Closed KomputerMaster64 closed 5 months ago
Let's focus on this problem here, not elsewhere.
I think we need to look at something you've removed in the output pasted above. "ERROR: Could not build wheels for pytorch3d, which is required to install pyproject.toml-based projects" is just a generic error from a build failure. It isn't the specific part. The important, specific part ends with
or, T=c10::impl::ScalarTypeToCPPTypeT<c10::ScalarType::Float>, minN=1L, maxN=8L, curN=1L, minM=1L, maxM=4L, curM=1L, Args=<size_t, size_t, c10::impl::ScalarTypeToCPPTypeT<c10::ScalarType::Float> *, c10::impl::ScalarTypeToCPPTypeT<c10::ScalarType::Float> *, int64_t *, int64_t *, c10::impl::ScalarTypeToCPPTypeT<c10::ScalarType::Float> *, int64_t *, int64_t, int64_t, int64_t, int>]"
(345): here
instantiation of "void DispatchKernel2D<Kernel,T,minN,maxN,minM,maxM,Args...>(int64_t, int64_t, Args...) [with Kernel=KNearestNeighborKernelV3Functor, T=c10::impl::ScalarTypeToCPPTypeT<c10::ScalarType::Float>, minN=1L, maxN=8L, minM=1L, maxM=4L, Args=<size_t, size_t, c10::impl::ScalarTypeToCPPTypeT<c10::ScalarType::Float> *, c10::impl::ScalarTypeToCPPTypeT<c10::ScalarType::Float> *, int64_t *, int64_t *, c10::impl::ScalarTypeToCPPTypeT<c10::ScalarType::Float> *, int64_t *, int64_t, int64_t, int64_t, int>]"
/tmp/pip-req-build-a42lrpsm/pytorch3d/csrc/knn/knn.cu(433): here
ninja: build stopped: subcommand failed.
We don't know if the beginning of this was an error or a warning.
I see two possibilities here. (1) There was a compiler error, perhaps ending with the nearest neighbor stuff here, or perhaps coming from somewhere else, which caused the failure. (2) The parallel compilation ran out of memory and died. If you pasted everything in the output (here or using pastebin or similar) we could tell.
You could also try building with MAX_JOBS=1 set in the environment, which would usually avoid any running out of memory.
Thank you for the response. I will go ovre the two suggestions (compiler error with nearest neighbour; MAX_JOBS=1).
__ I think that the official PyTorch3D guide suggests using Python 3.8, 3.9, and 3.10 here. I will try with Python version 3.9 (currently using Python 3.8.19)
__
__
I have also attached both the stderr and stdout (output) in a file foundationpose_log_2024.04.02_01.txt
as a Google Drive Link (File size too big: 25 MB are allowed, 37 MB were attempted to upload.)
There's no compilation error in that log: you have an out-of-memory problem. And "Python 3.8" is meant to include 3.8.19 etc.. There's no important difference between python versions for this - switching to 3.9 won't change anything.
Thank you for the resposne. I was able to do the set-up foundationpose.
@KomputerMaster64 what do you mean by foundationpose? What changes did you do?
The suggested methods in Issue #1734 did not resolve my error i.e. installing
nvcc
andg++
did NOT result in the successful installation of PyTorch3D.(Thank you for sharing the issue #1734 and the possible resolution)
I am facing similar issue while installing PyTorch3D using the stable and the default build, using the command
pip install "git+https://github.com/facebookresearch/pytorch3d.git@stable"
andpip install "git+https://github.com/facebookresearch/pytorch3d.git"
I am using an NVIDIA RTX A4500 and an Intel Xeon Silver CPU (details about the hardware have been mentioned at the very end)
I am getting the main error as:
Note: Given below are the outputs for PyTorch3D installation command, and the following commands:
conda info
,conda list
,g++ --version
,nvcc --version
,lscpu
, andnvidia-smi
.Here is the output for the command
pip install "git+https://github.com/facebookresearch/pytorch3d.git@stable"
:Note: The output was too long with recurrent errors so I have added the last few lines of the output
Here is the output for the command
conda info
:Here is the output for the command
conda list
:Here is the output for the command
g++ --version
:Here is the output for the command
nvcc --version
:Extra Information about the CPU and the GPU
Intel CPU Information
NVIDIA GPU Information