DeepSeek/DeepGEMM
1
0
Fork 0
mirror of https://github.com/deepseek-ai/DeepGEMM synced 2025-06-26 23:15:49 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Performance: reducing the percentage of FFMA interleaving yields a slight performance gain, roughly 0.5%

Browse Source
This commit is contained in:
sazc 2025-03-05 17:50:22 +08:00
parent 68fc742572
commit fcd1dcd99d
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
deep_gemm/jit/interleave_ffma.py
View File

@ -73,7 +73,7 @@ def parse_registers(line):
def modify_segment(m, name, ffma_lines):
num_lines = len(ffma_lines)
num_lines = (len(ffma_lines) * 9 // 16) // 2 * 2
assert num_lines % 2 == 0
le_bytes, new_le_bytes = [], []