vLLM cu129 aarch64 build — 0e3ac490

generated 2026-09-05 11:47 +07 · wheel directory · how to install · upstream commit

Build

wheelvllm-0.28.1rc1.dev323+cu129.20260903.g0e3ac490-cp38-abi3-manylinux_2_28_aarch64.whl
variantcu129 (CUDA 12.9.1, torch 2.13.0+cu129)
vLLM commit0e3ac4907d21e77cb4781338c49fef17bfea8f2b
build date20260903
build timenot recorded
ninja steps463
wheel size696.9 MB
peak memory? GiB cgroup / ? GiB anon
builder[email protected] (aarch64, 20x2 jobs, cpuset 0-19)
inspected object-
ELFELF 64-bit LSB shared object, ARM aarch64
cubinssm_100, sm_100a, sm_103, sm_103a, sm_121, sm_121a, sm_80, sm_89, sm_90, sm_90a
max glibc2.17
max libstdc++3.4.21

GPU coverage

Compiled cubins, no JIT needed. Verified present: sm_100, sm_100a, sm_103, sm_103a, sm_121, sm_121a, sm_80, sm_89, sm_90, sm_90a
GPUrequested archcubin
GH200 / H200 (Grace Hopper)9.0asm_90a
GB200 (Blackwell)10.0asm_100a
GB300 (Blackwell Ultra)10.3asm_103a
GB10 (DGX Spark)12.1asm_121a

Validation

PASSED

Structural checks only — neither machine has an aarch64 GPU. Functional proof comes from validate-gb10.py, which rents an NVIDIA GB10 and runs a real generation.

Build log (tail)

PEAK_ANON_BYTES 49018413056
BUILD_OK /home/build/vllm-nightly/dist/cu129/vllm-0.28.1rc1.dev323+g0e3ac4907.d20260902.cu129-cp38-abi3-linux_aarch64.whl
Warning: Permanently added '[192.168.1.141]:30422' (ED25519) to the list of known hosts.

VALIDATION {
  "wheel": "vllm-0.28.1rc1.dev323+g0e3ac4907.d20260902.cu129-cp38-abi3-linux_aarch64.whl",
  "checks": {
    "shared_objects": 13,
    "file": "ELF 64-bit LSB shared object, ARM aarch64",
    "cubin_archs": [
      "sm_100",
      "sm_100a",
      "sm_103",
      "sm_103a",
      "sm_121",
      "sm_121a",
      "sm_80",
      "sm_89",
      "sm_90",
      "sm_90a"
    ],
    "max_glibc": "2.17",
    "max_glibcxx": "3.4.21"
  }
}