|
CPColin posted:*thinks about a piece of unwritten code* "wow it's beautiful"
|
# ¿ Mar 3, 2020 18:26 |
|
|
# ¿ May 9, 2024 10:13 |
|
Subjunctive posted:my friend, my god, is there some assumption that we're double checking posts around here??
|
# ¿ Apr 6, 2020 22:14 |
|
tinaun posted:rust used to have a rule where no keyword or std type should have a name longer than 5 letters, so you had ret for return and loop instead of continue (and even sillier stuff like iface instead of trait)
|
# ¿ Jul 13, 2020 21:59 |
|
software downloadable content
|
# ¿ Jul 14, 2020 18:08 |
|
one of my favorite lines from a giant postmortem was "there was a point when IC's stopped believing in the overall schedule" because I could've narrowed when that switch flipped for me down to a week, possibly a day
|
# ¿ Jul 14, 2020 21:42 |
|
Cybernetic Vermin posted:i am honestly perplexed why you'd be especially doubtful. not that cl is some panacea, but it does not seem that hard to improve on verilog, and starting from a language with excellent macro facilities seems a good plan? HDL's are writing a description of a circuit, what particular strengths of lisp would lend well to it? in the context of "improve on verilog" my first step would be to check out system verilog, SVTB, OVM, you know the actual improvements to the language, tooling, ecosystem based on practitioner feedback over the past few decades? like making named argument lists into bundles was one of those whack-whole-class-of-bug improvements at the time. but sure, lisp with it's quick preference for infinite recursion, makes sense too
|
# ¿ Jul 29, 2020 14:45 |
|
Zlodo posted:can't wait circa ten years from now when horribly large piles of poo poo built with it turn up everywhere 8 years ago i worked on an embedded product where the testbench GUI was written in excel
|
# ¿ Mar 3, 2021 15:57 |
|
Cybernetic Vermin posted:yeah, it might not be a good idea even medium-term, but to just knock out a macro assembler for initial experiments it actually sounds like a pretty sane workflow, and uses excel features largely as intended. you might be able to find worse in GPU land. there's only like ~5 global teams that build them, it's all incestuous tribal knowledge bouncing around and plenty of "its our driver talking to our FW talking to our HW" where you can really hide some ghastly edge cases with "dont do that"
|
# ¿ Mar 10, 2021 19:03 |
|
Cybernetic Vermin posted:otherwise there is presumably some way (on a "real" avx-512 cpu) of regaining this 2k of register file wasted in a thread that has once run a bit of avx but no longer cares about the register content? iirc "no"
|
# ¿ Aug 20, 2021 22:22 |
|
BobHoward posted:as forsyth mentions, intel management was just hostile to gpus. even using larrabee for graphics was politically difficult; not only did it ultimately get killed but as forsyth mentions, intel's regular gpu team was on a leash and not allowed to even try to build something bigger. (there was likely no technical limitation preventing them from scaling up, by their nature gpus scale up with relative ease) ehhhhh idk about this? kinda loses the thread
|
# ¿ Aug 22, 2021 19:35 |
|
pseudorandom name posted:well, do they?
|
# ¿ Apr 25, 2022 19:35 |
|
FlapYoJacks posted:c will exist so long as kernels, bare metal, and RTOS’ exist. Which is forever. that said, idk it's not nearly as relevant as the FFI stuff
|
# ¿ Apr 26, 2022 17:42 |
|
gonadic io posted:what on earth weren't you one of less than five (5) humans on earth trying to do basic bare metal cortex m3 bringup with rust and hitting all sorts of issues?
|
# ¿ Apr 26, 2022 19:13 |
|
gonadic io posted:like, 4 years ago yeah. things are 100x better now including linux kernel contributions in rust approved by linus, like 5 different fully featured rtoses that exist. alright? i dont think this situation requires the breathless surprised you feigned at the outset that "MOST" folks banging on about this poo poo don't work in embedded and don't think about it
|
# ¿ Apr 26, 2022 21:29 |
|
BobHoward posted:e.g. one of the lessons of risc is that more than one memory reference per instruction is Bad
|
# ¿ Apr 28, 2022 23:59 |
|
Athas posted:I had a PL experience last week. I was external examiner at a university, for a project course where the students had to design and implement their own language from scratch. This is in many ways an ill conceived idea, as these are second year students who have seen only C#, Java, and maybe a bit of Javascript. ... back in school (in the dark ages) I took an ISA class somewhere in the 300 range for the EE track one assignment was optimizing a layout, like how big and how many sets/ways should the icache vs. dcache be, how many gates on the branch target buffer, etc. and apparently so many students complained that they didn't "get to design their own ISA" that the prof felt a need to address it. he explained that like... maybe a dozen architects get to do that greenfield fresh ISA design ever in their careers, optimization is a heck of a lot easier to grade, and you'd probably gently caress up real ISA design anyway
|
# ¿ Jul 5, 2023 13:08 |
|
my knowledge is hopelessly out of date, but I recall the x87 registers having a huge penalty? perhaps it's just when you try to switch back and forth between that and the XMM/YMM stuff, but I really thought that nuked the pipeline at some point
|
# ¿ Oct 18, 2023 01:04 |
|
big v6 fan here, v6 is always the best
|
# ¿ Oct 18, 2023 22:35 |
|
BobHoward posted:leaving the past behind might happen sooner than you think BobHoward posted:leaving the past behind might happen sooner than you think BobHoward posted:even if intel manages to get buy-in from microsoft BobHoward posted:microsoft holdin my breath here
|
# ¿ Oct 19, 2023 17:29 |
|
akadajet posted:I mean. it’s happened before for amd64. and windows has been built for different cpus for a long time now jumping to a new thing isn't the same as not requiring support for the old thing
|
# ¿ Oct 19, 2023 19:07 |
|
leper khan posted:A lot better tbh
|
# ¿ Nov 6, 2023 17:46 |
|
|
# ¿ May 9, 2024 10:13 |
|
having a hard time reading the python complaints from a few pages back as someone who used to get paid to maintain perl written by EE's "oooh my modules took me an hour to sort out file paths" there were functions with no call sites, they'd just bodge up the name on the spot and eval() it. you'd weep
|
# ¿ Jan 17, 2024 21:21 |