In 2017, Anaconda Accelerate was discontinued. The GPU CUDA, cuDNN and NCCL functionality are accessed in a Numpy-like way from CuPy. CuPy also allows use of the GPU is a more low-level fashion as well.
Before starting GPU work in any programming language realize these general caveats:
- I/O heavy workloads may make realizing GPU benefits more difficult
- Consumer GPUs (GeForce) can be > 10x slower than workstation class (Tesla, Quadro)
You must have a discrete Nvidia GPU in your laptop or desktop. Check for existence of an Nvidia GPU in your computer by:
Linux: a blank response means an Nvidia GPU is not detected.
lspci | grep -i nvidia
Windows: Look under the “render” tab to see if an Nvidia GPU exists.
- Determine the Compute Capability of your model GPU and install the correct CUDA Toolkit version.
- CuPy is installed distinctly depending on the CUDA Toolkit version installed on your computer.
- reboot or
import cupywill fail with errors like:
AttributeError: type object ‘cupy.core.core.Indexer’ has no attribute ‘reduce_cython’
import cupy dev = cupy.cuda.Device() print('Compute Capability', dev.compute_capability) print('GPU Memory', dev.mem_info)
The should return like:
Compute Capability 75
If you get error like
cupy.cuda.runtime.CUDARuntimeError: cudaErrorInsufficientDriver: CUDA driver version is insufficient for CUDA runtime version
This means the CUDA Toolkit version is expecting a newer Nvidia driver. The Nvidia driver can be updated via your standard Nvidia update program that was installed from the factory. “Table 1” of the CUDA Toolkit release notes gives the CUDA Toolkit required Driver Versions.
Alternatives to CuPy include Numba.cuda, which is a lower-level C-like CUDA interface from Python.
CUDA for Julia is provided in JuliaGPU.