New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[DOC] Better example of dask.neighbors in README #5850
Comments
Hi @wuxianliang, thanks for the issue! Could you run this script https://github.com/rapidsai/cuml/blob/branch-24.06/print_env.sh and put the output of it here? This problem is unexpected and would help us triage |
Click here to see environment details
|
Sorry for my carelessness. I am testing the code with a Maxwell card and a 2070. It is my fault. I will change to use 4 Tesla V100. If I succeed, I will close this issue. |
Thanks for the update @wuxianliang! The Maxwell card unfortunately is unsupported, which could be the root of the issue indeed, but glad to look deeper if you still run into problems! |
Now I am trying to use one 2070 and one 3080ti and run the program as .ipynb in VS code, the problem appears still. Because the water-cooler system has not been finished, I will try 4x V100 later. My working condition now is temporary just for trying the distributed codes, I think rarely people use it like me. I will report 4x V100 case weeks later. |
import cudf
import dask_cudf
from cuml.dask.neighbors import NearestNeighbors
from dask_cuda import LocalCUDACluster
from dask.distributed import Client
cluster = LocalCUDACluster() #use 2 GPUs
client=Client(cluster)
print(client)
df = cudf.DataFrame(
{
"a": range(20),
"b": reversed(range(20)),
"c": range(20)
}
)
print(df)
ddf = dask_cudf.from_cudf(df, npartitions=2) #use 2 GPUs
print(ddf.head(5))
nn = NearestNeighbors(n_neighbors=3, clinet=client)
print(ddf.values)
nn.fit(ddf) #.values
distances, indices = nn.kneighbors(ddf) #.values
print(distances.head(5))
print(indices.head(5))
|
Click here to see environment details
|
@wuxianliang I also have issues with dask NN. And I run into the same error code as you. |
Report incorrect documentation
Location of incorrect documentation
In README, there is an example of cuml.dask.neighbors, I want to try it with tiny data.
Describe the problems or issues found in the documentation
I met a problem.
Steps taken to verify documentation is incorrect
The example in README should work without any conditions I think. If not, I think it is incorrect.
Suggested fix for documentation
Give us an example with tiny data which can work with any two GPUs.
Steps taken to search for needed documentation
I ever tried dask-cuml years ago following the example in tests which was archived for a long time. Now I think cuml.dask should be much easier to use and the tests seem far more advanced to follow. But there are newbies like me still be a newbie never archived. So please keep a working example and let us play!
The text was updated successfully, but these errors were encountered: