/opt/dtk-23.10/cuda/bin/nvcc    -DWMMA -O2 -g -DNDEBUG -O3   -fuse-ld=gold CMakeFiles/llama_triton_example.dir/llama_triton_example.cc.o  -o ../../../bin/llama_triton_example   -L/opt/mpi/lib  -L/opt/dtk-23.10/cuda/targets/x86_64-linux/lib/stubs  -Wl,-rpath,/opt/mpi/lib: -lcublas -lrocblas -lcudart ../../../lib/libLlamaTritonBackend.a ../../../lib/libTransformerTritonBackend.a ../../../lib/libmpi_utils.a ../../../lib/libnccl_utils.a ../../../lib/libnvtx_utils.a ../../../lib/libword_list.a -lpthread ../../../lib/libLlama.a ../../../lib/libnccl_utils.a /opt/dtk-23.10/cuda/lib64/libnccl.so ../../../lib/libcublasMMWrapper.a ../../../lib/libcublasAlgoMap.a -lcublas -lcurand ../../../lib/libDynamicDecodeLayer.a ../../../lib/libnvtx_utils.a ../../../lib/libTopKSamplingLayer.a ../../../lib/libTopPSamplingLayer.a ../../../lib/libBaseSamplingLayer.a ../../../lib/libsampling_penalty_kernels.a ../../../lib/libsampling_topk_kernels.a ../../../lib/libsampling_topp_kernels.a ../../../lib/libban_bad_words.a ../../../lib/libstop_criteria.a ../../../lib/libactivation_kernels.a ../../../lib/libdecoder_masked_multihead_attention.a ../../../lib/libdecoder_multihead_attention.a ../../../lib/libbert_preprocess_kernels.a ../../../lib/libdecoding_kernels.a ../../../lib/libunfused_attention_kernels.a ../../../lib/libcustom_ar_comm.a ../../../lib/libcustom_ar_kernels.a ../../../lib/libgpt_kernels.a ../../../lib/libmemory_utils.a ../../../lib/libtensor.a ../../../lib/libcuda_utils.a -lmpi ../../../lib/liblogger.a -lcudart -ldl 
