Python interface to GPU-powered libraries

Problem

step 1: install scikit-cuda==0.5.2 step 2: Just do import skcuda.linalg as linalg

Error

File "C:\Users\cegprakash.virtualenvs\similarity_articles-JuZA6FGA\lib\site-packages\skcuda\linalg.py", line 23, in from . import cublas File "C:\Users\cegprakash.virtualenvs\similarity_articles-JuZA6FGA\lib\site-packages\skcuda\cublas.py", line 55, in raise OSError('cublas library not found') OSError: cublas library not found

Environment

• OS platform : Windows 10
• Python version : 3.6.3
• CUDA version : V10.0.130 (Installed it from https://developer.nvidia.com/cuda-downloads)
• PyCUDA version : 2018.1.1
• scikit-cuda version : 0.5.2

Hey,

I have the following code:

    # measurement
    from timeit import timeit
    import pycuda.gpuarray as gpuarray
    import pycuda.autoinit
    import numpy as np
    from skcuda.linalg import svd
    import skcuda
    skcuda.misc.init()
    N = 6400
    Y = np.random.randn(N, N) + 1j*np.random.randn(N, N)
    X = np.asarray(Y, np.complex64)
    a_gpu = gpuarray.to_gpu(X)
         
    tm = timeit("svd(a_gpu, jobu='A', jobvt='A', lib='cusolver')", 
                    globals={'a_gpu': a_gpu, 'svd': svd}, 
                    number=1)

You should be able to run it if you have skcuda, pycuda and pandas installed.

What I am trying to do is to compare the speed with numpy's SVD version:

    import numpy as np
    import pandas as pd
    from timeit import timeit
    N = 10000
    X = np.random.randn(N, N).astype('float32')
    tm = timeit('svd(X, full_matrices=True)', 
                    globals={'X': X, 'svd': np.linalg.svd}, 
                    number=1)

but when I compare the results, it is terrible for GPU. On CPU, I compute SVD matrix about 10000 rows in 15 minutes on single CPU with numpy, while on skcuda SVD it takes about 90 minutes for matrix about 6400 rows.

Am I doing something wrong? I can see that skcuda is using the card (at least it is warming up and taking up memory on the card).

I'm consistently getting this problem

raise OSError('%s not found' % _load_err) OSError: libcula_lapack.dylib, libcula_core.dylib not found

I had to change the default library names in cula.py since CULA 16 gives libcula_lapack.dylib, libcula_core.dylib for Mac OSX 10.9.5. CULA is installed and works. All CULA libraries are included in the DYLD_LIBRARY_PATH. CUBLAS is working fine, but the scikits.cuda just can't find CULA, even though the relevant .dylib files are in the library path.

Hi, I am new user to Scikits.cuda and I am trying to set it up on a machine running Os x 10.8.2 with nvidia cuda 5.0 already installed and working on it. I installed scikits.cuda using easy_install and when I try to import it I get an error message saying that "CUDA driver not found".

Please find the log below,

Python 2.7.5 (default, Aug 1 2013, 01:01:17) [GCC 4.2.1 Compatible Apple Clang 4.1 ((tags/Apple/clang-421.11.66))] on darwin Type "help", "copyright", "credits" or "license" for more information.

import pycuda.autoinit import scikits.cuda.fft as gpufft Traceback (most recent call last): File "", line 1, in File "/opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/scikits.cuda-0.042-py2.7.egg/scikits/cuda/fft.py", line 19, in import misc File "/opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/scikits.cuda-0.042-py2.7.egg/scikits/cuda/misc.py", line 16, in import cuda File "/opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/scikits.cuda-0.042-py2.7.egg/scikits/cuda/cuda.py", line 8, in from cudadrv import * File "/opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/scikits.cuda-0.042-py2.7.egg/scikits/cuda/cudadrv.py", line 29, in raise OSError('CUDA driver library not found') OSError: CUDA driver library not found

Please help me to resolve this issue.

Thanks & Regards, Kartheek.

Hi there,

Ive been wanting to use cufft on windows using pycuda; I noted scikits cuda doesnt support windows, but monkey-patching it so that at least cufft works on win7 64bit proved quite trivial (see below).

import platform if platform.system() == "Windows": _libcufft = ctypes.windll.LoadLibrary('cufft32_42_9.dll')

I am anything but an expert on these matters, so I dont know how hard it is to generalize this; but if at least cufft would be made to have windows support this easily, id love to see it get merged into the main branch, so I dont have to ship my project with this monkey-patch of mine.

Thanks for the great work on this!

I am failing to run my program that starts with some standard imports.

test.py

import pycuda.driver as cuda
import pycuda.autoinit
import pycuda.gpuarray as gpuarray
import pycuda.cumath
from pycuda.compiler import SourceModule
import numpy as np
from skcuda import linalg
linalg.init()

Error output:

File "test.py", line 10, in <module>
    linalg.init()
  File "/usr/lib/python2.7/site-packages/skcuda/misc.py", line 177, in init
    from . import cusolver
  File "/usr/lib/python2.7/site-packages/skcuda/cusolver.py", line 55, in <module>
    raise OSError('cusolver library not found')
OSError: cusolver library not found

All pyCUDA and scikit-CUDA dependencies are installed freshly and all CUDA paths are set up according CUDA documentation.

Cheers

I am trying to run scikits.cuda on my OS10.8.3 apple laptop. I modified utils.py to look for libdl.dylib rather than libdl.so, and used homebrew to install gobjdump (I believe) help find the cublas version. I also tried changing the objdump invocation to gobjdump/llvm-objdump.

I'm now stuck on line 227 in utils.py, above which it says:

XXX This approach to obtaining the CUBLAS version number may break Windows/MacOSX compatibility XXX

This appears to have happened. I'll probably keep plugging away at this throughout the day, but any chance of help?

Hi. I am facing an issue with the cublas library. I cannot import cublas library.

Problem

After installing the dependences: setuptools and installing scikit-cuda with: "pip install scikit-cuda" I run the following line:

              import skcuda.linalg as linalg

The execution give the following:

Traceback (most recent call last):

File "", line 1, in import skcuda.linalg as linalg

File "C:\Users\victo\Anaconda3\envs\spikesorting_tsne\lib\site-packages\skcuda\linalg.py", line 21, in from . import cublas

File "C:\Users\victo\Anaconda3\envs\spikesorting_tsne\lib\site-packages\skcuda\cublas.py", line 55, in raise OSError('cublas library not found')

OSError: cublas library not found

Environment

List the following info:

• Microsoft Windows [Versión 10.0.17713.1002]
• Python 3.6.6
• Cuda Toolkit 9.0. I installed downloading the package from Nvidia web page, including the cudnn Library (cudnn-9.0-windows10-x64-v7.2.1.38). I made sure the environment variables CUDA_PATH and CUDA_PATH_V9_0 were set.
• scikit-cuda 0.5.1

OS X noob and have never encountered this one on LINUX machines with similar software configurations. Running skcuda version 0.5.1 in ANACONDA env with CUDA toolkit 7.5 & pycuda installed on OS X 10.11.2.

from skcuda.cuda import fft

returns an returns OSError: cufft library not found, also affecting other python programs using CUDA e.g., mne.cuda @Eric89GXL

Hi, the installation of scikits.cuda via pip failed: "nvcc not in path"

[email protected]:~$ sudo pip install scikits.cuda Downloading/unpacking scikits.cuda Downloading scikits.cuda-0.042.tar.gz (97kB): 97kB downloaded Running setup.py (path:/tmp/pip_build_root/scikits.cuda/setup.py) egg_info for package scikits.cuda

Requirement already satisfied (use --upgrade to upgrade): numpy in /usr/local/lib/python2.7/dist-packages (from scikits.cuda) Downloading/unpacking pycuda>=0.94.2 (from scikits.cuda) Downloading pycuda-2014.1.tar.gz (1.6MB): 1.6MB downloaded Running setup.py (path:/tmp/pip_build_root/pycuda/setup.py) egg_info for package pycuda *** WARNING: nvcc not in path. ************************************************************* *** I have detected that you have not run configure.py. ************************************************************* *** Additionally, no global config files were found. *** I will go ahead with the default configuration. *** In all likelihood, this will not work out. *** *** See README_SETUP.txt for more information. *** *** If the build does fail, just re-run configure.py with the *** correct arguments, and then retry. Good luck! ************************************************************* *** HIT Ctrl-C NOW IF THIS IS NOT WHAT YOU WANT ************************************************************* Continuing in 1 seconds... ..
Traceback (most recent call last): File "", line 17, in File "/tmp/pip_build_root/pycuda/setup.py", line 216, in main() File "/tmp/pip_build_root/pycuda/setup.py", line 88, in main conf["CUDA_INC_DIR"] = [join(conf["CUDA_ROOT"], "include")] File "/usr/lib/python2.7/posixpath.py", line 77, in join elif path == '' or path.endswith('/'): AttributeError: 'NoneType' object has no attribute 'endswith' Complete output from command python setup.py egg_info: *** WARNING: nvcc not in path.

*** I have detected that you have not run configure.py.

*** Additionally, no global config files were found.

*** I will go ahead with the default configuration.

*** In all likelihood, this will not work out.

*** See README_SETUP.txt for more information.

*** If the build does fail, just re-run configure.py with the

*** correct arguments, and then retry. Good luck!

*** HIT Ctrl-C NOW IF THIS IS NOT WHAT YOU WANT

Continuing in 1 seconds...

Traceback (most recent call last):

File "", line 17, in

File "/tmp/pip_build_root/pycuda/setup.py", line 216, in

main()

File "/tmp/pip_build_root/pycuda/setup.py", line 88, in main

conf["CUDA_INC_DIR"] = [join(conf["CUDA_ROOT"], "include")]

File "/usr/lib/python2.7/posixpath.py", line 77, in join

elif path == '' or path.endswith('/'):

AttributeError: 'NoneType' object has no attribute 'endswith'

Cleaning up... Command python setup.py egg_info failed with error code 1 in /tmp/pip_build_root/pycuda Storing debug log for failure in /home/marco/.pip/pip.log

Yesterday I had the same problem with Theano. Before installing some packages with Anaconda distribution and then removing Anaconda (didn't want to rely on proprietary distribution), everything with Theano went fine. But just right after anaconda's removal, exactly the same error's message appeared. For Theano I solved the problem, putting the path into .theanorc file (theano's configuration file-"tableau de bord" file).

Any hints to solve the problem for scikits.cuda?

Looking forward to your kind help. Kind regards. Marco

Hi guys,

I recently installed scikits.cuda, the latest version. I simply got the zip from github and executed sudo python setup.py install, so I have the latest version.

The code that I am running stops when it need cublasCgemmBatched, that have been added recently. I obtain:

AttributeError: 'module' object has no attribute 'cublasCgemmBatched' Apply node that caused the error: BatchedComplexDotOp(GpuContiguous.0, GpuContiguous.0) (I am running the code of Sander Dieleman fft for convolution in theano http://benanne.github.io/2014/05/12/fft-convolutions-in-theano.html).

Is there something obvious that could explain the fact that it seems not to find it?

Thanks by advance, Sam

Problem

I want to compute the skcuda equivalent to A.T @ b for numpy arrays (A, b) when A has 2 dimensions and b has one dimension. Here's some simple test data

import numpy as np
import pycuda.gpuarray as gpuarray
import skcuda.linalg as skla

A_gpu = gpuarray.to_gpu(np.ones((3, 2)))
b_gpu = gpuarray.to_gpu(np.ones(3))

I would expect that the function call below produces the desired output. But instead it raises a ValueError:

skla.dot(A_gpu, b_gpu, transa='T')

Traceback (most recent call last):
  File "/home/riley/anaconda3/envs/rla39a/lib/python3.9/site-packages/IPython/core/interactiveshell.py", line 3441, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)
  File "<ipython-input-27-3e66e0c2b9c5>", line 1, in <module>
    skla.dot(A_gpu, b_gpu, transa='T')
  File "/home/riley/anaconda3/envs/rla39a/lib/python3.9/site-packages/skcuda/linalg.py", line 1060, in dot
    return out.reshape(out_shape)
  File "/home/riley/anaconda3/envs/rla39a/lib/python3.9/site-packages/pycuda/gpuarray.py", line 912, in reshape
    raise ValueError("total size of new array must be unchanged")
ValueError: total size of new array must be unchanged

The function call below computes the expected result

c_gpu = skla.dot(b_gpu, A_gpu)

Environment

• Ubuntu 18.04
• Python 3.9
• CUDA 9.1.0 (I forget how I installed it, sorry!)
• PyCUDA 2021.1
• scikit-cuda 0.5.3.

Problem

I am using Colab to do GPU programming. I came up with a problem when I am using matrix multiplication function Sgemm. The three inputs of the function would be gpuarrays. Once I got the result, I tried to transform the result gpuarray (which is a flatten one) by first transposing (.T) and second flattening (ravel()). I think this could get a matrix that is completely different from the raw output since I rearrange the order. But the truth is that I still got the identical matrix. Therefore, I want to know how to rearrange the order of gpuarray. Thanks in advance.

Environment

I think the environment is Linux 5.4.104+ and Python version is 3.7.12. CUDA is integrated in Colab with version Cuda 11.1. Pycuda version is 2021.1, and scikit-cuda version is 0.5.3.

Probably caused by a misprint in cusolver.py, "11" is missed in _win32_version_list.

in line 26 two "10":

_win32_version_list = [110, 10, 10, 100, 92, 91, 90, 80, 75, 70]

I think it should be

_win32_version_list = [110, 11, 10, 100, 92, 91, 90, 80, 75, 70]

Problem

I found pycuda.gpuarray.dot() is different from numpy.dot(). So I want to use linalg.dot() by import skcuda.linalg. (The calculation method of linalg.dot() is equivalent to numpy.dot()?).

Then I got ImportError: CUSOLVER library only available in CUDA 7.0 and later

Someone know why this will happen? Thanks for your help.

Environment

OS platform: Windows 10 Python version: 3.6.8 CUDA version: V10.2.89 PyCUDA version: 2021.1 scikit-cuda version: 0.5.3

Problem

while importing skcuda.linalg, OSError: cublas library not found.

I tried with other released pycuda, still not working.

Environment

List the following info:

• OS platform (including distro if you are on Linux):windows10
• Python version ==3.8.8
• CUDA version==11.2.142 + downloaded and installed through nvidia webset
• PyCUDA version == 2021.1+cuda114
• scikit-cuda version (including GitHub revision if you have installed it from there) ==0.5.3

Are there any plans to get a 0.5.4 release out of the door? It looks like there has been some good work on scikit-cuda over the last two years (since 0.5.3 was released), and it would be great to be able to get it in a proper release rather than installing from git.