Community Profile

photo

Joss Knight

MathWorks

Last seen: 8 days ago Active since 2013

Although I cannot be contacted directly, if you would like to ask me a question all you have to do is mention "GPU" somewhere in your MATLAB Answers question.

Statistics

  • 36 Month Streak
  • Knowledgeable Level 5
  • Pro
  • Revival Level 2
  • First Answer

View badges

Content Feed

View by

Answered
minibatchqueue function cannot generate the expected MiniBatchSize
You've asked your arrayDatastore to iterate over the rows because that's the default. So as far as arrayDatastore is concerned, ...

18 days ago | 1

| accepted

Answered
RTX 3090 vs A100 in deep learning.
According to the spec as documented on Wikipedia, the RTX 3090 has about 2x the maximum speed at single precision than the A100,...

18 days ago | 0

| accepted

Answered
GPUCoder does not generate parallelized code
This looks about right to me, because your kernel is too simple and you're transferring data from and to the CPU on every call. ...

25 days ago | 1

Answered
Can I run custom Matlab function or gpuArray on another GPU?
You can use parallel syntax to process other arrays on other GPUs at the same time, or to process some data on the CPU at the sa...

1 month ago | 0

| accepted

Answered
How can run and upload my Deep Learning model in cloud?
<https://uk.mathworks.com/help/deeplearning/ug/deep-learning-in-the-cloud.html Start here>

1 month ago | 0

| accepted

Answered
Error in minibatchqueue (line 290) numVariables = numel(getPreviewFromDatastore(originalDatastore));
Make sure the accompanying file augmentDataForLD2HDCT.m that comes with this example is on the path when you run your code.

1 month ago | 0

| accepted

Answered
Reorganizing current data structure in order to take advantage of GPU
Evidently with your nested structure you are not required to have uniform data - every element at each level can have a differen...

1 month ago | 1

| accepted

Answered
Saving images quickly for huge datasets
It's hard to say what will speed things up, since we don't know which part of the process is slow. Is saving slow? Is computing ...

1 month ago | 1

Answered
Help me for this CUDA Error
I see two calls to predict which don't specify ExecutionEnvironment and therefore will be trying to use your college computer's ...

1 month ago | 0

| accepted

Answered
REINFORCE algorithm- unable to compute gradients on latest toolbox version
What is underlyingType(observationBatch) underlyingType(lossData) ?

2 months ago | 1

| accepted

Answered
Retraining Alexnet - works on windows 7 not on windows 10?
The difference here is not your OS but your GPU. You have a newer GPU on your Windows 10 machine. Probably it is an Ampere GPU w...

2 months ago | 1

Answered
How to use GPU only on the nodes that have it
|canUseGPU| is the favoured way to guard your code.

2 months ago | 0

Answered
Why do I receive the error "CUDA_ERROR_ILLEGAL_ADDRESS" despite of successfully compiling the cuda mex function.
I suspect your kernel is writing into out-of-bounds memory. The next time the device synchronizes the error will be reported, wh...

2 months ago | 0

Answered
Not using GPU for trainNetwork?
Nearly always in cases like this you just need to install your GPU drivers from NVIDIA: https://www.nvidia.co.uk/Download/driver...

2 months ago | 0

| accepted

Answered
Develop WGAN-GP for 3-D image
You definitely need to use EnableHigherOrderDerivatives here because you are including computed gradients in the loss term. With...

2 months ago | 0

| accepted

Answered
How can I validate CNN after training?
The easiest way to validate after training for classification is to do exactly what you do in your example code to check the acc...

2 months ago | 0

| accepted

Answered
(Windows 10) Limiting GPU access: What is the default file location of the functional MATLAB application?
Deselecting the device for use for computation doesn't stop MATLAB using it for graphics. The actual exe you're looking for is ...

2 months ago | 0

Answered
Tutorial / classes / training for developing asynchronous CUDA and MEX code?
Make all the variables involved gpuArray objects and those lines of code will run as asynchronously as the GPU allows. This mean...

3 months ago | 0

| accepted

Answered
Error with averagePooling1dLayer (pooling dimension lower than pooling size)
The time dimension is variable so just because you happen to be passing a long sequence into the network doesn't mean you're not...

3 months ago | 0

| accepted

Answered
pixelLabelImageDatastore not partitionable despite underlying imageDatastore and pixelLabelDatastore are
For parallel training of a semantic segmentation network, you're supposed to use combine to join the two input datastores rather...

3 months ago | 0

| accepted

Answered
Error in helperModClassTrainingOptions (line 29) 'CheckpointPath',checkpointPath,...
You need to pass the checkpointPath variable to your function.

3 months ago | 0

| accepted

Answered
How to increase gpu usage?
To get GPU utilization high you need to give it a lot of high resolution work to do. Often RL problems, especially the ones in o...

3 months ago | 1

| accepted

Answered
Increasing effective GPU memory through parallel GPUs?
The only way to do this is to divide your computation up manually and spread the work across your GPUs by using a parallel pool ...

4 months ago | 0

Answered
Parallel CPU computing for recurrent Neural Networks (LSTMs)
That doc page is about shallow networks (using train) rather than deep networks (using trainNetwork). Parallel training in train...

4 months ago | 1

Answered
CNN-LSTM validation data underperforming compared to training data
I don't know why your model seems to be overfitting, but I can confirm that your validation data is being run through the exact ...

4 months ago | 1

Answered
Error using pagefun with mldivide
This is a known bug in cublas introduced in CUDA 11 and fixed in the next MATLAB release. The workaround, as you correctly disco...

4 months ago | 0

| accepted

Answered
Is it possible to use variable length arrays with SequenceInputLayer in a custom training loop, i.e. as a dlarray?
dlnetwork objects do not take your entire dataset as input, they expect to receive a single batch at a time. You need to loop ov...

4 months ago | 0

| accepted

Answered
Can I use my CPU and GPU simultaneously for Numerical Simulation?
canUseGPU is the easiest way to write code that diverges based on whether a supported GPU is available. A good design should nee...

4 months ago | 2

Answered
Force CPU Parallel pool
You should use 'ExecutionEnvironment','cpu' for training on your local machine. This is multithreaded and will use all your core...

4 months ago | 2

Answered
Increrasing the validation accuracy
There's no way to do this automatically, but in your case just using the default 'OutputNetwork' setting ( last-iteration ) woul...

5 months ago | 0

Load more