JuliaGPU/CUDA.jl

Exponential of CuArray

Open

#1,418 opened on Mar 1, 2022

View on GitHub
 (7 comments) (0 reactions) (0 assignees)Julia (274 forks)batch import
cuda arrayenhancementgood first issue

Repository metrics

Stars
 (1,408 stars)
PR merge metrics
 (Avg merge 5d 5h) (16 merged PRs in 30d)

Description

Calculating the exponential of a GPU matrix yields errors:

using CUDA
A = cu([1.0, 1;0 1])
CUDA.exp(A)

results in an error: ERROR: ArgumentError: cannot take the CPU address of a CuArray{Float32, 2, CUDA. Mem.DeviceBuffer}

Trying to do this via ExponentialUtils also throws errors. It seems that the function called does not work properly for GPU arrays, either because indexing is done explicitly somewhere, or because some precompiled library is called that can only handle CPU arrays. This seems like a basic functionality, so I'm not sure if I'm requesting a new feature, if this is a bug, or if I'm just doing something wrong.

Contributor guide

Research direction
Inspect the implementation of exp for CuArray in CUDA.jl to identify where CPU addresses are accessed. Look for explicit indexing or calls to CPU only linear algebra libraries. Consider using GPU compatible exponential functions or implementing a GPU kernel.
Tech stack
julia
Domain
backend
Issue type
Bug
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
3
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
1-3 hours
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
Active
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
Mostly clear
Prerequisites
JuliaCUDA
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
50