| {% set version = "0.5.dev" %} |
| |
| package: |
| name: tvm-libs |
| version: {{ version }} |
| |
| source: |
| path: ../.. |
| |
| build: |
| number: 1 |
| string: cuda{{ cuda_version }}_{{ PKG_BUILDNUM }} # [cuda] |
| |
| requirements: |
| build: |
| - {{ compiler('cxx') }} # [linux] |
| - llvmdev ==6.0.0 # [osx] |
| host: |
| # The OS X build will require some manual setup or it will break |
| # See https://conda.io/docs/user-guide/tasks/build-packages/compiler-tools.html#macos-sdk |
| # It is also ass-backward because of llvm brokeness when mixed with the |
| # conda OS X compiler |
| - {{ compiler('cxx') }} # [osx] |
| - cmake |
| - llvmdev ==6.0.0 # [linux] |
| - zlib # [linux] |
| run: |
| - {{ pin_compatible('cudatoolkit', lower_bound=cuda_version, max_pin='x.x') }} # [cuda] |
| |
| about: |
| home: https://github.com/dmlc/tvm |
| license: Apache2 |
| summary: a low level domain specific language for compiling tensor computation pipelines |
