The teaser for

@LukasHozda The smallest hello World in C is about 800 bytes. A long way to go.

@_chenglou @cmuratori "you" is not ""you"" , JK not JK, and all other usual disclaimers apply

@cmuratori LOL, this is awesome.

I did a comparison of Posit32 vs Float32 numbers. Posit gives such better precision for the same bit size, but is 1.7x slower (I compared software implementations).

@ThePrimeagen Please video of you sensually mumbling opcode names into the mic. Title, "ASM-R"

@cmuratori Makes sense, "muro" means outdoors wall (in spanish so I assume it does in Italian). Like a "mural" (literally "on the wall").

@Jonathan_Blow but.... but, they look so slick in the marketing material with the buttons hidden from view.

@games_inu @swadeart I mean, often times if you are using sin/cos means you are still something "wrong" and that they can be removed (they are fast, fat not faster than not computing them).

@games_inu @swadeart In GPU, sin() and cos() are incredibly fast, while atan(), acos() are super super slow (do not use). In CPU, they are all slow.

@SebAaltonen NVIDIA is under no obligation to disclose their tech (investment), and like Apple, often chooses not to. So it's really up to the research community to step up. Problem is Jensen bought much controll of that very community by hiring all engineers/researchers he could. Smart.

@Rogost_ It doesn't have to, it's 10805488/8 that should be divisible by 8.

@skdh Yes, americans are optimists by culture. Temperature begins at 32 already, you start climbing buildings at the 1st floor for some reason, every kid is a winner until proven otherwise, and a crappy job is met with "awesome, let's think of ways to improve it".

The reason I've never been able to use any open source library is that "you only need python, npm, node, yarn, make, ninja and ssh, but only some pecific versions, and then type these things in the command line like it's still the 90s, only they never work, fuck this shit".

@80brew @DJSnM I made this one (in 2009). It's an exercise on creative mathematics and hardcore programing. For the math part, you can watch this:

@SK83RJOSH @Jonathan_Blow The actual value is sqrt(1+sqrt(1/2)), close enough. Fixed the shader. Thanks! The dot(v,v) will be more expensive than abs(v.x)+abs(v.y), it involves two multiplications and one addition, instead of jus tone addition (abs values don't take any instruction)

@kmett @modelprint02 @Jonathan_Blow Not only that, the aren't absolute value instruction/calls either, those become instruction modifiers.

@LFTH_Alex @Jonathan_Blow Yes. I updated the explanation in the shader, but it's too late for the video.

@s7nfo @Jonathan_Blow Very nice! I wasn't aware of this competition.