[yocto] Review request 0/13: Contribute meta-tensorflow to Yocto

[Stephen Lawrence]

> -----Original Message-----
> From: Khem Raj <raj.khem at gmail.com>
> Sent: 23 February 2019 17:05
> To: Richard Purdie <richard.purdie at linuxfoundation.org>
> Cc: Manjukumar Harthikote Matha <MANJUKUM at xilinx.com>; Stephen Lawrence
> <stephen.lawrence at renesas.com>; Hongxu Jia <hongxu.jia at windriver.com>;
> mhalstead at linuxfoundation.org; ross.burton at intel.com;
> paul.eggleton at linux.intel.com; yocto at yoctoproject.org; lpd-cdc-core-
> dev at windriver.com; Zhangle.Yang at windriver.com
> Subject: Re: [yocto] Review request 0/13: Contribute meta-tensorflow to Yocto
> 

[snip]

> > I believe that meta-oe is too large to be maintainable and that we need
> > a larger number of smaller layers.
> >
> 
> There is a fine balance to be had, that I have come to realize over years now
> but AI is large enough and segmented enough to have a layer of its own.
> 
> > Having tensorflow in its own layer which as a specific purpose and its
> > specific maintainers who understand it is in my view much more
> > desirable and sustainable.
> 
> I think its a good idea to have various AI infras in one layer
> including tensorflow
> unless we have large enough dev community to maintain each of them so I like
> meta-ai conceptually.

From a brief discussion with the team here one issue is the low backwards compatibility 
between models from different Tensorflow versions. I don't know how fast upstream is moving,
but there may be demand to support more than one version per YP version. That's not 
unsurmountable of course, but I thought I would mention it.

meta-renesas-ai has a MIT license btw if it’s a help in creating something more generic
and shared.

Regards

Steve