efficient way of cuda file organization: .cpp .h .cu .cuh .curnel files
19,017
Solution 1
- h, cpp, c, hpp, inc - files that don't contain CUDA C code (e.g. __device__ and other keywords, kernel calls, etc.) and do not make any cuda runtime calls (cuda... functions). It is perfectly fine to call CUDA driver API (cu...) functions from these files. Note that it is possible to compile these files with compilers other then NVCC.
- cu - CUDA C source files. These files are passed to the NVCC compiler to be compiled into linkable (host/device) objects.
- cuh, cuinc - files that are included in .cu files. These files can have CUDA C keywords and/or call CUDA runtime functions.
Solution 2
As an example, suppose to have a GPU-based FDTD code. I usually do the following (Visual Studio 2010).
- main.cpp file, including CPU-GPU memory transfers;
- FDTD.cu, including an
extern "C" void E_update(...)function which contains the kernel<<< >>>call; - main.h file, including the
extern "C" void E_update(...)prototype; - FDTD.cuh, including the
__global__ void E_update_kernel(...)function.
Related videos on Youtube
15 : 21
Hough Transforms with OpenCV Cuda on GPU in C++
33 : 30
10. CUDA C++ Basics
01 : 18 : 20
GPU Programming with CUDA and Thrust | Foundations | Basics | C++ | Short Course!
21 : 54
GPU Raytracer in C++ Part 1
37 : 32
Thread execution in CUDA program scheduling and memory access
02 : 31 : 37
Short Course: Introduction to CUDA (Fall 2021)
25 : 42
Cách tách code C++ thành file .h và .cpp
51 : 47
C++ CRUDS -Part 3 Complete Tutorial for Creating, Updating, Deleting and Search record on textfile
22 : 51
GPU Raytracer in C++ Part 4
![[Lập trình C/C++] Bài 41. Kĩ thuật chia tách file(file.h và file.cpp) trong lập trình C/C++(phần 1)](https://i.ytimg.com/vi/BPO5HyynLmg/hq720.jpg?sqp=-oaymwEcCNAFEJQDSFXyq4qpAw4IARUAAIhCGAFwAcABBg==&rs=AOn4CLD-GC8rrOL0JRAFnr-_sNBFU8vfXA)
19 : 25
[Lập trình C/C++] Bài 41. Kĩ thuật chia tách file(file.h và file.cpp) trong lập trình C/C++(phần 1)
01 : 08 : 32
CUDA Programming
06 : 01
Intro to CUDA (part 4): Indexing Threads within Grids and Blocks
![[Lập trình C/C++] Bài 43. Kĩ thuật chia tách file(file.h và file.cpp) trong lập trình C/C++(phần 3)](https://i.ytimg.com/vi/HT55D8vGb34/hq720.jpg?sqp=-oaymwEcCNAFEJQDSFXyq4qpAw4IARUAAIhCGAFwAcABBg==&rs=AOn4CLAZXvL-TTeZl8yszTknJWpvzxMH-A)
10 : 54
[Lập trình C/C++] Bài 43. Kĩ thuật chia tách file(file.h và file.cpp) trong lập trình C/C++(phần 3)
10 : 28
SYCL 006 - (SETUP) Install NVIDIA CUDA Toolkit 11.1 on Windows
01 : 01 : 30
cudaFlow: Modern C++ Programming Model for GPU Task Graph Parallelism - CppCon 2021
22 : 55
Basics of CUDA Programming | CUDA Terminologies | Host, Device, Kernel, Thread, Block, Grid, Warp
16 : 51
Configuring system's GPU for Theano & Theano dependent Libraries part-2
02 : 30
How to ensure the Mercury Playback Engine GPU Acceleration (CUDA) is turned on in Adobe Premiere Pro
![CUDA GUI VS 2015 + CLR FORM [Simple Adding]](https://i.ytimg.com/vi/6Ed-RRQoPWU/hqdefault.jpg?sqp=-oaymwEcCOADEI4CSFXyq4qpAw4IARUAAIhCGAFwAcABBg==&rs=AOn4CLBVZCaqdhyO08LZyN6UjaIo3j7oNQ)
05 : 11
CUDA GUI VS 2015 + CLR FORM [Simple Adding]
25 : 35
New Features in CUDA 6 Make GPU Acceleration Easier
20 : 24
How Should You Install CUDA and CuDNN || The Correct Way
46 : 09
Simplifying NVIDIA GPU Access: A Polyglot Binding for GPUs with GraalVM
Author by
Ian Decks
Updated on June 05, 2022Comments
-
Ian Decks almost 2 years
What is the most easy to understand/efficient etc. code organization for cuda. After some investigation i found that cuda function declarations should be in .cuh file and implementations reside in .cu file and kernel function implementations in .curnel files. Other c++ stuff in .cpp and .h files ordinarily. Recently i posted a question visual studio .cu file shows syntax error but compile successfully . Is this organization correct? where .cpp calls .cu and it calls kernel function that in .curnel.
-
talonmies about 11 years
curnelfiles? I count myself as reasonably experienced with CUDA and have looked at lot of other peoples projects, and I can't ever remember seeing a singlecurnelfile extension. It certainly isn't anything NVCC understands or has a predefined trajectory for....
-
-
Yola about 6 yearsCould you provide bodies for the files with empty functions implementations? Thanks!
-
polwel over 3 yearsWhy would the driver API be allowed but not the runtime? AFAICT the runtime just a set of regular C functions that I have to link to. Why would any code calling them need to be passed through a special compilation step?
-
BiMo about 2 yearswhat is FDTD ? is it the name of your project or something else in CUDA? sorry to ask, new in CUDA.
