Answered
why imresize() is slower with gpuArray?
It's hard to tell because your code is very confusing, but it looks to me like you are resizing 900 images one image at a time. ...

7 years ago | 1

Answered
Parallelizing MATLAB code using many GPU cores
GPU cores do not work like CPU cores. They cannot run independent tasks. To use the GPU with MATLAB, <https://uk.mathworks.co...

7 years ago | 1

| accepted

Answered
Increase GPU Throughput During training
The problem with using a |ReadFcn| is that it prevents MATLAB from being able to do I/O in batch and in the background, because ...

7 years ago | 0

Answered
What kind of toolboxes do I need to implement Deep neural network
For Deep Learning plus GPU you need: MATLAB, Parallel Computing Toolbox, Neural Network Toolbox (soon to be renamed Deep Learnin...

7 years ago | 1

| accepted

Answered
error when using multi-gpu option for trainingNetwork
You have a recurrent network. So you have a network with a SequenceInputLayer and at least one LSTM layer. For these networks, m...

7 years ago | 0

| accepted

Answered
Matlab is able to utilize only a part of actual available VRAM.
The division of GPU memory into compute and graphics is dictated by the driver - it isn't something that a CUDA application has ...

7 years ago | 0

Answered
How can I adapt arrayfun() to operate on vectors rather than scalars?
Look at <https://www.mathworks.com/help/distcomp/pagefun.html |pagefun|> for mixing in calls to vector operations with your ordi...

7 years ago | 0

| accepted

Answered
Deep Learning using GPU
The most likely explanation is a kernel timeout, see <https://uk.mathworks.com/matlabcentral/answers/410236-cuda-error-after-cal...

7 years ago | 0

Answered
The computation result of imresize of Matlab R2017b is wrong?
It's not a problem with |imresize|, it's a problem with save/load and indexing. You can eliminate your problem by deleting |ss| ...

7 years ago | 2

| accepted

Answered
How can I use the GPU to parallelize computation of average pairwise distances?
See also <https://www.mathworks.com/matlabcentral/answers/347838-computing-the-distance-between-a-grid-of-points-and-n-points-wi...

7 years ago | 0

Answered
GPU imdilate of 3D volume
No, this isn't possible unless you wish to write your own implementation. You are not the first to request this feature, it has ...

7 years ago | 1

| accepted

Answered
Calculate lsqr of two arrays on GPU
The least squares solution to a system of linear equations of one variable is pretty simple to work out, so you shouldn't really...

7 years ago | 0

| accepted

Answered
How can I use the GPU to parallelize computation of average pairwise distances?
You can do it with basic arithmetic but really you're just looking for the function <https://uk.mathworks.com/help/stats/pdist2....

7 years ago | 0

| accepted

Answered
Cuda error after calculating mahalanobis distance for large dataset.
It looks like you are suffering from a kernel execution timeout due to a long-running kernel. The easiest way to fix this is to ...

7 years ago | 0

| accepted

Answered
I am unable to convert my brain tumor detection code done in matlab to gpu coder. I want to perform GPU parallel processing of that code. Please help.
Read the documentation and try some things. Here's a place to start: <https://uk.mathworks.com/help/distcomp/gpu-computing-in-ma...

7 years ago | 0

Answered
Debugging: finding all instances of gpuArray in large code base [or finding all instances of other function calls]
On Linux? Use |grep|. On Windows? Just use Windows Explorer and type gpuArray into the search bar in the parent folder. Or write...

7 years ago | 0

| accepted

Answered
Multiply each 3D gpuArray page column with each other
It's hard to be sure whether you are literally after what you wrote, or whether you have just articulated it wrong. If you liter...

7 years ago | 0

| accepted

Answered
How can a create a vector directly on GPU, just like we can create a vector on CPU?
It's a little bit far-fetched to expect to be able to create an _arbitrary_ array directly on the GPU - after all, the only way ...

7 years ago | 1

| accepted

Answered
Matlab Online with GPU
We're working on a GPU version of MATLAB online, but I don't know when it's expected. The hardware sharing is complicated.

7 years ago | 3

| accepted

Answered
Default floating point precision with mexcuda
No, the compiler never casts anything to float. There are options to use some fast math operations that are slightly less confor...

7 years ago | 1

| accepted

Answered
GPU memory overhead dependent on fft dimension.
MATLAB uses cufft, so the behaviour is whatever its behaviour is. The implication of the batching API as described by the doc - ...

7 years ago | 0

| accepted

Answered
Failing to free GPU memory on cleared/replaced/gathered variables: 2017a
It's difficult to say exactly what is going on here, however I'm fairly sure that it is the cublas context that is occupying mem...

7 years ago | 0

| accepted

Answered
What does this mean when I run Matlab in Ubuntu 16.04?
Try starting MATLAB without hardware rendering using the |softwareopengl| option, see if that fixes it: matlab -softwareope...

7 years ago | 2

Answered
Help in next error: Undefined variable "net" or class "net.Layers". instruction: montage(mat2gray(gather(net.Layers(2).Weights)));
Well, you've called your network |net4clas23|, not |net|, so no variable of that name exists.

7 years ago | 0

Answered
Advice on when to start writing custom CUDA
This particular problem is a good example of something that is solved better with a custom kernel, but many real problems aren't...

7 years ago | 1

| accepted

Answered
Parallel computing toolbox...gpu choice
Compute Capability just tells you what generation of NVIDIA GPUs the device comes from, so its capability is related to its inst...

8 years ago | 0

| accepted

Answered
X'SX were X is a column vector and S is a matrix.
I suppose what you mean is that |X| is d-by-N, not d-by-1? So it seems like what you want is SX = S * X; for i = 1:N ...

8 years ago | 1

| accepted

Answered
SIFT features in GPU
Sorry about that. Your best bet is to use the <https://www.mathworks.com/discovery/matlab-opencv.html OpenCV interface> to lever...

8 years ago | 0

Answered
Is it possible to use both CPU and GPU to train convolution neural network (CNN)?
Yes it is. Are you averse to reading documentation? If not, try: * Setting the <https://www.mathworks.com/help/nnet/ref/predi...

8 years ago | 0

Answered
Back propagation on a deep learning custom layer with a maxpooling "alike" functionality
Don't write this: function [Z, memory] = forward(layer, X) N = size(X,4); X_reshaped = res...

8 years ago | 0