Garbage Reduction, Dynamic/Temp Draw Optimizations, Int2Int Collection#882
Draft
RuffledPlume wants to merge 2 commits into117HD:masterfrom
Draft
Garbage Reduction, Dynamic/Temp Draw Optimizations, Int2Int Collection#882RuffledPlume wants to merge 2 commits into117HD:masterfrom
RuffledPlume wants to merge 2 commits into117HD:masterfrom
Conversation
RuffledPlume
force-pushed
the
AhslPredicateCache
branch
4 times, most recently
from
eca39a2 to
acfe05b
Compare
RuffledPlume
changed the title
RuffledPlume
changed the title
Contributor
Author
|
Master 
PR: 
At a glance Differences: Garbage Scene Frame Time ProcGen |
RuffledPlume
force-pushed
the
AhslPredicateCache
branch
from
ccd38a3 to
21733f4
Compare
Contributor
Author
|
Remerged with Master + further optimizations to Int Collections: |
RuffledPlume
force-pushed
the
AhslPredicateCache
branch
5 times, most recently
from
99a7049 to
ac3413b
Compare
RuffledPlume
force-pushed
the
AhslPredicateCache
branch
from
df506ab to
44017aa
Compare
RuffledPlume
force-pushed
the
AhslPredicateCache
branch
3 times, most recently
from
17b2adb to
8e16f10
Compare
Contributor
Author
|
Rebased onto Main, latest perf comparisons: Scene Load Time, Main Vs PR: Varrock Perf, main Vs PR: |
Reduced ProcGen Garbage Added Mummur Hash to reduce collisions Improved LinearProbe by caching the accessed key, instead of reading it twice Added IntHashSet Switched Maps over to IntHashSet since the additional value was unnecessary Pass in worldPos when getting tileOverride since it's already calculated Fixed Iterator generating 500MB of garbage during scene load Cache is read on multiple threads, we cannot guarantee the last alhsl value will be fully written in time Cache the last searched key result Added Bounds around multiple AABBs to allow for early out Fixed Double Linear Probe Implemented Robin Hood Hashing Dynamic Model Upload Optimizations Calculate Int Bits of Vertex position once instead of multiple times per face Added Normalized value fast path for fp32 -> fp16 conversion Calculate invLength instead of diving each axis undoVanillaShading Tweaks colorAdjust only needs to be calculated within the inner part of the if, otherwise the value isn't used We can use nDotL instead of the length Early out if value should be ignored Early out on nDotL Made FindIndex static since its consistent across all collections Refactored LastRead into a 32 Byte ReadCache findIndex optimizations Switch GLState IntSet to use IntHashSet & not the generic HashSet Tweaks Pre-Compute SINE/COSINE Float Arrays matching Perspective.SINF/COSF Tweaks Add Object Support to Int Map Collections Fixes due to master rebase Sync Sync Sync
RuffledPlume
force-pushed
the
AhslPredicateCache
branch
from
854f0d0 to
b2dc8d6
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
A FIFO cache around theModelOverride::AhslPredicate results for given Ahsl keys, cache is made up of Longs, 32 bits for the ahsl, 1 bit for the result, are 31 bits for the age used during eviction.
Cache by default starts at 16 & has an upper limit of 512 values before it'll begin evicting values by the oldest first.