[libre-riscv-dev] Vulkanizing

Luke Kenneth Casson Leighton lkcl at lkcl.net
Wed Feb 19 23:19:13 GMT 2020

On Wed, Feb 19, 2020 at 11:04 PM Scheming Pony
<scheming-pony at protonmail.com> wrote:
> "I'm back," -- Ty Cobb


> From CEPL's project page (made a trip to the GitStar to retrieve):
> CEPL (Code Evaluate Play Loop)
> > Its definition of success is making the user feel that GPU programming has always been part of the languages standard.
> It uses a modern shader language called Varjo (Lisp), which compiles to some antique C thing called GLSL.  Which is then compiled to IR.  Why?

honestly?  you might not like this answer: "don't know, don't care".

the reason for that answer is very simple: if i "worried" about every
single level of detail, why, why, why, why, absolutely nothing would
get done.

this is what allowed me to be a successful network-reverse-engineer
when everyone i was working with was going "what the f*** why oh god
why did they do that, surely they couldn't possibly have been that
stupid, therefore this must not possibly be valid, therefore i am
going to ignore it, therefore i cannot progress".

> Why can't we just walk Varjo, an AST already, ("Backwards, I think") to generate the custom op codes we need?

i am tempted to ask "why would be do that?" but i'm not going to...
because once that "tree of conversation" is investigated, the answer
*still will not matter*, because we chose SPIR-V and Vulkan.

we chose SPIR-V because the entire industry, as a whole, has chosen
Vulkan (and SPIR-V).  why did they choose that?  answer: "don't know,
don't care".  they just did, they likely had good reasons: personally,
the fact that the industry *has* chosen it is "good enough for me".
therefore we go with it.

we *have* to be quite pathologically focussed, here, stephen.

> CEPL ported to MIT/Scheme would give a cool graphics monitor I think, for the dev board.  The MIT/Scheme dev list has given me a wall of papers to read (which I already have) I guess in payback for sending them a wall of my project links.
> Man this is getting exciting.  @Doug, do you know "cbaggers" on GitHub?  Could you forward this to him?  Promise him I won't contact him directly--he is busy on some really great video game.

More information about the libre-riscv-dev mailing list