Paul Khuong

@pkhuong

Stripping one layer of data update logging at a time. Views are my own, etc., etc.

New York, NY
Vrijeme pridruživanja: lipanj 2010.

Tweetovi

Blokirali ste korisnika/cu @pkhuong

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

  1. Prikvačeni tweet
    15. velj 2017.
    Odgovor korisnicima

    Gödwin's law.

    Poništi
  2. prije 5 sati

    Today, I tweaked three instructions and structured some code to work around this random speed bump :x

    Poništi
  3. 4. velj

    I had to help student sue ex-supervisors who wouldn't share datasets, but that's a whole new level of nightmare.

    Poništi
  4. 3. velj

    What makes a chicken soup "BTS inspired?"

    Poništi
  5. 3. velj

    Couldn't agree more with this post on hourglass interfaces for data structures!

    Poništi
  6. 3. velj

    I once had a cool optimisation for classifying geolocations by Zip code that happened to move Kansas City to the wrong state. My american colleagues had a good laugh :x

    Poništi
  7. 1. velj

    It's not every day that a read-after-write makes a server spend 4% of its CPU cycles in one instruction. Wrote that one up here:

    Poništi
  8. 1. velj

    I rarely wish I had a MS Project license, but when I do, it's to visualise out of order execution with Gantt charts.

    Poništi
  9. 1. velj

    Type erasure and GC are orthogonal… (might as well add oil to the generics fire)

    Poništi
  10. proslijedio/la je Tweet
    31. sij
    Odgovor korisnicima

    My point here being: exactness is hard, yes, but also practically irrelevant. And OoO makes approximations easier because its whole point is that it smooths out all the jagged edges. Things get more continuous and simple models give better matches.

    Poništi
  11. 31. sij

    Would we have a broader generic coding toolbox if CS people had to spend some time coding directly with the BLAS interface?

    Poništi
  12. 31. sij

    Worst math of the travelling salesman movie so far: complexity theory expert refers to "NPSPACE" as though that Savitch didn't prove that was the same as PSPACE in 1970.

    Poništi
  13. 30. sij

    Measuring progress on this rewrite by the size of the perf recorded data for the same input ;)

    Poništi
  14. 29. sij

    A new tiny CW equilibrium! Reminds me of an old Stone-Spiral experiment described in one of the old newsletters.

    Poništi
  15. 28. sij

    Feels like I regularly don't want likely / unlikely as much as fallthrough-when-true / fallthrough-when-false :|

    Poništi
  16. 26. sij

    Le serpent de passe-partout qui mange une pomme... référence biblique inattendue.

    Poništi
  17. 25. sij

    In C99 and C11, a non-static inline function (i.e., one with a guaranteed single out of line definition) can't refer to static inline functions. N.B., the out of line definition doesn't have to be identical to any/all the inline ones. Spec badness?

    Poništi
  18. 22. sij

    RIP fairway :/

    Poništi
  19. 22. sij

    Is there a pithy name or cool reference for this idea?

    Prikaži ovu nit
    Poništi
  20. 22. sij

    Sometimes you want to generalise something, and you say to yourself: "I know, I'll just abstract out behaviour with closures/function pointers/late binding." Maybe try to figure out how to use different behaviour to generate a much less abstract piece of intermediate data…

    Prikaži ovu nit
    Poništi
  21. 21. sij

    Correction, that's s/sqrt log n/7/, but I still stand behind the simplification: no retry logic in Las Vegas algorithms that should fail much less frequently than hardware.

    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:

    ·