What about RDRAND does not meet your requirements (other than it's not yet in all CPUs :-)
-
-
-
Was thinking about GPUs, where the standard practice is to generate a huge table of values in memory for randomness.
Kraj razgovora
Novi razgovor -
-
-
I like this method of generating random numbers between 0 and a prime congruent to 3 (mod 4): https://preshing.com/20121224/how-to-generate-a-sequence-of-unique-random-integers/ … I'm using this to shuffle point clouds at up to 800 million points/second, or 100M points/sec while streaming from an SSD.
- Još 6 drugih odgovora
Novi razgovor -
-
-
Ugh, as someone who's studied noise for years I've been bugging them to do this - thanks for saying it. There's even a noise() in the language specs; usually returns 0. So much for specs.
-
Yeah its totally unspecified. I swear it returned perlin noise at some point hehe
- Još 1 odgovor
Novi razgovor -
-
-
x86 has the `rdrand` instruction for this stuff. Maybe GPUs can have something like this too somehow.
-
Wasn't "rdrand" the instruction that made Destiny 2 go poof when launched on Ryzen CPUs a while back?
- Još 2 druga odgovora
Novi razgovor -
-
-
Any quick/simple demo of how this would work? Don’t quite get it.
-
I'm assuming he means you can generate data procedurally on the fly, so you don't need texture lookups? Shadertoy has a lot of nice examples https://www.shadertoy.com/view/4ttSWf
- Još 1 odgovor
Novi razgovor -
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.