python (65.2k questions)

javascript (44.3k questions)

reactjs (22.7k questions)

java (20.8k questions)

c# (17.4k questions)

html (16.3k questions)

r (13.7k questions)

android (13k questions)

Questions - cuda

CUDA do deprecated functions still function in newer versions?

Does the function cudaGLRegisterBufferObject (deprecated after version 3.0) still work in newer versions (ie 6.X) ? (I know that cudaGraphicsGLRegisterBuffer exists, however I'm doing some work on an ...
test-img

user18029267

c++

windows

cuda

deprecated

Votes: 0

Answers: 1

Latest Answer

It should still work as of CUDA 11.6.2 At the moment, it is documented here so it is still present/available and should be still usable. Deprecated means that it may be dropped in a future CUDA versio...
test-img

Robert Crovella

Does __global__ have overhead over __device__?

This question asks the difference between __device__ and __global__. The difference is: __device__ functions can be called only from the device, and it is executed only in the device. __global__ func...
test-img

user589321

c++

performance

cuda

gpu

Votes: 0

Answers: 1

Latest Answer

Yes, __global__ has overhead, compared to __device__, but there are additional details to be aware of. What you're proposing probably isn't a good idea. __global__ is the device code entrypoint, from...
test-img

Robert Crovella

This CUDA code throws the following error when trying to compile. I changed the matrix dimensions to constants but it still doesn't work

ERRORS: matrix_mul.cu(35): error: expression must have a constant value matrix_mul.cu(35): note #2689-D: the value of variable "BX" (33): here cannot be used as a constant matrix_mul.cu(35...
test-img

JindaJatt

parallel-processing

cuda

Votes: 0

Answers: 1

Latest Answer

Please don't post pictures of code. Please post code as properly formatted text. Those numbers BX and BY must be compile-time-discoverable constants. So you would need to do something like this: cons...
test-img

Robert Crovella

Can you do C++-style CUDA?

I follow some guidelines to deal with memory management in C++. Some examples: I never use malloc. I almost never need or use new or delete. I use smart pointers, and almost never need to write destru...
test-img

user589321

c++

c

cuda

malloc

Votes: 0

Answers: 1

Latest Answer

CUDA started out (over a decade ago) as a largely C style entity. Over time, the language migrated to be primarily a C++ variant/definition. For understanding, we should delineate the discussion bet...
test-img

Robert Crovella

Posts

Questions

Blogs

Jobs