microsoft / LightGBM

A fast, distributed, high performance gradient boosting (GBT, GBDT, GBRT, GBM or MART) framework based on decision tree algorithms, used for ranking, classification and many other machine learning tasks.
https://lightgbm.readthedocs.io/en/latest/
MIT License
16.69k stars 3.83k forks source link

[Off-topic / Do not answer] GPU installation pictures #398

Closed Laurae2 closed 7 years ago

Laurae2 commented 7 years ago

issue used for adding pictures for GPU installation on Windows

Changing PATH:

image

image

image

MinGW install:

image

Checking whether we are using R's MinGW or stand-alone MinGW (we are using R here):

image

What you get at the end of Boost installation (this was for MinGW 5.3, while Rtools is not 5.3 currently):

image

This is what happens when we link to Rtools's MinGW:

image

Here, downloading Git for Windows:

image

Here, cmake download:

image

Here I use cmake on C:/xgboost/LightGBM:

image

image

Scroll down, and check USE_GPU:

image

image

In Git Bash, we can run make and we clearly see it compiled successfully with GPU!:

image

Some hack to do on CLI/Python GPU to avoid segmentation fault:

image

Alternative hack:

image

When we run it with CLI with GPU support:

image

We can even compile for Python without any issue:

image

In Python and Windows, we cannot see the GPU so you have to rely on the delay between the issued command and the real training:

image

Getting Makeconf in R:

image

Makevars looks like that:

image

Installation in R working peacefully:

image

Now we can try with GPU support in R:

image

Laurae's LightGBM package:

image

Debugging:

remove this:

image

add debug in cmake:

image

compile with debug:

image

run:

image

Laurae2 commented 7 years ago

(ignore this if subscribed)

random stuff (pictures for PR)

image

image

image

image

image

image

image

github-actions[bot] commented 1 year ago

This issue has been automatically locked since there has not been any recent activity since it was closed. To start a new related discussion, open a new issue at https://github.com/microsoft/LightGBM/issues including a reference to this.