MCPcopy Create free account
hub / github.com/NVIDIA/TensorRT-LLM / get_sm_version

Function get_sm_version

tensorrt_llm/_utils.py:754–756  ·  view source on GitHub ↗
()

Source from the content-addressed store, hash-verified

752
753@lru_cache(maxsize=1)
754def get_sm_version():
755 prop = torch.cuda.get_device_properties(0)
756 return prop.major * 10 + prop.minor
757
758
759@lru_cache(maxsize=1)

Callers 15

allreduce_benchmarkFunction · 0.90
_woq_moe_groupwise_matmulMethod · 0.90
test_llama_verification_with_kv_cache_relocationMethod · 0.90
test_fused_moeFunction · 0.90
run_fused_moe_nvfp4Function · 0.90
__init__Method · 0.90
calculate_reference_output_generationFunction · 0.90
test_forward_sparse_mla_unifiedFunction · 0.90
_get_moe_config_for_blackwellFunction · 0.90
test_matmul_activation_int4_inputFunction · 0.90
test_w4a16_linearFunction · 0.90
test_w4a8_linearFunction · 0.90

Calls

no outgoing calls

Tested by 11

_woq_moe_groupwise_matmulMethod · 0.72
test_llama_verification_with_kv_cache_relocationMethod · 0.72
test_fused_moeFunction · 0.72
run_fused_moe_nvfp4Function · 0.72
__init__Method · 0.72
calculate_reference_output_generationFunction · 0.72
test_forward_sparse_mla_unifiedFunction · 0.72
_get_moe_config_for_blackwellFunction · 0.72
test_matmul_activation_int4_inputFunction · 0.72
test_w4a16_linearFunction · 0.72
test_w4a8_linearFunction · 0.72