Troels Henriksen

@sigkill_dk

Postdoc hacking away at data-parallel functional languages on GPUs and others ().

Vrijeme pridruživanja: prosinac 2015.

Tweetovi

Blokirali ste korisnika/cu @sigkill_dk

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @sigkill_dk

  1. 21. sij

    I answered some Futhark questions for a YouTube channel. Do you ever get used to hearing your own voice?

    Poništi
  2. 19. sij

    "The Spirit of Futhark", 2020

    Poništi
  3. 14. sij

    Note that chmod'ing +x works fine, so it's not obviously a permissions issue. Also weird that I have no problem on my user account, but others have on theirs.

    Prikaži ovu nit
    Poništi
  4. 14. sij

    I have a problem where open("foo", O_CREAT, 777) fails to set the execute bit on "foo" for some users, when they create files in their NFS-mounted file systems. umask looks fine. What could be the reason?

    Prikaži ovu nit
    Poništi
  5. proslijedio/la je Tweet
    5. sij

    Join me! ACM members, sign a petition supporting open access, asking to withdraw signature from anti-OA letter and make OA available at cost. OA is good for science, and good for ACM!

    Prikaži ovu nit
    Poništi
  6. 5. sij
    Poništi
  7. 5. sij

    PhD position for research in high-performance "deep probabilistic programming":

    Poništi
  8. proslijedio/la je Tweet
    4. sij
    Odgovor korisniku/ci

    Nah.

    Poništi
  9. proslijedio/la je Tweet
    1. sij

    With Futhark, you can speed up your Python data science code with a little bit of functional programming:

    Poništi
  10. 31. pro 2019.

    Most significant personal development in the past decade: accepting that "decade" is just a cultural convention, and that the decade thus really was 2010-2019, not 2011-2020.

    Poništi
  11. 28. pro 2019.

    Actually, it looks like GCC has more trouble than I do: tests/tiling/notile_loopvariant.c:2604:5: internal compiler error: Segmentation fault

    Prikaži ovu nit
    Poništi
  12. 28. pro 2019.

    My dependent type system implementation can type-check all my test programs and benchmarks now! Of course, that doesn't mean it's correct...

    Prikaži ovu nit
    Poništi
  13. 23. pro 2019.

    Finished our newly reworked course on Parallel Functional Programming. Maybe the material can be useful to others as well:

    Poništi
  14. 15. pro 2019.

    Hey , , go back to marketing GPUs like this:

    Poništi
  15. 9. pro 2019.

    ISPC releases are fairly infrequent, and mostly fix bugs and add new code generator backends. Language changes are few, because it's (almost) *done*. That is a good thing!

    Prikaži ovu nit
    Poništi
  16. 9. pro 2019.

    ISPC is an awesome example of a language that does just one (hard!) thing, does it well, and makes it easy to integrate with the outside world.

    Prikaži ovu nit
    Poništi
  17. 9. pro 2019.

    Giving a lecture on ISPC soon. The more I look at it, the more I fall in love with both the language, but also the excellent documentation.

    Prikaži ovu nit
    Poništi
  18. 24. stu 2019.

    Cool! I wrote and directed this song! I have no idea what an English speaker would see in it. (It's a song in Danish about being frustrated at not being allowed to program in Standard ML)

    Poništi
  19. 9. stu 2019.

    I ported another ray tracer to Futhark. This one stretched the language limitations a bit more:

    Poništi
  20. 6. stu 2019.

    How can I in tyool 2019 still crash my GPU by launching a kernel that contains an infinite loop?!

    Poništi

Čini se da učitavanje traje već neko vrijeme.

Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.

    Možda bi vam se svidjelo i ovo:

    ·