kanghui0204 6f3922f318 feat: Low Precision Allreduce for PCIe based GPU (#4344) ... This PR adds a customized allreduce to TensorRT-LLM. The new allreduce is used for communication on PCIe-based GPUs via low-precision quantization, which can accelerate the PCIe allreduce process. Signed-off-by: Hui Kang <hkang@nvidia.com> Co-authored-by: Hui Kang <hkang@nvidia.com>		2025-05-20 06:53:46 +08:00
..
test_allreduce.py	Fallback to NCCL for various patterns when input size is large. (#4009)	2025-05-01 15:17:16 -07:00
test_ar_residual_norm.py	refactor: remove ParallelConfig in tensorrt_llm._torch.distributed module (#3370)	2025-04-11 15:34:20 -07:00
test_embedding.py	refactor: remove ParallelConfig in tensorrt_llm._torch.distributed module (#3370)	2025-04-11 15:34:20 -07:00
test_linear.py	refactor: remove ParallelConfig in tensorrt_llm._torch.distributed module (#3370)	2025-04-11 15:34:20 -07:00
test_lowprecision_allreduce.py	feat: Low Precision Allreduce for PCIe based GPU (#4344)	2025-05-20 06:53:46 +08:00
test_star_attention_input.jsonl	Update TensorRT-LLM (#2936)	2025-03-18 21:25:19 +08:00
test_star_attention.py	Breaking change: perf: Enable scheduling overlap by default (#4174)	2025-05-15 14:27:36 +08:00
test_user_buffers.py	fix: Remove ParallelConfig. (#3678)	2025-04-21 14:14:08 +08:00