Skip to content

Latest commit

 

History

51 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Logismos

λογισμός - reckoning, calculation. The step-by-step numerical mode of reasoning.

GPU inference stack for transformer embedding models, built ground-up on HIP + hipBLASLt, targeting AMD gfx1100 (W7900).

Status: Phases 0-3 complete - Stella 1.5B v5 runs end-to-end on CPU with golden-fixture parity. Phase 4 (GPU cutover) is blocked on hardware availability: the AMD W7900 host is down for recovery, so GPU paths are unverified until it returns.

Why

Kanon's knowledge substrate (mnemosyne) needs a GPU-accelerated embedder. Candle has no ROCm backend. AMD deprecated ONNX Runtime's ROCm support. Rolling our own keeps the hardware boundary owned in-repo. One transformer family, one GPU family, written from the device upward.

Scope

  • In: tensor ops on HIP. Transformer inference. Stella 1.5B end-to-end. The EmbeddingModel contract mnemosyne consumes.
  • Out: training. Non-AMD GPUs. Runtime graph optimization. Multi-GPU.

Build configuration

crates/kernels/build.rs compiles HIP kernel sources via hipcc. When hipcc is absent from PATH this falls back automatically to a CPU-only, logismos_no_gpu_kernels-cfg'd build (this is the normal path on a GH-hosted CI runner and any box without ROCm).

On a box that has ROCm installed but where a HIP kernel compile needs to be skipped anyway - bisecting a hipcc regression, working around a broken local ROCm install, or a fast CPU-only iteration loop - set LOGISMOS_SKIP_HIP_BUILD to force the same CPU-only fallback the hipcc-absent path takes:

LOGISMOS_SKIP_HIP_BUILD=1 cargo build

Cost: the resulting build has no GPU kernels - kernels compiles against its CPU reference path only, and any HIP-backed op returns Error::NoGpuBuild at runtime instead of running on-device.

crates/hipcore/build.rs is different: it resolves the HIP runtime header and links amdhip64 unconditionally, with no LOGISMOS_SKIP_HIP_BUILD equivalent. cargo check --workspace therefore fails at hipcore on any box without ROCm headers installed — including a fresh developer machine. That failure is expected, not a defect (#14); do not look for a local workaround.

It does not block CI or merging. gate-attestation (.github/workflows/gate-attestation.yml, via forkwright/.github's hybrid-gate.yml) installs Ubuntu's libamdhip64-dev on the GH-hosted runner before building. That universe-component package ships the two headers hipcore's wrapper includes, plus libamdhip64.so. A PR with no local Gate-Passed trailer therefore still gets a real cargo check/clippy/nextest pass across the whole workspace, hipcore included. On a non-ROCm host, push without a trailer and let that CI path attest the change.

What it does not prove: the GH-hosted runner has no AMD GPU. This path proves the workspace compiles and links against real HIP headers/ABI — it never executes a HIP kernel. A change touching .hip sources, or hipcore/kernels FFI surface, still needs verification on real hardware (the forge-primary gate on menos, ROCm 6.4, per .kanon-ci.toml) before anyone can trust it at runtime.

Layout

Planning canonical lives in kanon, under projects/logismos/; see CLAUDE.md for how to resolve the kanon checkout root on this box:

  • projects/logismos/vision.md - what this is and what it is not.
  • projects/logismos/ROADMAP.md - phased plan.
  • projects/logismos/STATE.md - current state.
  • projects/logismos/gnomon.md - naming discipline inherited from the ecosystem.
  • projects/logismos/research/ - research dossiers.
  • projects/logismos/phases/NN-*/PLAN.md - per-phase implementation specs.

Repo-local:

  • CLAUDE.md - working instructions for AI assistants.
  • AGENTS.md - cross-tool bootstrap.
  • crates/ - the workspace.
  • phases/03-stella/golden/ - runtime test fixtures for Phase 3 parity test + Stella throughput bench.

License

PolyForm Shield 1.0.0 with AI-training prohibition. See LICENSE.

Repository Metadata

  • Registry name: logismos
  • Description: Kanon-managed forkwright repository logismos.
  • Repository identity: forkwright/logismos
  • Hosting: github
  • Push authority: GitHub-primary - push and PR through GitHub
  • Kanon prefix: lo
  • Config source: workflow/kanon.toml [projects.logismos]
  • Planning state: projects/logismos/STATE.md
  • Last state update: 2026-08-03

Run kanon docs sync --check --repo logismos to verify this generated section and kanon docs sync --apply --repo logismos to refresh it.

Blast zone

  • Paths explicitly named by the rendered prompt, role, or template input.

Acceptance verifier

kanon gate

About

GPU inference stack in Rust + HIP for AMD hardware - transformer embedding models; GPU cutover pending hardware access

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages