| 12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667 |
- # Authoritative build inputs — GLM-5.3-Flash r12 (jovian-judgement) clean build.
- # Every entry is a tag or exact version — no :latest, no main, no ranges.
- # Pattern: solar-open2-nvfp4 (raw CUDA base, staged wheel builds, ccache/uv
- # buildx caches) — content r12-specific per the rtx6kpro source contract.
- # CUDA base (aarch64; digest-pinned, tested by solar-open2-nvfp4)
- CUDA_BASE_IMAGE=nvidia/cuda:13.2.0-devel-ubuntu24.04
- CUDA_BASE_DIGEST=sha256:a5b6256e470196fc1d5f8f62139d57d3662867746dfe1cb352d7652024047020
- # Python / uv
- UV_VERSION=0.11.28
- PYPI_INDEX_URL=https://pypi.org/simple
- PYTORCH_INDEX_URL=https://download.pytorch.org/whl/cu130
- # Torch stack — the fork's requirements/cuda.txt pins (r12 era, cu130 aarch64)
- TORCH_VERSION=2.13.0
- TORCHVISION_VERSION=0.28.0
- TORCHAUDIO_VERSION=2.11.0
- # FlashInfer — the fork's cuda.txt uses flashinfer-python==0.6.17 (PyPI);
- # flashinfer-cubin is NOT a build dep (setup.py skips it; the runtime
- # fetches cubins when the package is absent).
- FLASHINFER_VERSION=0.6.17
- # vLLM — local-inference-lab fork, dev/jovian-judgement base + PR heads.
- #
- # r18 composition (2026-09-05): branch tip f564dffe9a66 — supersedes the r17
- # base 9c4dd054 (which itself superseded r12's 54f6e9826c + PR set 550-557;
- # do NOT re-merge 550-557, they conflict). The r17 PR set 579/580/581/586 is
- # now merged into the tip; only 582 is still open, pinned by head SHA.
- # New in this window (9c4dd054..f564dffe + b12x below): RoCEnante one-shot
- # RoCE collectives (PR #597; env VLLM_ENABLE_ROCE_ALLREDUCE, off by default),
- # B12X NVFP4 KV cache (fp8_ds_mla / nvfp4_ds_mla), DFlash split-KV paged
- # decode attention, L2 weight-prefetch decode, DSpark input-budget split,
- # spec-decode KV-connector + prefix-cache-block fixes, GLM tool-delimiter
- # parser hardening, prefill fairness engines, KDA prefill backend,
- # FlashKDA stable-inverse fix. No requirements/setup.py drift vs r17.
- # Staged tree on the build host: sm121/r18 @ 725e36d3189cc7cbd8c27a16323c4b9643fd8cf2
- # (base + pr582 merge + staged-cutlass cmake patch; see /tmp/vllm-src.tar.gz).
- VLLM_REPO=https://github.com/local-inference-lab/vllm.git
- VLLM_BASE_REF=f564dffe9a6625ae3d87706530ce35974f1d2989
- VLLM_PR_HEADS="582:4df4cf5e123b55aab2fbcb310468b076995e9410"
- # B12X — master + PR heads (pure-python wheel; CuTe-DSL JIT kernels).
- # r18: master tip 613cfd9b (supersedes r17 base 2bbab479; PR 290 bf16 PCIe
- # two-shot now merged into it). Still open: 284 (SM120 mhc decode split),
- # pinned by head SHA. New in window: RoCEnante one-shot RoCE all-reduce/
- # all-gather (#295, both-rail striping, DSV4 graph collectives), GLM_NEXT
- # NVFP4 KV cache (PR #309), KDA prefill extension (full op surface),
- # dense-GEMM/MoE A16 activation-precision selection on SM121/GB10 (relevant
- # to the linear-b12x quality toxin — re-bisect with this build), QSA large-
- # prefill + paged-CuTe specialization, DSA indexer plan-owned lifecycle.
- # Staged tree on the build host: sm121/r18 @ c594d447d3129e717f8aa0a7c56d7b777912bec3
- # (base + pr284 merge; see /tmp/b12x-src.tar.gz).
- B12X_REPO=https://github.com/local-inference-lab/b12x.git
- B12X_BASE_REF=613cfd9bbc542e2a7a9e5e216d1f62a8becda87f
- B12X_PRS="284"
- # DeepGEMM — fetched by vllm's cmake; pre-cloned (raw-SHA fetches of a
- # non-default-branch tip 404 over GitHub's git protocol)
- DEEPGEMM_REPO=https://github.com/deepseek-ai/DeepGEMM.git
- DEEPGEMM_BRANCH=nv_dev
- DEEPGEMM_REF=8b1392b978f5a03c828dd1711090d7fb50958b8a
- # Target GPU architecture (DGX Spark = sm_121a)
- TORCH_CUDA_ARCH_LIST=12.1a
- BUILD_JOBS=4
|
