<ydalton>
eiln: maybe a dumb question, but how different are the m1 neural engine and the m2 neural engine? you can't just slap "apple,t8112-ane" in ane_of_match?
2023-07-05
<aep>
hi. any idea if there's a difference in the neural engine for the M1 vs M2 and if yes which has the better chance to be supported some day?
2023-05-14
<enthusiast>
ah but we don't inheirt neural acceleration
2023-04-06
<marcan>
and may even talk to the neural engine because reasons.
2022-08-20
<qcyrdqcpzg[m]>
hey im new here so sorry if its already been asked but is there a team for the apple neural engine (ANE)? i used to work in CV so im really interested in it. i cant find anything much other than https://github.com/geohot/tinygrad/tree/master/accel/ane
2022-07-27
<mps>
I'm trying to select new machine to buy, what is better M1pro or M2 for linux (I don't use macos). I don't need much GPU or neural engine
2022-05-11
<Serentty[m]>
Most of those features seem pretty relevant to me at least on occasion. I could probably go without the neural engine ever getting support. Stuff like Bluetooth and DisplayPort is more relevant to your average user I think.
2022-04-28
<ylebars[m]>
Actually, the so-called neural engine in M1 architecture corresponds to TPU architecture. It is made specifically to compute some specialize linear computing.
<mps>
side note: as no one know in this world how nerves actually works calling something neural network is misleading at least (but marketing :) )
2022-04-27
<xfim>
That is precisely my impression, Cy8aer, having immersed in asahi for the last month and having read some tutorials (the gentoo one was very exciting, I have to say, https://blog.devgenius.io/installing-gentoo-linux-in-apple-macbook-pro-m1-49e163534898). But that is precisely my need: I can wait for better display, I can wait for the neural engines to work, and I am ready for a bit of experimentation and
2022-03-22
<ngnboone>
I was wondering what sort of progress/knowledge gathering has been made as far as the neural engine
2022-03-09
<PthariensFlame[m]>
So what does the M1 Ultra officially having "32 Neural Engine core" mean for the ANE@1 on each die?
2022-01-29
<mini>
as a more generic m1 pro/max question - do we know why the other half of the neural engine seems to be disabled?
2021-11-07
<jannau>
mburge: the intention is to support all hardware but I don't think the neural engine is a priority for someone
<mburge>
Is there any plan to expose the "neural engine" part of the M1 chip? Probably a complete integration with e.g. TensorFlow would be out-of-scope, but perhaps as a block device with a small library
2021-09-25
<sorear>
by the time you've done a UI interaction to wake up the phone, open the camera, taken a few dozen preview frames, autofocused, run the neural network noise reducer, etc, etc, etc, the final JPEG encode before storing the image is lost in the noise
2021-02-01
<arnd>
yes, pretty much, though ${coworker} uses it for running neural nets, as they are dealing with relatively small matrices
<segher>
jix, arnd: so AMX is for training neural nets, and ANE is for running them, it seems
<arnd>
It looks like ANE (apple neural engine) is bfloat16 and float32, while AMX is int16/float32/float64