Module benchmarks 

Performance benchmark infrastructure for Rusty-Stack. Heavy lifting is delegated to a small embedded Python helper that runs ROCm-enabled PyTorch kernels to exercise the real hardware.

Structs

BenchmarkResult

Functions

run_deepspeed_benchmark

run_flash_attention_benchmark

run_flash_attention_ck_benchmark

run_gemm_benchmark

run_gpu_capability_benchmark

run_llama_cpp_benchmark

run_megatron_benchmark

run_memory_bandwidth_benchmark

run_onnx_benchmark

run_pytorch_benchmark

run_tensor_core_benchmark

run_vllm_benchmark