MasterJH5574 commented on issue #20388: URL: https://github.com/apache/tvm/issues/20388#issuecomment-5745374785
Verified the v0.27.0.rc0 candidate two ways — **(A)** building from the signed source release, and **(B)** installing the published wheel from PyPI. Kept separate since they are different installation paths. ## A. Source build check — `tests/scripts/release/test_release_package.sh` Run with the script shipped in the release candidate, with the version header set to `v0.27.0` and `rc=rc0`: https://github.com/apache/tvm/blob/v0.27.0.rc0/tests/scripts/release/test_release_package.sh ```bash $ bash test_release_package.sh + version=v0.27.0 + rc=rc0 [setup] Create an isolated virtualenv ... + python3 -m venv .venv ... IGNORE ... [1/10] Downloading from apache.org ... ... IGNORE ... [2/10] Downloading from github.com ... ... IGNORE ... [3/10] Check difference between github.com and apache.org ... + diff github/md5sum.txt ./apache/md5sum.txt [4/10] Checking asc ... + gpg --verify ./apache-tvm-src-v0.27.0.rc0.tar.gz.asc ./apache-tvm-src-v0.27.0.rc0.tar.gz gpg: using RSA key 3EA0F45C3223CF0D81C8F2CCA9F8369106601185 gpg: Good signature from "Ruihang Lai <[email protected]>" [unknown] [5/10] Checking sha512 ... + sha512sum -c ./apache-tvm-src-v0.27.0.rc0.tar.gz.sha512 apache-tvm-src-v0.27.0.rc0.tar.gz: OK [6/10] Unzip ... + tar -zxf apache-tvm-src-v0.27.0.rc0.tar.gz [7/10] Checking whether binary in source code ... [8/10] Compile and Python Import on Linux ... + cmake .. + make -j16 ... IGNORE ... [100%] Built target tvm_objs [100%] Linking CXX shared library lib/libtvm_compiler.so [100%] Built target tvm_compiler [9/10] Install Python runtime dependencies (declared in the source pyproject.toml) ... Installing declared runtime deps: apache-tvm-ffi>=0.1.14.post0 ml_dtypes numpy typing_extensions Successfully installed apache-tvm-ffi-0.1.14.post0 ... [10/10] Import TVM and print path ... + python3 -c 'import tvm; print(tvm.__path__)' ['.../github/apache-tvm-src-v0.27.0/python/tvm'] ``` LGTM — md5 (github vs apache.org) identical, GPG signature good (verified against the dev KEYS), sha512 OK, no binaries in source, full compile, and `import tvm` succeeds. The default build has LLVM disabled; import emits warnings about parsing Arm target features without LLVM support. The source build now reports `tvm.__version__ == "0.27.0"`; serialized IR metadata also reports `0.27.0`. The focused IR checks passed (19 tests). ## B. Wheel install check — PyPI All four platform wheels published for `apache-tvm==0.27.0rc0` (macOS arm64, Linux x86_64 + aarch64 manylinux_2_28, Windows AMD64). Installed the Linux x86_64 wheel in a clean venv — dependencies (including `apache-tvm-ffi`) resolve automatically, no source build: ```bash $ python3 -m venv venv && . venv/bin/activate $ unset PYTHONPATH $ pip install --only-binary=:all: apache-tvm==0.27.0rc0 ... Successfully installed apache-tvm-0.27.0rc0 apache-tvm-ffi-0.1.14.post0 ... $ python -c "import tvm; print(tvm.__version__)" 0.27.0rc0 ``` Basic CUDA runtime checks also passed on both NVIDIA GeForce RTX 4090 GPUs: device detection, allocation, host-to-device, device-to-device and device-to-host copies, and synchronization for float32 and int32 tensors. TVM libraries were loaded from the fresh venv, with PYTHONPATH and LD_LIBRARY_PATH unset. LGTM — wheel installs with the declared dependencies (`apache-tvm-ffi>=0.1.14.post0`) and imports at the expected version. (It is a pre-release, so a bare `pip install apache-tvm` still resolves the latest *stable*; testers opt into the RC by pinning `==0.27.0rc0` or using `--pre`.) -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
