Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/amd-agi/apex/rocprof-computenpx skills add AMD-AGI/Apex --skill rocprof-computegit clone --depth 1 https://github.com/AMD-AGI/ApexWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/amd-agi/apex/rocprof-compute)<a href="https://agentmods.dev/skills/amd-agi/apex/rocprof-compute"><img src="https://agentmods.dev/badge/skills/amd-agi/apex/rocprof-compute.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00037 | $0.00657 |
| Opus 5 | $0.00018 | $0.00329 |
| Sonnet 5 | $0.00007 | $0.00131 |
| Haiku 4.5 | $0.00004 | $0.00066 |
Grade A, and why
rocprof-compute scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 40 lines — stays where its author put it; the contents beside it link to each section on GitHub.
rocprof-compute Profiling
Purpose
- Capture AMD GPU kernel metrics, roofline data, and traces with rocprof-compute.
- Analyze collected workloads to identify bottlenecks (compute vs memory, cache/TCP/TCC/SQ utilization).
When to Use
- Need kernel-level performance diagnostics on AMD GPUs (MI200/MI300 family).
- Comparing different kernel implementations or launch configs.
- Triaging stalls/low occupancy indicated by runtime benchmarks.
How to Use
- Activate project venv and install rocprof-compute Python deps (once per environment):
source .venv/bin/activatepython -m pip install -r /opt/rocm-7.0.0/libexec/rocprofiler-compute/requirements.txt
- Profile a workload:
source .venv/bin/activate && rocprof-compute profile -n <name> --path <out_dir> --join-type kernel -b SQ -b TCP -b TCC -- <cmd> <args>- Example (paged attention ragged test):
rocprof-compute profile -n kernelgen --path rocprof_compute_profile --no-roof --join-type kernel -b SQ -b TCP -b TCC -- .venv/bin/python -O op_tests/test_pa_ragged.py -p Shomy -q none -c 128 - Prefer
--join-type kernelfor comparing same kernel across grids; switch togridif grid-sensitive. - Add/adjust
-bblocks to target specific hardware units; use--list-metrics <arch>if unsure. - Use
--no-roofto skip roofline if only counters are needed; remove it to gather roofline data.
- Analyze collected data:
rocprof-compute analyze --path <out_dir> -b <metric_ids>or--list-stats/--list-metrics <arch>to discover ids.- Example:
rocprof-compute analyze --path rocprof_compute_profile -b 2 - For interactive review, use
--gui(default port 8050 or--random-port) or--tui.
- Typical workflow checklist:
- Pick a short, reproducible workload and seed; pin
--name+--pathper experiment. - Collect counters (SQ/TCP/TCC) and optionally roofline in one run; avoid mixing many kernels in a single profile when isolating a hotspot.
- After analyze, inspect top stats, occupancy, LDS/HBM bandwidth, and hotspot kernels; rerun with filtered
--kernelor--dispatchif needed.
- Pick a short, reproducible workload and seed; pin
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 3d ago First seen · 40 lines · 37 tokens per session scan A 77d545769ae0
rocprof-compute is a skill published in the GitHub repository AMD-AGI/Apex (76 stars, last pushed 6d ago), licensed MIT. It adds 37 tokens to every session and 657 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
write-triton-kv-cache-append-kernel
Guide the agent through implementing a Triton kernel that writes newly computed K and V tensors into a pre-allocated KV cache during LLM inference. This covers two cache layouts (contiguous and paged / vLLM-style PagedAttention), unified prefill and decode handling via a slotmapping tensor, GQA/MQA where the cache…
write-triton-rope-kernel
Guide the agent through implementing a correct Triton kernel that applies Rotary Position Embeddings (RoPE) to query and key tensors before attention. This covers the two incompatible layout conventions (GPT-NeoX/HuggingFace-LLaMA vs GPT-J/original-paper), pre-computed cos/sin table consumption, per-token position…
choose-launch-configuration
Guide the agent through selecting the correct and efficient thread block dimensions and grid dimensions for a CUDA kernel, covering occupancy analysis, register and shared memory constraints, tail effects, persistent kernels, and when to use cudaOccupancyMaxActiveBlocksPerMultiprocessor as a decision tool.
optimize-shared-memory-tiling
Guide the agent through designing and tuning shared memory tiling strategies for CUDA kernels, covering bank conflict analysis and elimination, tile shape selection, double buffering with async copy, occupancy tradeoffs from shared memory allocation, and the decision of when smem tiling is worth the complexity.
write-cuda-layernorm-kernel
Guide the agent through designing and implementing a correct, efficient CUDA LayerNorm (and RMSNorm) kernel, covering mean/variance computation strategies, Welford online accumulation, epsilon placement, affine transform application, backward pass structure, and decomposition for non-power-of-two hidden dimensions.
write-cuda-reduction-kernel
Guide the agent through designing and implementing a correct, efficient CUDA reduction kernel for a given operator (sum, max, min, or custom binary associative op), covering warp-level primitives, block-level reduction, multi-block strategies, and when to use CUB instead.