opened 06:10AM - 25 Oct 21 UTC
closed 05:32PM - 10 Nov 21 UTC
stat:awaiting response
type:build/install
stale
subtype: ubuntu/linux
2.6.0
**System information**
- OS Platform: `Ubuntu 20.04.3 LTS`
- TensorFlow instal… led from (source or binary): `source`
- TensorFlow version: 487ce666eba608e3ac909db94c4211c2661f2828
- Python version: `Python 3.8.10`
- Bazel version: `bazel 3.7.2`
- GCC/Compiler version (if compiling from source): `gcc (Ubuntu 9.3.0-17ubuntu1~20.04) 9.3.0`
- CUDA/cuDNN version: `NaN`
- GPU model and memory: `NaN`
**Problem**
When I try installing `tensorflow` from source the build fails with a message saying `'error: 'npy_hash_t' does not name a type; did you mean 'npy_half'?'` which is the result of the error `'error: 'Py_hash_t' does not name a type'`
**Command that I used to build tensorflow from source**
```shell
$ bazel --host_jvm_args=-Xms512m build --local_ram_resources 2048 --jobs=2 //tensorflow/tools/pip_package:build_pip_package
```
**Other info**
There's a strange behaviour that I would like to inform you, when I examine the messages from the kernel ring buffer I found the `Out of memory` issue which is strange for me because I'm limiting both the `memory` and the `number of jobs` while building `tensorflow` from source.
I need your help. I can’t build tensorflow from source, gcc compiler is giving me the following error:
'error: 'npy_hash_t' does not name a type; did you mean 'npy_half'?'
And I don’t understand why… Am I missing something?
Hi @joshiayush
Welcome to the TensorFlow Forum!
This error shows up when the numpy version is not updated or compatible with the existing TensorFlow version in your system. Please try again by upgrading the numpy version atleast >=1.19.0 using
!pip install --upgrade numpy
Let us know if the issue still persists. Thank you.