Thank you for great and speedy answer. I should've probably searched for the issue a bit, apologies for being lazy
Search Criteria
Package Base Details: tensorrt
Package Actions
| Git Clone URL: | https://aur.archlinux.org/tensorrt.git (read-only, click to copy) |
|---|---|
| Keywords: | ai artificial intelligence nvidia |
| Submitter: | dbermond |
| Maintainer: | dbermond |
| Last Packager: | dbermond |
| Votes: | 15 |
| Popularity: | 0.088947 |
| First Submitted: | 2018-07-29 16:17 (UTC) |
| Last Updated: | 2024-04-30 19:32 (UTC) |
Packages (2)
FuzzyAtish commented on 2024-05-24 19:41 (UTC)
dbermond commented on 2024-05-24 19:28 (UTC)
@FuzzyAtish TLDR: it fails to build against cuda 12.5, downgrade cuda to 12.4.1 and it will work. The long story: there are two issues regarding this. Firstly, tensorrt 10.0.1 apparently does not support cuda 12.5, as you can see in the upstream documentation. Secondly, cuda 12.5 is not supported by the current version of the nvidia drivers that are available in the official repositories (cuda 12.5 was pushed without a driver that supports it), and you can read more details about this in the Arch Linux cuda package issues page at this link.
FuzzyAtish commented on 2024-05-24 17:52 (UTC)
I'm getting the following error when building
#17 2311.3 [ 36%] Building CUDA object
plugin/CMakeFiles/nvinfer_plugin.dir/roiAlignPlugin/roiAlignKernel.cu.o
#17 2320.2 home/builder/.cache/yay/tensorrt/src/TensorRT/plugin/common/common.cuh(321):
error: identifier "FLT_MAX" is undefined
#17 2320.2 float threadData(-FLT_MAX);
#17 2320.2 ^
#17 2320.2
#17 2320.2 home/builder/.cache/yay/tensorrt/src/TensorRT/plugin/common/common.cuh(373):
error: identifier "FLT_MAX" is undefined
#17 2320.2 float threadData(-FLT_MAX);
#17 2320.2 ^
#17 2320.2
#17 2320.3 2 errors detected in the compilation of "/home/builder/.cache/yay/tensorrt/src/TensorRT/plugin/roiAlignPlugin/roiAlignKernel.cu".
dbermond commented on 2024-04-04 18:45 (UTC)
@tbb @qsthy this is an issue with the python-onnx package, as it needs a rebuild against latest protobuf. The latest protobuf version did not have a soname version bump, but it seems to be ABI incompatible, thus needing a rebuild of some dependent packages. I've talked to the maintainer, and a fixed version is currently in [extra-testing]. Either wait for it to leave [extra-testing], or enable this repository. I've tested, and it's building fine with it.
qsthy commented on 2024-03-28 13:26 (UTC)
Concur, same build failure happens here. Looking it up on the Web it appears to be an issue with onnx and/or the the onnx/protobuf combination, see for example https://github.com/onnx/onnx/issues/4704
In case it matters I already have on my machine extra/protobuf and aur/anaconda installed.
tbb commented on 2024-03-26 05:51 (UTC) (edited on 2024-03-26 05:52 (UTC) by tbb)
can't build [code] ImportError: /usr/lib/python3.11/site- packages/onnx/onnx_cpp2py_export.cpython-311-x86_64-linux-gnu.so: undefined symbol: _ZN6google8protobuf8internal17AssignDescriptorsEPFPKNS1_15DescriptorTableEvEPN4absl12lts_202308029once_flagERKNS0_8MetadataE [/code]
dbermond commented on 2024-01-18 02:46 (UTC)
@gbin This warning is not related to the package itself. It means that the underlying makepkg command invoked by the AUR helper is not doing a clean build. Again, AUR helpers are not supported, use makepkg.
gbin commented on 2024-01-10 15:52 (UTC) (edited on 2024-01-10 15:53 (UTC) by gbin)
@dbermond Interestingly, something changed and it started to work with paru. Thank you for maintaining the package. It gives out a warning during install maybe the flakiness is related to it?
==> Checking runtime dependencies...
==> Checking buildtime dependencies...
==> WARNING: Using existing $srcdir/ tree
==> Starting pkgver()...
==> Starting build()...
dbermond commented on 2024-01-04 20:13 (UTC)
@gbin I've just checked, and the package is building fine. You are using an AUR helper, and this is not supported. Use makepkg.
gbin commented on 2024-01-04 03:02 (UTC)
It compiles but not install correctly for me:
-- Installing: /home/gbin/.cache/paru/clone/tensorrt/pkg/tensorrt/usr/lib/libnvonnxparser_static.a
==> Tidying install...
-> Removing libtool files...
-> Purging unwanted files...
-> Removing static library files...
-> Stripping unneeded symbols from binaries and libraries...
-> Compressing man and info pages...
==> Checking for packaging issues...
==> WARNING: Package contains reference to $srcdir
usr/lib/libnvinfer_plugin.so.9.2.0
usr/lib/libnvinfer_vc_plugin.so.9.2.0
usr/lib/libnvonnxparser.so.9.2.0
usr/lib/libnvonnxparser_static.a
usr/lib/libnvinfer_plugin_static.a
==> Creating package "tensorrt"...
-> Generating .PKGINFO file...
-> Generating .BUILDINFO file...
-> Generating .MTREE file...
-> Compressing package...
==> Starting package_python-tensorrt()...
/home/gbin/.cache/paru/clone/tensorrt/PKGBUILD: line 152: cd: /home/gbin/.cache/paru/clone/tensorrt/src/TensorRT/tools/onnx-graphsurgeon/dist: No such file or directory
==> ERROR: A failure occurred in package_python-tensorrt().
Aborting...
error: failed to build 'tensorrt-9.2.0.5-1':
error: packages failed to build: tensorrt-9.2.0.5-1
Pinned Comments
dbermond commented on 2024-05-24 19:28 (UTC)
@FuzzyAtish TLDR: it fails to build against cuda 12.5, downgrade cuda to 12.4.1 and it will work. The long story: there are two issues regarding this. Firstly, tensorrt 10.0.1 apparently does not support cuda 12.5, as you can see in the upstream documentation. Secondly, cuda 12.5 is not supported by the current version of the nvidia drivers that are available in the official repositories (cuda 12.5 was pushed without a driver that supports it), and you can read more details about this in the Arch Linux cuda package issues page at this link.