Vincent Huang 8505c3ad88 Optimize gemm perf for gpt-oss in Spark ... 1. support bias epilog for cublaslt backend 2. replace torch native mm with trtllm mm for sm121 gpt-oss 3. add lookup table for the gpt-oss gemm for sm121 Signed-off-by: Vincent Huang <vincenth@nvidia.com> Signed-off-by: list <58580514+farazkh80@users.noreply.github.com>		2025-09-30 17:25:58 +00:00
..
cmake	feat: reduce unnecessary kernel generation (#5476)	2025-07-04 14:37:49 +08:00
include/tensorrt_llm	Optimize gemm perf for gpt-oss in Spark	2025-09-30 17:25:58 +00:00
kernels	[fmha] fixes: fp8 kernels, assert removal, ampere-style sinks	2025-09-25 04:58:54 +00:00
micro_benchmarks	[TRTLLM-7319][perf] Fuse slicing into MoE. (#6728)	2025-08-25 16:52:30 -04:00
tensorrt_llm	Optimize gemm perf for gpt-oss in Spark	2025-09-30 17:25:58 +00:00
tests	pre-commit	2025-09-25 05:02:17 +00:00
CMakeLists.txt	[None][feat] Enable nanobind as the default binding library (#6608)	2025-08-22 09:48:41 +02:00
conandata.yml	infra: add conan (#3744)	2025-04-30 11:53:14 -07:00
conanfile.py	feat: large-scale EP(part 6: Online EP load balancer integration for GB200 nvfp4) (#4818)	2025-06-08 10:25:18 +08:00
libnuma_conan.py	fix cuda driver link issue with driver version less than 12.3 (#5025)	2025-06-10 15:27:39 +08:00