Richard Geldreich

@richgel999

Entrepreneur at Binomial, open source dev. When I coded at corps I was at SpaceX, Valve, and Ensemble Studios. SIBO survivor. Opinions my own. He/him.

United States
Vrijeme pridruživanja: ožujak 2011.

Medijski sadržaj

  1. 2. velj

    This video shows the time for a Commodore 64 vs. a new MacBook Pro to boot, load and print a document: The MacBook barely won.

  2. 1. velj

    I really like Ben Eater's hardware videos on YouTube:

  3. 1. velj

    After an early Falcon 1 failure (which had to be crushing to SpaceX's morale!), Elon Musk said this: "For my part, I will never give up and I mean never."

    Prikaži ovu nit
  4. 1. velj
    Odgovor korisnicima

    I know Elon Musk isn't liked in many circles, but he and his team created one the world's first successful reusable rockets (Falcon 9). When one of the early Falcon 1's crashed in 2008 he said this: "For my part, I will never give up, and I mean never":

  5. 31. sij

    I took this pic a few months ago. It reminded me of something I would see in a video game!

  6. 30. sij
    Odgovor korisniku/ci
  7. 29. sij

    Amazingly, Super Mario 64's C code wasn't compiled with compiler optimizations enabled: This helped fans decompile the game into C:

    Prikaži ovu nit
  8. 27. sij

    Left=ETC1 36.63 dB Right=UASTC 45.94

    Prikaži ovu nit
  9. 25. sij

    I'm changed the way the encoder balances UASC+BC7 error - much better!

    Prikaži ovu nit
  10. 25. sij

    Close up of the UASTC alpha artifacts on the left edge I'm talking about:

    Prikaži ovu nit
  11. 25. sij

    Here's our current UASTC results (left=RGB right=alpha). Pretty good, but there are some subtle blockiness in alpha (on the left gradient) that is annoying. 16 mode UASTC: 55.71 (RGB) 54.63 (A) dB PSNR Near-optimal BC7: 55.91 52.35 Near-optimal BC1: 43.6

    Prikaži ovu nit
  12. 25. sij

    Currently debugging alpha artifacts with the new HQ universal codec. This test texture with an alpha channel from Simon Brown's website is actually really useful:

    Prikaži ovu nit
  13. 24. sij

    ASTC->BC7 transcoding failure.

    Prikaži ovu nit
  14. 24. sij

    As our HQ universal encoder gets more complex it's starting to break in odd ways. The latest issue only occurs when OpenMP is enabled on BC1 encoded blocks. We just use stb_dxt on these blocks. The code is straightforward and was working fine, so I'm not sure what happened.

    Prikaži ovu nit
  15. 23. sij

    Good stuff - thanks for applying this documentation fix!

  16. 22. sij

    Least squares refinement works great! 2 steps gets 33.91 dB, 1 step gets 33.6 dB. Well that was easy.

    Prikaži ovu nit
  17. 22. sij

    Fixed a remaining issue. stb_dxt gets 34.35 dB, the quick transcode directly from our ASTC mode 0 to BC1 gets 32.52 dB. Promising!

    Prikaži ovu nit
  18. 22. sij

    I was using the wrong lookup table to dequantize the [0,191] ASTC endpoints. I was using the table sorted by unquantized intensity, not the table for encoded ASTC endpoints (which aren't linear!).

    Prikaži ovu nit
  19. 22. sij

    First attempt at experimenting with transcoding our ASTC mode 0 data directly to BC1. Near-hopeless failure.

    Prikaži ovu nit
  20. 21. sij
    Odgovor korisnicima

    My eyes! Ok this must be disabled or this font is dead to me.

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:

    ·