[Libre-soc-dev] video assembler

[Lauri Kasanen]

On Tue, 11 May 2021 13:23:51 +0100
Luke Kenneth Casson Leighton <lkcl at lkcl.net> wrote:

> On Tuesday, May 11, 2021, Lauri Kasanen <cand at gmx.com> wrote:
> > Does it at least count the sub-steps separately? So they can be divided
> > by say 8, for an 8-parallel cpu. After all that's the point of simple-v,
> > N of these happen at the same time in parallel. Without that there'd be
> > no point at all to write anything at this point.
>
> this i would consider to be a false objection: in english, "cart before the
> horse".  with no unit tests we cannot even begin to know where the areas
> are to make improvements.

If it can't measure even roughly what is better, that is no false
objection. It's essential to writing optimized codecs. Even at that
level we could point to the measurements and say "in theory this is x%
faster for this use, everything else being the same". It's far from
what could be done with a higher-level simulator, not upstreamable in
any way, but not useless.

Ie, without it, there is nothing for me to write. The grant was about
writing optimized codecs, incl new instructions.

> 2) they help show where things should go for the next phase (full
> algorithms)

I disagree. A blindly written inner loop is of zero use for the actual
code. It most likely needs a full rewrite once proper measurements can
be done.

> my biggest concern is that by not starting we have nothing at all by the
> Grant deadline which is only around six months away.
...
> we have to cut through that and begin somewhere, before it is too late and
> we lose the funding.
...
> 1) they act as more complex unit tests, that help ensure the HDL and
> simulator is functional where needed
>
> to give you some context, Lauri: the largest unit test currently ever run
> is only 7 instructions.  "only" 20 opcodes would be a *200*% increase on
> that.
>
> so the question is: would you be ok to do *really* basic unit tests, that
> *at a later date* may be adopted and adapted to create the full
> algorithm(s)?  i can make it easier to set up.

I understand the time constraints, and that you want unit tests.
However I can't justify those under the video grant, as they will be
practically useless for the grant's purpose.

For unit tests, it's not useful to have inner video data either. It's
superfluous, a waste of time, when good tests test boundary conditions
and a few samples between.

I'm not opposed to writing tests, I know there's not many people
around, but under which budget could they be justified?

Or, to put it another way, is it a lot of work for you to have the
sub-steps counted separately? That would enable the work to not be a
waste from the video grant's perspective.

- Lauri