iron
iron
### Problem Description // Define a map of rsmi status codes to amdsmi status codes const std::map rsmi_status_map = { {RSMI_STATUS_SUCCESS, AMDSMI_STATUS_SUCCESS}, {RSMI_STATUS_INVALID_ARGS, AMDSMI_STATUS_INVAL}, {RSMI_STATUS_NOT_SUPPORTED, AMDSMI_STATUS_NOT_SUPPORTED}, {RSMI_STATUS_FILE_ERROR, AMDSMI_STATUS_FILE_ERROR}, {RSMI_STATUS_PERMISSION, AMDSMI_STATUS_NO_PERM},...
### Problem Description I need a api to get compute units and FP32 and FP64 units in amdsmi, I need this api to get compute performance, like nvidia's nvmlDeviceGetNumGpuCores to...
When I execute tritonserver --model-repository=./repo-backend-src/examples/model_repos/recommended_models, a Segmentation fault was occured, error message are as follows: src/examples/backends/recommended/build# /opt/tritonserver/bin/tritonserver --model-repository=/home/kylin/triton/server/build/triton-server/_deps/repo-backend-src/examples/backends/minimal/build/_deps/repo-backend-src/examples/model_repos/recommended_models I1009 07:47:05.764978 59477 model_lifecycle.cc:461] loading: batching:1 I1009 07:47:05.766123 59477 recommended.cc:63] TRITONBACKEND_Initialize: recommended...
We must free ctx and model after program exited.
### Git commit 1682e39aa5bb1699fae3f760450be2e76d35a6a1 ### Operating systems Linux ### GGML backends CUDA ### Problem description & steps to reproduce Tell CMake where to find the compiler by setting either the...
When shutting down linux operating system, triton server needs taking 1 to 2 seconds to stop triton server，especially when the Python backend has already been started， it leads to a...