43 comments

> 11.08× faster and generated tokens 16.36× faster than the same workload in the same stock VM.

So this was the comparison, for me the title was a bit confusing

frabonacci OP
yeah fair point. it's always tricky to get the whole idea across within HN's title limit. tldr: we ran the same workload in the same Lume macOS VM on the same Apple Silicon host, first with stock Metal capability reporting and then with our process-scoped dynamic library. The 11.08x figure is prompt processing, while 16.36x is token generation. the mechanism technically extends to graphics workloads too but these figures are specifically from llama.cpp
I don’t understand what Apple 1-9 are. At first I thought it was M series chips but there is no M9 (yet)
wtallis
So those generation numbers aren't really anchored to Apple's hardware designs. It's just counting from when Apple introduced the Metal API, and the first several generations were when the GPU cores Apple was using were still nominally PowerVR designs.
frabonacci OP
yeah the naming is confusing. Apple family 9 isnt M9, it's a Metal GPU feature family. Apple maps family 7 to M1, family 8 to M2, family 9 to M3/M4, and family 10 to M5
simonw
It looks to me like this won't speed up llama.cpp for everyone, just for users running it in this particular kind of Virtualization.framework VM.

The fix here works around a problem where the VM was causing llama.cpp to select the wrong kernels.

frabonacci OP
> this won't speed up llama.cpp for everyone, just for users running it in this particular kind of Virtualization.framework VM.

correct. these figures apply to llama.cpp inside the macOS guest configuration we tested. Lume is the VM frontend we used, while Apple's Virtualization.framework provides the virtual GPU. bare-metal llama.cpp is unaffected.

> The fix here works around a problem where the VM was causing llama.cpp to select the wrong kernels.

mostly, with one nuance: llama.cpp is selecting the correct kernels for the capability answers it receives. the stock guest reports an older Apple GPU family and a 32 KB threadgroup memory limit, so llama.cpp chooses slower kernels. Our process-scoped layer reports the tested Apple 9 and 64 KB values while allowing llama.cpp to select newer paths that the paravirtual GPU successfully execute

the layer itself though works at the Metal API boundary, independently of llama.cpp. other Metal compute and graphics apps now may select newer paths from the same capability answers, although this is still preliminary and each app needs separate testing. for example, MLX-LM stayed flat in our tests

historically related limitations have been coming up across Apple Silicon VM frontends for a while e.g. Tart tracked MPS/GPU support back in 2023: - https://github.com/openai/tart/issues/501 - https://github.com/openai/tart/issues/1032

UTM also has related cases where apps detect the Apple paravirtual Metal device but falls back to software rendering: https://github.com/utmapp/UTM/issues/7671

sitkack
why do use ai to write your posts ?
Aldo_MX
Why do you expect an AI engineer to manually write prose?
Why do you think an AI engineer would go through the trouble lower casing everything except for proper nouns and abbreviations?
frabonacci OP
the better question is why a throwaway account is doing capitalization forensics
b112
Don't post generated text or AI-edited text. HN is for conversation between humans.

Because it is not allowed here, that's why. See the guidelines.

octocop
a win is still a win
That makes sense. The title initially sounded like a general llama.cpp speedup on Apple Silicon, but if the improvement comes from fixing kernel selection inside Virtualization.framework VMs, that distinction is pretty important.
frabonacci OP
agreed on the title. added more context below on the exact scope and why this is really a VM capability-reporting issue: https://news.ycombinator.com/item?id=49260087
I recall there was another YC startup that was working on Mac-specific ML optimizations for local inference (and perhaps fine-tuning).

I wonder if their work is related?

frabonacci OP
RunAnywhere or Conifer?
The Claudish in the blogpost makes it really hard to ready. Also, TinyLlama 1.1B lol.
my whole setup is buy more RAM, run it on CPU, and tell myself the GPU is just a personality trait I'm working on.
I'm hoping AMD wins when the RAM bubble bursts and their integrated AMD 395+ platform can keep getting faster and higher bandwidth.
Forget the 395+, I want the MI350P (or similar) long term.

PCIe card is the way forward IMO, AI keeps changing so you don't want static hardware.

petu
How PCIe card is better?

"Static hardware" is still fully featured computer with lots of RAM, could be easily reused for other purposes.

The unified memory architecture is interesting for toying with medium size models but will never offer as much bandwidth as a dedicated GDDR memory bank. Conversely, GDDR can't be used for general CPU purposes because access latency is just too high. Unless someones also comes up with dynamically programmable memory banks, something I'm not sure would even be possible.
aeriose
What I don't get, which this article doesn't talk about, why would Apple’s Virtualization.framework expose a lesser Metal profile instead of reporting all capabilities supported by the host GPU?
bestham
Because it cannot be safely virtualised?
chorizo
All M-series chips support Metal 4. Wonder if we can fix this with a simple override somewhere.
Because nobody knows.

Apple doesn't let you "pass" the GPU through to a VM like most other ARM/x86_64 processors (forwarding interrupts and PCIe memory regions). There are symbols defined to do this within the kernel (if you dump the binary) but they aren't used in retail macos.

Instead you end up creating a paravirtual device that emulates the GPU acting like a 'normal PCI device' which you give to clients. This is usually reserved (by other hardware vendors) for when you're doing multi-tenat time sharing of higher end GPUs (like Nvidia enterprise cards can do).

These paravirtualized GPUs then just have 'less features' and Apple (being Apple) states no reason why.

frabonacci OP
my guess is apple chose a conservative profile for compatibility across different chips and guest releases
b112
QEMU/kvm does this as a default, because keeping a more generic CPU / etc makes moving VMs between machines with different hardware possible. If you try to move a VM it won't work, of course, if the new machine doesn't support what the old did.

Not sure of this is why Apple does it. With KVM, you tend to pick a baseline that all your machines support.