skavi

11 ms·

1. 211.▲
   by skavi 2y ago

   Homeless shelters really do suck from what I’ve read. Theft in particular is an issue. I don’t think it’s fair to extrapolate from that statement that people wouldn’t take up the offer of shelter if quality and safety were improved. There w
2. 212.▲
   by skavi 2y ago

   That’s the first I’ve heard of the S III Neo. What an odd product. Came out 2 years after the original S III, in practically the same chassis, but with a few parts swapped. I wonder why it was made.
3. 213.▲
   by skavi 2y ago

   I believe they mean “physical” as in not implementing in software. So integrated GPUs are perfectly fine as well.
4. 214.▲
   by skavi 2y ago

   The two generation old Gracemont already beat Skylake [0]. Skymont can beat Raptor Cove [1] (The big core that was paired with Gracemont). [0]: https://www.anandtech.com/show/16881/a-deep-dive-into-intels... [1]:
5. 215.▲
   by skavi 2y ago

   That story is from 2021, which is before Musk purchased Twitter.
6. 216.▲
   by skavi 2y ago

   that sounds like a super fun class. wonder if it’s still offered?
7. 217.▲
   by skavi 2y ago

   It’s been a while since I looked, but I remember SVE2 being much more usable than SVE. A64FX was SVE IIRC. I think SVE did not do a great job of fully replacing NEON.
8. 218.▲
   by skavi 2y ago

   I think it's somewhat unfair to ask for real world examples when there really aren't many people writing optimized SVE code right now. Probably because there are hardly any devices with the extension. I think the transition from A
9. 219.▲
   by skavi 2y ago

   This isn’t a convincing explanation to me. There are plenty of variable latency instructions on existing high performance arm64 cores.
10. 220.▲
    by skavi 2y ago

    I misremembered. Looks like it is only Apple. I appreciate the correction.
11. 221.▲
    by skavi 2y ago

    Masked instructions primarily. But apart from that it’s just a more complete ISA vs NEON. More comparable to AVX512/AVX10. > 2x256b is only beneficial over 4x128b if you're limited by decode width This is only true if we ignore
12. 222.▲
    by skavi 2y ago

    Does anyone have insight into why arm CPU vendors seem so hesitant about implementing SVE2? ~They seem~ *Apple seems to have no issue with SSVE2 or SME. Edit: Only Apple has implemented SSVE and SME I think.
13. 223.▲
    by skavi 2y ago

    consumer robotics
14. 224.▲
    by skavi 2y ago

    I wouldn’t expect most companies using Rust to require significant prior Rust experience. At least, mine does not.
15. 225.▲
    by skavi 3y ago

    What’s the relationship between llamafile and llama.cpp these days? IIRC, it was originally llama.cpp compiled with Cosmopolitan Libc. I’ve read more recently of optimizations in llamafile with comparisons against llama.cpp. Are those pushe
16. 226.▲
    by skavi 3y ago

    I think it’s incredibly useful today to have both of these annotations. I frequently scan for all instances of `?` or `.await` in functions (though, unfortunately, for various reasons this won’t show you everywhere these effects are produce
17. 227.▲
    by skavi 3y ago

    SVE2 includes BDEP and BEXT which should let you do this absurdly quickly.
18. 228.▲
    by skavi 3y ago

    See footnote 1
19. 229.▲
    by skavi 3y ago

    You’d rather they be a transparent conduit for state propaganda? Better to remind readers that the military sees value in this article existing.
20. 230.▲
    by skavi 3y ago

    I think you’ll find far better answers than I can provide with a web search.
21. 231.▲
    by skavi 3y ago

    If you manage to click on the correct image, you’ll get to this page [0]. Then you can find a link to the “interactive demo” [1]. Weirdly well hidden. [0]: http://daslab.seas.harvard.edu/projects/evolution/ [1]:
22. 232.▲
    by skavi 3y ago

    One (admittedly obvious) difference is that it runs on Node/V8 instead of Bun/JSC.
23. 233.▲
    by skavi 3y ago

    You can still buy an Apple USB “SuperDrive” today. Getting a large, rarely used part out of the laptop chassis is a good thing.
24. 234.▲
    by skavi 3y ago

    What does power usage look like? Is it retained or immediate? Does the API feel similar to any existing UI library?
25. 235.▲
    by skavi 3y ago

    This is exciting. I genuinely enjoyed working with Druid (though my time with it was limited). I was wondering if Xilem would get to that level of usability. Sounds like they’re planning to go way further.
26. 236.▲
    by skavi 3y ago

    Really cool idea. Unfortunately, the proportions are somewhat broken on the Oasis (the text bar stretches a bit less than half way across the screen). Was the page designed to be pixel perfect for the Paperwhite?
27. 237.▲
    by skavi 3y ago

    what does “a synergy of mobile design and desktop design” mean? It’s the same cores, but more of them. iirc, they’ve made cache structures larger as well.
28. 238.▲
    by skavi 3y ago

    Arm’s teams work at the same time. One does a longer term design while the other is shipping.
29. 239.▲
    by skavi 3y ago

    Is this finally what the Sophia-Antipolis team has been working on? IIRC, the last few years have all been Austin.
30. 240.▲
    by skavi 3y ago

    Per chip? Not the full story when discussing a system which can integrate multiple. The Orin has more memory bandwidth than an RTX 4050 even though the latter uses GDDR6. The M3 Max has double the bandwidth of the Orin, but also uses LPDDR5