Hi, this is a friendly reminder:
Put "higher is better" or "lower is better" next to you benchmark data or graphs.
Thank you so much.
Remote
dram🎀
@dramforever@mastodon.social
mastodon 4.8.0-nightly.2026-10-06caps, mosfets, and the happiest thoughts in the universe
https://dram.page
https://github.com/dramforever
https://cohost.org/dram
https://ko-fi.com/dramforever
https://codeberg.org/dramforever
0 Followers
0 Following
45 Posts
Joined June 26, 2024
pro-tip: instead of foo/bar/baz use meow/woof/floof. it's just better.
your password must include an upper case letter, a lower case letter, a combining diacritic, a cjk unified ideograph, an emoji, four different whitespaces, three different scripts, and a partridge in a pear tree
the existence of "backwards long jump" implies the existence of "forwards setjmp"
the phrase "peroxyacetic acid" taken literally means "most-acid acid acid"
i'm not even that big a touhou fan but today i managed to misspell "slander"
I don't know how to say this more clearly.
If you're asking me something, having "asked an AI" counts *negative* points towards doing your own research.
If you don't understand something and think it's okay to parrot what an LLM says you're *actively sabotaging* the integrity of your knowledge. Which I cannot stop but, like, if you ask me afterwards, you're basically telling me "I'm intentionally making it hard for me to learn. Teach me anyway."
Read the damn links. Play with your own code.
Replying to
@david_chisnall @whitequark there's a tom scott video about that! i've watched it!
... there's no way you *actually* make tea like that ... right?
Replying to
@0xabad1dea@infosec.exchange i'm like a minute and half in and the runner has already said "binary64" i think i know what i'm doing tonight
the existence of emudev implies the existence of simudev
Replying to
@CyReVolt@mastodon.social @unsafe@m.webtoo.ls can't believe i had my own "legend of zelda speedrun" moment
Replying to
@theartlav @0xabad1dea loongarch hasn't been here for that many years. if you have anything before 3A5000 (~2019) it was MIPS + lots of extensions
Replying to
Replying to
@joe@f.duriansoftware.com int(Range) gives you the interior of the range, making it an open a < x < b
Replying to
@iximeow@haunted.computer i like you're still going on and on about plt, gotpcrelx, and the lack of the latter, and i'm rooting for you
batt-man, like as in batteries, the kind that stores and provides electric energy
two fried shrimps that look *exactly* the same
"god bless ttbr0_el1"
Replying to
@whitequark ... apparently symlinks are handled as symlinks. i don't know what i expected.
Replying to
@dysfun @0xabad1dea oh hey LATX is open source now https://github.com/lat-opensource/lat
so it's just all source code and no docs now i suppose
Replying to
@whitequark@mastodon.social @mokomull@macaw.social in case you're wondering, you can tell by the lui a0, 0x10000
mips lui has only 16 bits of immediate, so this can't be mips. riscv has 20 bits so yay
Replying to
@chris@social.losno.co @thephd@pony.social kinda thinking, this could conceivably have been the case on 32-bit x86 with "cdecl" calling convention where calling functions with parameters at all was unduly expensive?
processors at the time had less architectural registers, probably less stack optimization, and had to deal with this args-on-stack thing
same goes for more ancient 16-bit and 8-bit machines, probably.
my current hypothesis is that if you ping copilot to work on someone else's pr, it proceeds as if it's a copilot pr, and just replaces their pr description and title.
not that this is by any means normal.
Boosted by @kkarhan@jorts.horse
Replying to
@whitequark@social.treehouse.systems I genuinely love this. This is a "front end" feature, right? So it's arguably just handling otherwise correctly formatted tar files better.
... and now I'm curious, what does git-pages do to symlinks
Replying to
@pikhq i think the "don't give a shit" part is so important. I keep hearing about things like how AI is getting better, but the reality is that doesn't even matter. "Vibe coding" is *already* long past the point of diminishing marginal perceived utility. The new models you've got in the past ~year are so much more feel-good than actually useful for anything.
Replying to
@AsahiLinux been trying out the bleeding edge 6.16, it's honestly great to see asahi linux picking some steam back up! keeping up with rebasing is not a small task, honestly, so thanks for the great work of y'all for keeping the linux updates coming
Replying to
@chris@social.losno.co @thephd@pony.social 💭 "the average number of arch regs, and phys regs, per cpu core in the entire world over time"
Replying to
@eri i'm so used to this not working your post made me realize there's actually a webcam on my computer
You've probably heard of this statement by now, but, well, until today I didn't know the source code still existed!
See:
- PDP-7 Unix resurrection project https://github.com/DoctorWkt/pdp7-unix/blob/master/src/sys/s3.s
- The scanned files https://www.tuhs.org/Archive/Distributions/Research/McIlroy_v0/
- The Unix Heritage Wiki https://wiki.tuhs.org/doku.php?id=systems:pdp7_unix
Replying to
add milk
i wonder if you can use zink on a zynq
Replying to
@regehr i don't have a concrete one that i can share immediately, but i had someone ask me "why are atomics needed? this works perfectly fine" on a concurrent spsc circular buffer queue that worked entirely relying on compilation unit barriers acting as compiler barriers, word sized load and store being atomic, and x86 tso behavior. running the same program with tests on arm immediately breaks, and so does turning on -flto
Replying to
@niconiconi@mk.absturztau.be maybe systemd-nspawn -D --console=interactive
fight fire with fire