic kernel-level Heterogeneous Memory Management (HMM).
So it does exist and is available for platforms that matter.
Re:Apple Silicon Exec Explains Mac Mini AI Demand and On-Dev...
by vkazanov on Friday July 10, 2026 @06:46AM
(#48858210)
It is interesting how apple claimed that "unified memory" is something special, and ppl believed them.
Intel and AMD had been doing this for years already, and had linux support for it from day 1.
Re:Apple Silicon Exec Explains Mac Mini AI Demand and On-Dev...
by throw1234567891 on Friday July 10, 2026 @07:00AM
(#48858291)
Cool. Apple was the only one who managed to ship a consumer device with UMA and RDMA support. 2TB VRAM max over RDMA.
Re:Apple Silicon Exec Explains Mac Mini AI Demand and On-Dev...
by vkazanov on Friday July 10, 2026 @08:17AM
(#48858866)
I think the REALLY cool thing about apple's shared memory implementation is the ultra-wide memory bus.
Otherwise, AMD is quite close to what Apple has, and Strix Halo is honestly incredible.
Not sure what RDMA brings to the table.
Re:Apple Silicon Exec Explains Mac Mini AI Demand and On-Dev...
by throw1234567891 on Friday July 10, 2026 @09:22AM
(#48859578)
RDMA increases the inference performance by a significant percentage across devices connected via Thunderbolt 5.4x512 is like a 2TB machine.
Re:Apple Silicon Exec Explains Mac Mini AI Demand and On-Dev...
by boredatoms on Sunday July 12, 2026 @02:24AM
(#48878829)
Thunderbolt RDMA is slower and higher latency than if apple just gave us PCIe, where we could put a (old) connectX card in for infiniband