TensorRT-LLMs/cpp/kernels
杨凯旋 61c5a53642
[#5403][perf] Conditionally enable SWAP AB for speculative decoding (#5404)
Signed-off-by: zoheth <z0heth@outlook.com>
Co-authored-by: Yao Yao <lowsfer@users.noreply.github.com>
2025-07-01 18:32:37 +08:00
..
fmha_v2 keep sm90 headsize 128 cubins (#5320) 2025-06-26 12:14:01 +08:00
xqa [#5403][perf] Conditionally enable SWAP AB for speculative decoding (#5404) 2025-07-01 18:32:37 +08:00