Profiling Your First CUDA App with GPUFlight Trace
A practical walkthrough of using gpufl trace to capture a CUDA application and upload the result to the GPUFlight dashboard.
Archive
Articles on GPU instrumentation, CUDA and ROCm profiling, and practical performance work.
A practical walkthrough of using gpufl trace to capture a CUDA application and upload the result to the GPUFlight dashboard.
GPUFlight adds AMD GPU profiling with ROCm, HIP kernel tracing, telemetry, occupancy analysis, ISA disassembly, and hardware metrics.