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 skills add AMD-AGI/Apex --skill gpu-architecture-fundamentalsgit 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/gpu-architecture-fundamentals)<a href="https://agentmods.dev/skills/amd-agi/apex/gpu-architecture-fundamentals"><img src="https://agentmods.dev/badge/skills/amd-agi/apex/gpu-architecture-fundamentals/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/amd-agi/apex/gpu-architecture-fundamentals"><img src="https://agentmods.dev/badge/skills/amd-agi/apex/gpu-architecture-fundamentals.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 5.1 | $0.00051 | $0.00602 |
| Opus 5 | $0.00026 | $0.00301 |
| Sonnet 5 | $0.00010 | $0.00120 |
| Haiku 4.5 | $0.00005 | $0.00060 |
Grade A, and why
gpu-architecture-fundamentals 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 9d 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 — 37 lines — stays where its author put it; the contents beside it link to each section on GitHub.
GPU Architecture Fundamentals
Purpose
- Reference core GPU concepts (memory hierarchy, execution model) and typical bandwidth/latency numbers to ground optimization choices.
- Provide block size heuristics and ready-to-use checklists before writing or tuning kernels.
- Map common optimization patterns across HIP, Triton, and PyTorch to pick framework-specific tactics quickly.
When to Use
- Planning or reviewing kernel designs where occupancy, memory bandwidth, or latency hiding are concerns.
- Selecting grid/block shapes, deciding on shared memory usage, or checking for coalesced accesses.
- Comparing optimization levers across frameworks when porting kernels.
How to Use
- Recall memory hierarchy: prefer registers > shared/L1 > L2 > HBM; treat HBM as ~400–800 cycle latency, registers ~0, shared ~20–30 cycles.
- Anchor bandwidth sense-checks with table values (e.g., MI300X HBM3 ~5.3 TB/s, A100 HBM2e ~2.0 TB/s).
- Choose block sizes by operation: element-wise 256–1024 threads, reduction 256–512, matmul tiles 128x128 or 256x128, conv 32x32 or 64x64.
- Apply execution model mapping: thread ↔ element/partial tile, warp/wavefront ↔ contiguous data segments, block/workgroup ↔ tiles sharing shared memory, grid ↔ full problem coverage.
- Run the optimization checklist before finalizing kernels:
- Ensure coalesced and vectorized memory access; avoid shared memory bank conflicts.
- Target occupancy >50%; watch register pressure and shared memory usage to avoid spilling.
- Fuse operations where possible; leverage mixed precision when valid.
- Overlap transfers with compute; tune block/grid dimensions; unroll small loops.
- Use pattern summaries to pick tactics per framework:
- Memory: HIP manual strides/shared, Triton
tl.arange/implicit tiling, PyTorch.contiguous()/compiler. - Compute: HIP manual fusion/unroll, Triton
@triton.jit+tl.constexpr, PyTorchtorch.compile/FlashAttention. - Parallelism: HIP block/grid + occupancy APIs, Triton autotune + constexpr block sizes, PyTorch compiler/automatic launch config.
- Memory: HIP manual strides/shared, Triton
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.
- 9d ago First seen · 37 lines · 51 tokens per session scan A 7c65b3b67ec2
gpu-architecture-fundamentals is a skill published in the GitHub repository AMD-AGI/Apex (76 stars, last pushed 6d ago), licensed MIT. It adds 51 tokens to every session and 602 once invoked, about $0.0003 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
html-ppt-zhangzara-pin-and-paper
A field-biology capstone on urban pollinator decline — the survey design, the data, the contribution, and the caveats. Built as a decision-grade coursework defense deck for faculty reviewers.
proof-writer
A tool for writing or checking rigorous mathematical proofs for claims such as theorems, lemmas, propositions, and corollaries.
tooluniverse-organic-chemistry
Organic chemistry reasoning guide for reaction product prediction, mechanism analysis (electrophilic/nucleophilic substitution, addition, elimination, pericyclic, radical), and spectroscopy interpretation (1H/13C NMR, IR, MS). Reasons from first principles (electron flow, kinetic vs thermodynamic) rather than…
scaffold-exercises
Scaffold a graded problem set with sections, problems, worked solutions, and short "why this matters" explainers across analytical, empirical, and coding types. Use when user says "make a problem set on X", "scaffold exercises for this lecture", "create practice problems", "generate homework with a solution key"…
edu-chem-reaction
A guide for turning a chemical reaction into a self-contained interactive 3D web demonstration. It shows molecules changing while also displaying the balanced equation, explanations, atom counts, and optionally an energy curve.
present-paper
Academic presentation preparation — paper-driven (journal club, grand rounds, seminar) and lecture/teaching decks (course material, workshop slides, conference talks). Analyzes source material, finds supporting references, drafts audience-adapted speaker scripts, generates or augments PPTX with speaker notes, and…