turboderp 1f220f6e50 GEMM/MGEMM: Add autotuning with disk cache, remove static tuning table

2026-04-29 15:55:54 +02:00

..

codebook_eval.py

GEMM: Lock MCG multiplier to 0xCBAC1FED and MUL1 to 0x83DCD12D. Make MCG the default codebook for new models.

2025-10-12 22:09:01 +02:00

gumbel_eval.py

Initial commit

2025-04-06 14:42:49 +02:00

kv_quant_exp.py

Add cache quantization

2025-04-22 21:52:33 +02:00

qgemm_benchmark.py

GEMM: Lock MCG multiplier to 0xCBAC1FED and MUL1 to 0x83DCD12D. Make MCG the default codebook for new models.

2025-10-12 22:09:01 +02:00