Conversation
RISC-V ISA string stuff makes my head hurt.
Claims that the spec definitions are stable just blows my mind. I'd swear that the biggest pain in the arse patches are those dealing with passing it that to a toolchain, particularly the removal of Zicsr/Zifencei from I crap.
1
0
2
@dickon I'm not really aware of the timing of those changes, but I assume they changed midway through your software support for the extensions. Did it also change late enough to affect hardware tapeout (although you probably cannot answer that) and/or after public review?
1
0
1
@dickon Yah, I saw Lawrence's latest iteration earlier.
I just hope that noone ends up with some pre- & non-1.0 compliant implementation.
2
0
0
@dickon I wish I was working on that sort of thing when I was an intern.
1
0
1
@dickon Oh I totally didn't interpret as figurehead. I know we talked about the crypto support patches at FOSDEM w/ him, so I figured it was a "he sent it, but we all worked on it" type of comment.
Sounds like you guys are doing a pretty good job of internships.

> He's also the team's best git wrangler, which is actually worrying me; he goes back to uni in September...

Oof. git-wranger-in-chief is kinda my job title too, although I've managed to gradually improve everyone's ability over the last few years.
1
0
1
@conor @dickon it happens all the time, the way RVI handle schedules is just wildly unsuitable for building hardware and pretty much everyone gets bit by it at some point
1
0
1
@dickon @conor that happens a lot, it's how the HW folks end up getting burned
0
0
1