Difference between revisions of "Eosc/easybuild/andreku"

From Nordic Language Processing Laboratory
Jump to: navigation, search
(To use:)
(Remaining issues)
Line 29: Line 29:
  
 
= Remaining issues =
 
= Remaining issues =
 +
*NVIDIA Bert implementation packaged as a module
 
*In order to keep the modules NLPL-branded, [https://source.coderefinery.org/nlpl/easybuild/-/issues/4#note_13895 environment variables must be added manually to the module files]. Without that, modules load fine, but cannot be used as dependencies in building other modules.
 
*In order to keep the modules NLPL-branded, [https://source.coderefinery.org/nlpl/easybuild/-/issues/4#note_13895 environment variables must be added manually to the module files]. Without that, modules load fine, but cannot be used as dependencies in building other modules.
 
*TensorFlow is built with CUDA 10.1.243, not CUDA 10.0.130. Attempts to use the latter [https://source.coderefinery.org/nlpl/easybuild/-/issues/9 failed]. Should find a way to make EasyBuild look for a non-standard CUDA location.
 
*TensorFlow is built with CUDA 10.1.243, not CUDA 10.0.130. Attempts to use the latter [https://source.coderefinery.org/nlpl/easybuild/-/issues/9 failed]. Should find a way to make EasyBuild look for a non-standard CUDA location.
 
* Check whether using '''gompi''' instead of '''gompic''' (with CUDA) [https://source.coderefinery.org/nlpl/easybuild/-/issues/9#note_13894 leads to problems with multi-node training]. Multi-GPU training on a single node is confirmed to work.
 
* Check whether using '''gompi''' instead of '''gompic''' (with CUDA) [https://source.coderefinery.org/nlpl/easybuild/-/issues/9#note_13894 leads to problems with multi-node training]. Multi-GPU training on a single node is confirmed to work.

Revision as of 17:54, 23 November 2020

Important stuff to remember

export EB_PYTHON=python3

module load EasyBuild/4.3.0

Playground on Saga: /cluster/shared/nlpl/software/easybuild_ak

export EASYBUILD_ROBOT_PATHS=/cluster/software/EasyBuild/4.3.0/easybuild/easyconfigs:/cluster/shared/nlpl/software/easybuild_ak

(or just source PATH.local)

Repository: https://source.coderefinery.org/nlpl/easybuild/-/tree/ak-dev

Status

03/11/2020: successfully built cython-0.29.21-foss-2019b-Python-3.7.4, numpy-1.18.1-foss-2019b-Python-3.7.4, SciPy-bundle-2020.03-foss-2019b-Python-3.7.4, Bazel-0.26.1-foss-2019b, h5py-2.10.0-foss-2019b-Python-3.7.4.

04/11/2020: TensorFlow 1.15.2 successfully built and installed, using CUDA 10.1.243

19/11/2020: gomkl toolchain built with Intel MKL 2019.1.144

21/11/2020: successfully built everything (including TensorFlow 1.15.2) with the gomkl toolchain.

To use:

module use -a /cluster/shared/nlpl/software/easybuild_ak/easybuild/install/modules/all/

module load NLPL-TensorFlow/1.15.2-gomkl-2019b-Python-3.7.4

Remaining issues

  • NVIDIA Bert implementation packaged as a module
  • In order to keep the modules NLPL-branded, environment variables must be added manually to the module files. Without that, modules load fine, but cannot be used as dependencies in building other modules.
  • TensorFlow is built with CUDA 10.1.243, not CUDA 10.0.130. Attempts to use the latter failed. Should find a way to make EasyBuild look for a non-standard CUDA location.
  • Check whether using gompi instead of gompic (with CUDA) leads to problems with multi-node training. Multi-GPU training on a single node is confirmed to work.