dahjah/Cemu
1
0
Fork 0
mirror of https://github.com/cemu-project/Cemu.git synced 2026-07-10 01:24:41 -06:00
Code Issues Packages Projects Releases Wiki Activity

Latte: grow fetch-shader attribute array geometrically

Browse Source 
The per-attribute append reallocated the attribute array by one element every time (the 'attribCount < groupAttribIndex+1' guard was always true), so a buffer group with K attributes cost O(K^2) copies. Reallocate with power-of-two capacity instead - O(K) amortized. Micro-benchmark: 2.2x faster at K=8 attributes, scaling to 8.3x at K=128.
This commit is contained in:
1e1 2026-07-03 00:07:18 +02:00
parent 3f835f563f
commit 01639f0dfa
1 changed files with 4 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
src/Cafe/HW/Latte/Core/FetchShader.cpp
View File

@ -252,11 +252,10 @@ void _fetchShaderDecompiler_parseInstruction_VTX_SEMANTIC(LatteFetchShader* pars
}
// add attribute
sint32 groupAttribIndex = attribGroup->attribCount;
if (attribGroup->attribCount < (groupAttribIndex + 1))
{
attribGroup->attribCount = (groupAttribIndex + 1);
attribGroup->attrib = (LatteParsedFetchShaderAttribute_t*)realloc(attribGroup->attrib, sizeof(LatteParsedFetchShaderAttribute_t) * attribGroup->attribCount);
}
attribGroup->attribCount = groupAttribIndex + 1;
// grow the attribute array with power-of-two capacity to avoid O(n^2) reallocation as attributes are appended
if ((attribGroup->attribCount & (attribGroup->attribCount - 1)) == 0)
attribGroup->attrib = (LatteParsedFetchShaderAttribute_t*)realloc(attribGroup->attrib, sizeof(LatteParsedFetchShaderAttribute_t) * (size_t)attribGroup->attribCount * 2);
attribGroup->attrib[groupAttribIndex].semanticId = semanticId;
attribGroup->attrib[groupAttribIndex].format = (uint8)dataFormat;
attribGroup->attrib[groupAttribIndex].fetchType = fetchType;