Create accountLogin

Recent searches

No recent searches

Search options

Only available when logged in.
mastodon.gamedev.place is one of the many independent Mastodon servers you can use to participate in the fediverse.
Mastodon server focused on game development and related topics.

Administered by:

Server stats:

5.4K
active users

mastodon.gamedev.place: AboutProfiles directoryPrivacy policy

Mastodon: AboutGet the appKeyboard shortcutsView source codev4.3.7

Public
thomastc | frozenfractal

Hmm, generating good hashes for use in simplex noise is surprisingly expensive. Has anyone ever tried using the CRC32 instructions that come with SSE4.2 for this? They only take 1 clock cycle, much faster than anything I've come up with.

Public *
thomastc | frozenfractal

I tried. It's bad. It *is* faster than permutation tables though.

Bottom center panel is my attempt.

A grid of 3x3 images of simplex noise. The one at the bottom center has a repeating pattern.
Public
thomastc | frozenfractal

AES instructions don't fare any better.

A similar image as before, but with a different repeating pattern.
Public
thomastc | frozenfractal

I guess I should bring out a profiler and actually see where the bottlenecks are.

Meanwhile, if someone knows a fast hash that takes 128 bits as input (x: i32, y: i32, seed: u64), works in AVX __mm256i registers and has good entropy in the lower bits, I'm all ears.

Public
thomastc | frozenfractal

@bitinn Thanks, that's the sort of thing I was looking for. XXH3 doesn't beat the permutation table for performance (38 ms total runtime instead of 34 ms), but the visual quality is good.

ExploreLive feeds
About