Help me for this CUDA Error
1 Ansicht (letzte 30 Tage)
Ältere Kommentare anzeigen
Akshay Sharma A S
am 12 Apr. 2022
Kommentiert: Akshay Sharma A S
am 16 Apr. 2022
This is regarding my Project Work doing on Matlab for detection of Diabetic Retinopathy stages using MATLAB.
trainingData=imageDatastore('datastorage',...
'IncludeSubFolders',true,'LabelSource','foldername');
[imdsTrain,imdsValidation]=splitEachLabel(trainingData,0.75,'randomized');
countEachLabel(trainingData);
categories={'Mild NPDR','Moderate NPDR','No DR','Proliferative DR','Severe NPDR'};
layer='fc1';
netLayers = [imageInputLayer([227 227 3],'Name','inp')...
,convolution2dLayer(12,25,'Name','conv1')...
,reluLayer('Name','rel_1')...,
maxPooling2dLayer(3,'Stride',2,'Name','pool_1')...
,convolution2dLayer(12,25,'Name','conv_2')...
,reluLayer('Name','relu_2')...,
maxPooling2dLayer(3,'Stride',2,'Name','pool_2')...
,batchNormalizationLayer('Name','bn1')...
,convolution2dLayer(12,25,'Name','conv_3')...
,batchNormalizationLayer('Name','bn2')...
,reluLayer('Name','relu_3')...,
maxPooling2dLayer(3,'Stride',2,'Name','pool_3')...
,fullyConnectedLayer(5,'Name','fc1')...
,softmaxLayer('Name','sm1')...
,classificationLayer('Name','cl_out')];
lgraph=layerGraph(netLayers);
analyzeNetwork(lgraph);
inputSize=[227 227 3];
augmidsTest=augmentedImageDatastore(inputSize(1:2),imdsValidation);
netOptions = trainingOptions("sgdm",'MiniBatchSize',64, ...
'InitialLearnRate',0.001, ...
'MaxEpochs',20, ...
'Plots','training-progress',...
'ValidationData',augmidsTest,...
'ExecutionEnvironment','cpu');%'cpu' for single cpu, 'gpu' for single gpu,'parallel' for multi-cpu/gpu
ASNet= trainNetwork(trainingData, netLayers, netOptions);
featuresTrain=activations(ASNet,imdsTrain,layer,'OutputAs','rows');
featuresTest=activations(ASNet,imdsValidation,layer,'OutputAs','rows');
svmtrain=fitcecoc(featuresTrain,imdsTrain.Labels);
YPred=predict(svmtrain,featuresTest);
plotconfusion(imdsValidation.Labels,YPred);
featuresTrain=double(featuresTrain);
trainingLabels=imdsTrain.Labels;
trainingLabels=cellstr(trainingLabels);
t=templateSVM('standardize',1);
mdl=fitcecoc(double(featuresTrain),cellstr(trainingLabels),'Learners',t,'FitPosterior',1,'ClassNames',{'Mild NPDR','Moderate NPDR','No DR','Proliferative DR','Severe NPDR'});
[cm,order] = confusionmat(imdsValidation.Labels,YPred);
tp=cm(1,1)+cm(2,2)+cm(3,3)+cm(4,4);
tn=cm(5,5);
fp=cm(1,2)+cm(1,3)+cm(1,4)+cm(1,5)+cm(2,1)+cm(2,3)+cm(2,4)+cm(2,5)+cm(3,1)+cm(3,2)+cm(3,4)+cm(3,5)+cm(4,1)+cm(4,2)+cm(4,3)+cm(4,5);
fn=cm(5,1)+cm(5,2)+cm(5,3)+cm(5,4);
precision = tp / (tp + fp);
recall = tp / (tp + fn);
accuracy=(tp+tn)/(tp+fp+fn+tn);
F1 = (2 * precision * recall) / (precision + recall);
predictedLabels=predict(mdl,featuresTest);
[predictedLabels,~,~,Posterior]=predict(mdl,featuresTest);
testLabels=imdsValidation.Labels;
testLabels=cellstr(testLabels);
[X,Y]=perfcurve(testLabels,Posterior(:,4),'No DR');
[A,B]=perfcurve(testLabels,Posterior(:,1),'Mild NPDR');
[C,D]=perfcurve(testLabels,Posterior(:,2),'Moderate NPDR');
[E,F]=perfcurve(testLabels,Posterior(:,5),'Severe NPDR');
[G,H]=perfcurve(testLabels,Posterior(:,3),'Proliferative DR');
p=linspace(0,1);
q=linspace(0,1);
figure
plot(p,q);
hold on;
plot(X,Y,'r-*');
hold on;
plot(A,B,'g-*');
hold on;
plot(C,D,'b-*');
hold on;
plot(E,F,'y-*');
hold on;
plot(G,H,'c-*');
xlabel('FPR');
ylabel('TPR');
legend(categories);
%ASNet=deerNet;
save ASNet;
The code runs perfectly on My Laptop, which has AMD Radeon 530
But, when I try to run the same code in college computer, it shows unexpected launch failure, which uses NVIDIA GT710, error is CUDA_ERROR_LAUNCHED_FAILURE
Please do help me
0 Kommentare
Akzeptierte Antwort
Joss Knight
am 12 Apr. 2022
I see two calls to predict which don't specify ExecutionEnvironment and therefore will be trying to use your college computer's NVIDIA card. Add " ExecutionEnvironment='cpu' " to those calls.
3 Kommentare
Joss Knight
am 13 Apr. 2022
You can't get a CUDA error on a call to trainingOptions since this function doesn't use the GPU. Can you show the complete error including the call stack (everything in red)?
Weitere Antworten (0)
Siehe auch
Kategorien
Mehr zu GPU Computing finden Sie in Help Center und File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!