Loading...
Loading...
Found 6 Skills
Collects and reads rocprof-compute roofline data to place an AMD GPU kernel against the hardware's peak compute and peak bandwidth ceilings. Use when the user asks whether a kernel is compute-bound or memory-bound, asks about arithmetic intensity or FLOPs per byte, asks how much performance is left on the table, or asks for a roofline. Not for CUDA tools, Windows, or per-instruction analysis.
Profiles an AMD GPU application with rocprof-compute and finds which kernel is slow and why, using performance metrics and PC sampling. Collects counters and selects kernels and dispatches, then analyzes GPU memory throughput, compute efficiency, wavefront occupancy and wavefront limits, scheduler stalls, etc. Use when the user asks to profile, benchmark, or speed up a HIP/ROCm kernel or application, or asks where a kernel's bottleneck is. Not for CUDA tools, Windows even when the user names rocprof-compute, system-wide CPU/MPI tracing, or writing kernel source.
Collects and analyzes rocprof-compute PC sampling to show which individual instructions in an AMD GPU kernel are hot and why wavefronts stall there, with ISA and source line attribution. Use when the user asks which instruction or source line is slow, asks about stall reasons, wants instruction-level or ISA-level detail, or wants hotspots inside a single kernel. Not for CUDA tools, Windows, or whole-application timing.
Analyzes the AMD GPU memory hierarchy with rocprof-compute, covering the Memory Chart, vL1D and L2 caches, LDS bank conflicts, Infinity Fabric traffic, and the experimental gfx950 memory bandwidth analysis. Use when the user asks about memory bandwidth, cache hit rates, coalescing, LDS conflicts, data locality, HBM traffic, or says a kernel is memory-bound. Not for CUDA tools, Nsight, Windows, host-memory leaks, GPU out-of-memory allocation failures, or system-wide tracing.
Attributes AMD GPU kernel performance counters back to the PyTorch or Triton operators that launched them, using rocprof-compute ML API tracing. Use when the user asks which PyTorch layer or operator is slow, wants to map GPU kernels to model code, profiles a training or inference script, asks about torch.compile and Inductor kernels, or wants to select user-defined ROCTx regions from a traced Python profile. Not for CUDA tools, Windows, compiled non-Python binaries, or starting and stopping a profile at ROCTx ranges.
Reads the rocprof-compute System Speed-of-Light panel to show how close an AMD GPU kernel runs to hardware peak for compute throughput, memory bandwidth, and utilization. Use when the user asks how close to peak a kernel is, what percent of peak it reaches, whether a measured percentage is good enough or normal, whether a utilization number should be higher, or asks for an overall efficiency summary. Not for CUDA tools, Windows, per-instruction analysis, or a datasheet or theoretical-peak question when there is no profile to read.