CODEWITHSUNDEEP
pythongoogle-colaboratoryrapidscuml
Gozdi
Gozdi

Reputation: 61

RAPIDS, CUML on google colab

I'm installing RAPIDS on google colab pro but it takes a lot of time, last 2 instalations took over an hour instead of about 15 minutes as said during instalation "Starting the RAPIDS install on Colab. This will take about 15 minutes". Is there any way to speed this process up? I only want to use cuML library I don't care about others that are beeing installed with it.

For instalation I'm using following steps

!git clone https://github.com/rapidsai/rapidsai-csp-utils.git
!python rapidsai-csp-utils/colab/env-check.py

!bash rapidsai-csp-utils/colab/update_gcc.sh
import os
os._exit(00)

import condacolab
condacolab.install()

import condacolab
condacolab.check()

!python rapidsai-csp-utils/colab/install_rapids.py stable
import os
os.environ['NUMBAPRO_NVVM'] = '/usr/local/cuda/nvvm/lib64/libnvvm.so'
os.environ['NUMBAPRO_LIBDEVICE'] = '/usr/local/cuda/nvvm/libdevice/'
os.environ['CONDA_PREFIX'] = '/usr/local'

Upvotes: 1

Views: 3035

Answers (1)

TaureanDyerNV
TaureanDyerNV

Reputation: 1291

5/14/24 update:

cuDF and cudf.pandas now comes preinstalled on Google Colab's GPU instances. You can still pip install the rest of the RAPIDS libraries, such as cuML, cuGraph, cuXfilter, and cuSpatial, onto your GPU enabled Colab instance using the RAPIDS+Colab Pip Installation Template

Previously:

The OSS code for this capability is here: https://github.com/rapidsai-community/rapidsai-csp-utils.

A good chunk of time is spend solving via conda. If you install mamba before RAPIDS, it will solver much faster.

You can use Colab to edit the file install_rapids.py to install what you want using the picker on https://rapids.ai/start.html#rapids-release-selector. you will still pull cudf, as cuml requires it.

Please understand that other Try It Now solutions exist (see website) and the latest compatible RAPIDS version on Colab is 21.12

Upvotes: 1

Related Questions