Doug Binks on Nostr: I managed another obvious small perf improvement to this code - do not launch a ...
I managed another obvious small perf improvement to this code - do not launch a shadow ray-cast if the contribution from path weight and in-scattering would be negligible. It's only about 1.04x perf improvement on my tests but it's very consistent and pure ALU to reduce ALU + memory traffic so likely a good future bet too.
Published at
2026-10-15 10:22:41 UTCEvent JSON
{
"id": "bbe814d33d3d6c75053fbdc2320e63d9f185eb7178a69812ed5601132ca4d6ea",
"pubkey": "5deba1e5cf488b37481efab30607c66bc56dec47b20a640c2f8d9d0d97e63f5e",
"created_at": 1792059761,
"kind": 1,
"tags": [
[
"proxy",
"https://mastodon.gamedev.place/users/dougbinks/statuses/117444428559271942",
"activitypub"
],
[
"client",
"Mostr",
"31990:6be38f8c63df7dbf84db7ec4a6e6fbbd8d19dca3b980efad18585c46f04b26f9:mostr",
"wss://relay.ditto.pub"
]
],
"content": "\n\nI managed another obvious small perf improvement to this code - do not launch a shadow ray-cast if the contribution from path weight and in-scattering would be negligible. It's only about 1.04x perf improvement on my tests but it's very consistent and pure ALU to reduce ALU + memory traffic so likely a good future bet too.",
"sig": "75eb5ee5047a8434854f6a83832a43043f74ce156867a536e7a9d534b80f8452263ecfed481baa4ed6137a484a99964747ee4b112bb5e4b79df489b758fc173f"
}
