Tool/software:
I have installed Python 3.5.2 on the 66AK2E05 and successfully installed basic Python libraries such as pysqlite3 and numpy. It is currently running on the company’s production line. To improve convenience and computational efficiency, we need to install the scipy library on Arm-Linux. To this end, I have cross-compiled the following software on PC-Linux, which successfully runs on Arm-Linux:
- GCC 4.7.4 with gfortran
- OpenBlas0.3.22
- Python 3.7.5
- Numpy1.18.5 compiled with OpenBlas
- Scipy1.3.3
Unfortunately, when using Scipy, importing OpenBlas results in the following error. I have consulted various sources, and most indicate that the version of Numpy is insufficient (I cannot verify this). However, higher versions of Numpy require GLIBC 2.16 and above, while the Arm-Linux currently supports only up to GLIBC 2.15. Upgrading GLIBC can easily lead to system instability.
root@k2hk-evm:/mnt/nfs_client_share# python3.7Python 3.7.5 (default, Sep 25 2024, 08:31:08)[GCC 4.7.4] on linuxType "help", "copyright", "credits" or "license" for more information.>>> from scipy.linalg import solveTraceback (most recent call last): File "<stdin>", line 1, in <module> File "/mnt/nfs_client_share/install-arm/lib/python3.7/site-packages/scipy/linalg/__init__.py", line 195, in <module> from .misc import * File "/mnt/nfs_client_share/install-arm/lib/python3.7/site-packages/scipy/linalg/misc.py", line 5, in <module> from .blas import get_blas_funcs File "/mnt/nfs_client_share/install-arm/lib/python3.7/site-packages/scipy/linalg/blas.py", line 214, in <module> from scipy.linalg import _fblasImportError: /mnt/nfs_client_share/install-arm/lib/python3.7/site-packages/scipy/linalg/_fblas.cpython-37m-arm-linux-gnueabihf.so: undefined symbol: npy_PyErr_ChainExceptionsCause
I would like support for the following two issues:
- The highest supported version of glibc in the Linux of 66AK2E05 is 2.15. Is there a suitable solution for a stable upgrade to a higher version?
- A solution for using Python and successfully running Scipy calculations on 66AK2E05 under any conditions.