photo

Joss Knight

MathWorks

Last seen: 2 days ago |  Active since 2013

Followers: 0   Following: 0

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

  • Guiding Light
  • 36 Month Streak
  • Knowledgeable Level 5
  • Revival Level 2
  • First Answer

View badges

Feeds

View by

Answered
Matlab Support for float32/single and float16/half datatypes in GPU Sparse Matrix Multiplication
Half support is not currently planned for sparse.

7 days ago | 3

Answered
Error: "Incorrect type of 'Z' for 'predict' in Layer 'samplelayer'. Expected 'gpuArray', but instead was 'single'." This error is shown during using a custom deep layer.
Hello. You need to return an array of the same data type and storage type as the input. Somehow your data is no longer on the GP...

16 days ago | 0

| accepted

Answered
Speeding up matrix expotentials by using GPU
Your GTX 1650 is designed for single precision computing. In single precision it has a peak performance of about 3 teraflops, wh...

20 days ago | 3

| accepted

Answered
GPU Array Max Dimensions/Size (i.e., int32 of ~2e9) Not Same as Other CUDA-Enabled Languages (i.e., Python 3.9 - Tensorflow 2.11)
In CUDA 12 NVIDIA introduced a limited set of 64-bit integer APIs in cublas and other libraries, allowing arrays with more eleme...

23 days ago | 1

| accepted

Answered
Multilayer Perceptron with More Than One Output and Data Interpretation?
You can continue to use trainNetwork if you don't want to use dlnetwork. dlnetwork obviously provides much more flexibility as w...

1 month ago | 0

Answered
Matlab continuously launching on GPU
Using the NVIDIA Control Panel, disable hardware acceleration for MATLAB. Alternatively, type |opengl software| on startup.

1 month ago | 0

Answered
Using transformer neural network for classification task
You've passed |layers| instead of |lgraph| to |trainNetwork|.

1 month ago | 0

| accepted

Answered
I get the illegal instruction error while running Matlab code with GPU
On the face of it it does seem like you've hit some kind of bug in CUDA 11. If you can upgrade your MATLAB to a more recent vers...

1 month ago | 0

| accepted

Answered
GPU memory usage for Hadamard product
I can't reproduce this. You say you have 48Gb of GPU memory available...have you checked this? Try running gpuDevice and lookin...

2 months ago | 0

Answered
Initializing LSTM which is imported using ONNX
This code is suspect % initialize input arrays obs = dlarray(rand(obs_size),"BS"); state = dlarray(rand(state_size),"SBS"); ...

2 months ago | 0

Answered
trainnet gives training loss is NaN
Do your network weights contain NaNs? Try this nansInMyNetwork = ~(all(cellfun(@allfinite, net.Learnables.Value)) && all(cellfu...

2 months ago | 0

Answered
how to use printf inside a CUDA kernel?
Just use it, and launch MATLAB from a terminal. On Linux, the output will appear in the terminal window. On Windows you will nee...

2 months ago | 0

| accepted

Answered
mexcuda compiler error: "__global__" does not apply here
You cannot call a function declared __global__ from another function declared __global__. Declare the second function as __devic...

3 months ago | 0

| accepted

Answered
CUSOLVER_STATUS_INTERNAL_ERROR with RTX 4080 Super in MATLAB 2022a C MEX Function
This is a known issue with CUDA 11.2, Ada cards, and cusolver. The solution is to upgrade MATLAB to R2023a or above, because it ...

3 months ago | 1

| accepted

Answered
CUSOLVER_STATUS_INTERNAL_ERROR with RTX 4080 Super in MATLAB 2022a C MEX Function
It's probably necessary for you to provide your code, since any number of things could have gone wrong prior to intializing cuso...

3 months ago | 0

Answered
why do I got such a low score when i use bench function in matlab2024a
Do check your computer's power settings, to make sure they are set to "best performance".

3 months ago | 0

Answered
Matab is very slow in processing figures: might it be caused by NVidia Optimus?
Check your laptop's power settings and make sure they are in 'best performance' mode.

3 months ago | 0

| accepted

Answered
Which is the difference between 'multi-gpu' and 'parallel-gpu' in 'trainingOption' function of the DeepLearning Toolbox?
The purpose of 'multi-gpu' is effectively to try to ensure you are using a local pool with numGpus workers, rather than needing ...

3 months ago | 1

Answered
MATLAB does not detect NVIDIA T400 GPU despite updated drivers
The 516 driver is too old for R2024a, which is on CUDA 12.2. Download the recommended 552 driver for your T400 from https://www....

3 months ago | 0

Answered
MATLAB GPU UBUNTU 20.04 compatibility CUDA driver
Try the 555 and 550 drivers recommended for your device available from https://www.nvidia.com/en-gb/drivers/, rather than relyin...

3 months ago | 0

Answered
Memory leak occurring when repeatedly performing matrix left division on GPU
Thank you very much for reporting this. This appears to be a bug in MATLAB's gpuArray support for overdetermined solves. For now...

4 months ago | 0

| accepted

Answered
How do I use multiple inputs in mexcuda?
You have passed in two inputs but you have a check if (nrhs!=1), so you are erroring because you don't have one input. You need ...

5 months ago | 1

Answered
Why training stops showing error (out of memory) during implementing transfer learning with pre-trained network despite having a laptop of well configuration?
You have a very small GPU I'm afraid. You need to lower your |MiniBatchSize| until it works. Also, set your |ExecutionEnvironmen...

5 months ago | 1

| accepted

Answered
Parallel Processing is better than buying another MATLAB licence?
There may be a misunderstanding here. Your typical license is to a user or a computer. That user, or that computer, can open as ...

5 months ago | 0

Answered
I can`t solve out this problem, there is always Output argument "varargout{2}" (and possibly others) not assigned a value in the execution with "dlarray/dlgradient" function.
Your mistake is |dlfeval(@dlgradient,...)|. You need to put your code that computes the loss and the loss gradients into a funct...

5 months ago | 0

| accepted

Answered
AlexNet (filter groups) cannot be well resolved inside dlfeval (dimension mismatch error)
Sorry about this. This is because you have an old version of alexnet (although yes, it shouldn't error). For now, please get you...

5 months ago | 0

| accepted

Answered
arrayfun vs loops again
I wish it were a cleverer answer, but I'm afraid that it's simply that MATLAB has been heavily optimized for |for| loops over th...

5 months ago | 3

Answered
Extreme memory usage when using gather
As others have worked out, it looks like the issue is the indexed assignment into u: u(TERRAIN:end,m) = abs(gather(head(ffield,...

5 months ago | 0

| accepted

Answered
How can i fix the error mismatch format when convert multiplication layer from LayerGraph to dlnetwork in prune and quantize network example
For one solution, replace the fully connected layers with convolution layers with filters the size of the input and num filters ...

5 months ago | 0

Answered
How can i fix the error mismatch format when convert multiplication layer from LayerGraph to dlnetwork in prune and quantize network example
That's an annoying limitation which is hopefully fixed in current releases. Is there any particular reason why you can't use |mu...

5 months ago | 0

Load more