H2O AutoML GPU Support
484 views
Skip to first unread message
blwh...@ncsu.edu
May 28, 2018, 8:13:18 PM5/28/18
to H2O Open Source Scalable Machine Learning - h2ostream
Hello,
I am considering building a Linux ML workstation. I was wondering if I bought a CUDA-supported GPU, would I be able to use this extra compute power for AutoML? I see the compatibility with H2O4GPU, Driverless AI, and Deep Water online but not much for AutoML. If anyone has had success with this or knows the limitations I would appreciate the advice.
Thanks!
Erin LeDell
May 29, 2018, 12:44:17 AM5/29/18
to blwh...@ncsu.edu, H2O Open Source Scalable Machine Learning - h2ostream
H2O AutoML is part of H2O, which is largely CPU-powered. XGBoost is the
only algo inside of H2O that can make use of GPUs. Once we add XGBoost
to AutoML, then you can make use of GPUs in AutoML:
https://0xdata.atlassian.net/browse/PUBDEV-4507?filter=20700
-Erin
only algo inside of H2O that can make use of GPUs. Once we add XGBoost
to AutoML, then you can make use of GPUs in AutoML:
https://0xdata.atlassian.net/browse/PUBDEV-4507?filter=20700
-Erin
Reply all
Reply to author
Forward
0 new messages