[libre-riscv-dev] presentation on graphics drivers & doing it right the first time

[Luke Kenneth Casson Leighton]

On Friday, October 4, 2019, Jacob Lifshay <programmerjake at gmail.com> wrote:

> found these slides interesting:
> https://xdc2019.x.org/event/5/contributions/325/attachments/
> 416/666/How_to_not_write_a_back-end_compiler.pdf


We have a minor problem of not knowing what it is that we don't know, so
consequently are highly likely to make design mistakes.

Intel has so many engineers that they can afford the time and money to
iterate a la "three revisions" which go from

Don't know what and don't know how
Know what but still don't know how
FINALLY know what AND how

>From those slides it looks like Intel is at stage 2.

We really cannot afford to repeat that by arrogance of "we know best" or
"doing everything from scratch is best because other people made the wrong
decisions".

So, very good find - one for the subconscious backburner, to use to avoid
those costly mistakes, even though half the language is stuff that we don't
yet have direct experience with.

This is why I spend a lot of time searching for other people's high profile
working code and splice it or adapt it.

On the gcc port that will be critical. Hiring compiler engineers at
business to business cost levels is half a MILLION dollars per engineer per
year.

L.



-- 
---
crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68