IVGCVSW-6806 Add Unidirectional Sequence Lstm support to Neon

 * Corrected TensorInfo order for IsUnidirectionalSequenceLstmSupported
    * outputStateOut TensorInfo is not optional.
    * cellStateOut TensorInfo is not optional.
    * TensorInfo Order matches other QLSTM/LSTM layers.
 * Added missing parameters to UnidirectionalSequenceLstmOperator for
   delegate.
 * Added quantized UnidirectionalSequenceLstm support to Neon

!android-nn-driver:7457

Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: I26dde1bb96793dd25eb9081ca5ae5f63752288c4

1 files changed, 28 insertions, 12 deletions

diff --git a/src/armnnSerializer/test/LstmSerializationTests.cpp b/src/armnnSerializer/test/LstmSerializationTests.cpp
index d8f8967bcd..ae2d813fc0 100644
--- a/src/armnnSerializer/test/LstmSerializationTests.cpp
+++ b/src/armnnSerializer/test/LstmSerializationTests.cpp
@@ -2299,6 +2299,8 @@ TEST_CASE("SerializeDeserializeUnidirectionalSequenceLstmCifgPeepholeNoProjectio
     armnn::TensorInfo inputTensorInfo({ batchSize, timeSize, inputSize }, armnn::DataType::Float32);
     armnn::TensorInfo cellStateTensorInfo({ batchSize, numUnits}, armnn::DataType::Float32);
     armnn::TensorInfo outputStateTensorInfo({ batchSize, outputSize }, armnn::DataType::Float32);
+    armnn::TensorInfo outputStateOutTensorInfo({ batchSize, timeSize, outputSize }, armnn::DataType::Float32);
+    armnn::TensorInfo cellStateOutTensorInfo({ batchSize, outputSize }, armnn::DataType::Float32);
     armnn::TensorInfo outputTensorInfo({ batchSize, timeSize, outputSize }, armnn::DataType::Float32);
 
     inputLayer->GetOutputSlot(0).Connect(unidirectionalSequenceLstmLayer->GetInputSlot(0));
@@ -2310,8 +2312,10 @@ TEST_CASE("SerializeDeserializeUnidirectionalSequenceLstmCifgPeepholeNoProjectio
     cellStateIn->GetOutputSlot(0).Connect(unidirectionalSequenceLstmLayer->GetInputSlot(2));
     cellStateIn->GetOutputSlot(0).SetTensorInfo(cellStateTensorInfo);
 
-    unidirectionalSequenceLstmLayer->GetOutputSlot(0).Connect(outputLayer->GetInputSlot(0));
-    unidirectionalSequenceLstmLayer->GetOutputSlot(0).SetTensorInfo(outputTensorInfo);
+    unidirectionalSequenceLstmLayer->GetOutputSlot(0).SetTensorInfo(outputStateOutTensorInfo);
+    unidirectionalSequenceLstmLayer->GetOutputSlot(1).SetTensorInfo(cellStateOutTensorInfo);
+    unidirectionalSequenceLstmLayer->GetOutputSlot(2).Connect(outputLayer->GetInputSlot(0));
+    unidirectionalSequenceLstmLayer->GetOutputSlot(2).SetTensorInfo(outputTensorInfo);
 
     armnn::INetworkPtr deserializedNetwork = DeserializeNetwork(SerializeNetwork(*network));
     CHECK(deserializedNetwork);
@@ -2319,7 +2323,7 @@ TEST_CASE("SerializeDeserializeUnidirectionalSequenceLstmCifgPeepholeNoProjectio
     VerifyLstmLayer<armnn::UnidirectionalSequenceLstmDescriptor> checker(
         layerName,
         {inputTensorInfo, outputStateTensorInfo, cellStateTensorInfo},
-        {outputTensorInfo},
+        {outputStateOutTensorInfo, cellStateOutTensorInfo, outputTensorInfo},
         descriptor,
         params);
     deserializedNetwork->ExecuteStrategy(checker);
@@ -2436,6 +2440,8 @@ TEST_CASE("SerializeDeserializeUnidirectionalSequenceLstmNoCifgWithPeepholeAndPr
     armnn::TensorInfo inputTensorInfo({ batchSize, timeSize, inputSize }, armnn::DataType::Float32);
     armnn::TensorInfo cellStateTensorInfo({ batchSize, numUnits}, armnn::DataType::Float32);
     armnn::TensorInfo outputStateTensorInfo({ batchSize, outputSize }, armnn::DataType::Float32);
+    armnn::TensorInfo outputStateOutTensorInfo({ batchSize, timeSize, outputSize }, armnn::DataType::Float32);
+    armnn::TensorInfo cellStateOutTensorInfo({ batchSize, outputSize }, armnn::DataType::Float32);
     armnn::TensorInfo outputTensorInfo({ batchSize, timeSize, outputSize }, armnn::DataType::Float32);
 
     inputLayer->GetOutputSlot(0).Connect(unidirectionalSequenceLstmLayer->GetInputSlot(0));
@@ -2447,8 +2453,10 @@ TEST_CASE("SerializeDeserializeUnidirectionalSequenceLstmNoCifgWithPeepholeAndPr
     cellStateIn->GetOutputSlot(0).Connect(unidirectionalSequenceLstmLayer->GetInputSlot(2));
     cellStateIn->GetOutputSlot(0).SetTensorInfo(cellStateTensorInfo);
 
-    unidirectionalSequenceLstmLayer->GetOutputSlot(0).Connect(outputLayer->GetInputSlot(0));
-    unidirectionalSequenceLstmLayer->GetOutputSlot(0).SetTensorInfo(outputTensorInfo);
+    unidirectionalSequenceLstmLayer->GetOutputSlot(0).SetTensorInfo(outputStateOutTensorInfo);
+    unidirectionalSequenceLstmLayer->GetOutputSlot(1).SetTensorInfo(cellStateOutTensorInfo);
+    unidirectionalSequenceLstmLayer->GetOutputSlot(2).Connect(outputLayer->GetInputSlot(0));
+    unidirectionalSequenceLstmLayer->GetOutputSlot(2).SetTensorInfo(outputTensorInfo);
 
     armnn::INetworkPtr deserializedNetwork = DeserializeNetwork(SerializeNetwork(*network));
     CHECK(deserializedNetwork);
@@ -2456,7 +2464,7 @@ TEST_CASE("SerializeDeserializeUnidirectionalSequenceLstmNoCifgWithPeepholeAndPr
     VerifyLstmLayer<armnn::UnidirectionalSequenceLstmDescriptor> checker(
         layerName,
         {inputTensorInfo, outputStateTensorInfo, cellStateTensorInfo},
-        {outputTensorInfo},
+        {outputStateOutTensorInfo, cellStateOutTensorInfo, outputTensorInfo},
         descriptor,
         params);
     deserializedNetwork->ExecuteStrategy(checker);
@@ -2592,6 +2600,8 @@ TEST_CASE("SerializeDeserializeUnidirectionalSequenceLstmNoCifgWithPeepholeWithP
     armnn::TensorInfo inputTensorInfo({ batchSize, timeSize, inputSize }, armnn::DataType::Float32);
     armnn::TensorInfo cellStateTensorInfo({ batchSize, numUnits}, armnn::DataType::Float32);
     armnn::TensorInfo outputStateTensorInfo({ batchSize, outputSize }, armnn::DataType::Float32);
+    armnn::TensorInfo outputStateOutTensorInfo({batchSize, timeSize, outputSize}, armnn::DataType::Float32);
+    armnn::TensorInfo cellStateOutTensorInfo({batchSize, timeSize, outputSize}, armnn::DataType::Float32);
     armnn::TensorInfo outputTensorInfo({ batchSize, timeSize, outputSize }, armnn::DataType::Float32);
 
     inputLayer->GetOutputSlot(0).Connect(unidirectionalSequenceLstmLayer->GetInputSlot(0));
@@ -2603,8 +2613,10 @@ TEST_CASE("SerializeDeserializeUnidirectionalSequenceLstmNoCifgWithPeepholeWithP
     cellStateIn->GetOutputSlot(0).Connect(unidirectionalSequenceLstmLayer->GetInputSlot(2));
     cellStateIn->GetOutputSlot(0).SetTensorInfo(cellStateTensorInfo);
 
-    unidirectionalSequenceLstmLayer->GetOutputSlot(0).Connect(outputLayer->GetInputSlot(0));
-    unidirectionalSequenceLstmLayer->GetOutputSlot(0).SetTensorInfo(outputTensorInfo);
+    unidirectionalSequenceLstmLayer->GetOutputSlot(0).SetTensorInfo(outputStateOutTensorInfo);
+    unidirectionalSequenceLstmLayer->GetOutputSlot(1).SetTensorInfo(cellStateOutTensorInfo);
+    unidirectionalSequenceLstmLayer->GetOutputSlot(2).Connect(outputLayer->GetInputSlot(0));
+    unidirectionalSequenceLstmLayer->GetOutputSlot(2).SetTensorInfo(outputTensorInfo);
 
     armnn::INetworkPtr deserializedNetwork = DeserializeNetwork(SerializeNetwork(*network));
     CHECK(deserializedNetwork);
@@ -2612,7 +2624,7 @@ TEST_CASE("SerializeDeserializeUnidirectionalSequenceLstmNoCifgWithPeepholeWithP
     VerifyLstmLayer<armnn::UnidirectionalSequenceLstmDescriptor> checker(
             layerName,
             {inputTensorInfo, outputStateTensorInfo, cellStateTensorInfo},
-            {outputTensorInfo},
+            {outputStateOutTensorInfo, cellStateOutTensorInfo, outputTensorInfo},
             descriptor,
             params);
     deserializedNetwork->ExecuteStrategy(checker);
@@ -2697,6 +2709,8 @@ TEST_CASE("SerializeDeserializeUnidirectionalSequenceLstmCifgPeepholeNoProjectio
     armnn::TensorInfo inputTensorInfo({ timeSize, batchSize, inputSize }, armnn::DataType::Float32);
     armnn::TensorInfo cellStateTensorInfo({ batchSize, numUnits}, armnn::DataType::Float32);
     armnn::TensorInfo outputStateTensorInfo({ batchSize, outputSize }, armnn::DataType::Float32);
+    armnn::TensorInfo outputStateOutTensorInfo({batchSize, timeSize, outputSize}, armnn::DataType::Float32);
+    armnn::TensorInfo cellStateOutTensorInfo({batchSize, timeSize, outputSize}, armnn::DataType::Float32);
     armnn::TensorInfo outputTensorInfo({ timeSize, batchSize, outputSize }, armnn::DataType::Float32);
 
     inputLayer->GetOutputSlot(0).Connect(unidirectionalSequenceLstmLayer->GetInputSlot(0));
@@ -2708,8 +2722,10 @@ TEST_CASE("SerializeDeserializeUnidirectionalSequenceLstmCifgPeepholeNoProjectio
     cellStateIn->GetOutputSlot(0).Connect(unidirectionalSequenceLstmLayer->GetInputSlot(2));
     cellStateIn->GetOutputSlot(0).SetTensorInfo(cellStateTensorInfo);
 
-    unidirectionalSequenceLstmLayer->GetOutputSlot(0).Connect(outputLayer->GetInputSlot(0));
-    unidirectionalSequenceLstmLayer->GetOutputSlot(0).SetTensorInfo(outputTensorInfo);
+    unidirectionalSequenceLstmLayer->GetOutputSlot(0).SetTensorInfo(outputStateOutTensorInfo);
+    unidirectionalSequenceLstmLayer->GetOutputSlot(1).SetTensorInfo(cellStateOutTensorInfo);
+    unidirectionalSequenceLstmLayer->GetOutputSlot(2).Connect(outputLayer->GetInputSlot(0));
+    unidirectionalSequenceLstmLayer->GetOutputSlot(2).SetTensorInfo(outputTensorInfo);
 
     armnn::INetworkPtr deserializedNetwork = DeserializeNetwork(SerializeNetwork(*network));
     CHECK(deserializedNetwork);
@@ -2717,7 +2733,7 @@ TEST_CASE("SerializeDeserializeUnidirectionalSequenceLstmCifgPeepholeNoProjectio
     VerifyLstmLayer<armnn::UnidirectionalSequenceLstmDescriptor> checker(
         layerName,
         {inputTensorInfo, outputStateTensorInfo, cellStateTensorInfo},
-        {outputTensorInfo},
+        {outputStateOutTensorInfo, cellStateOutTensorInfo, outputTensorInfo},
         descriptor,
         params);
     deserializedNetwork->ExecuteStrategy(checker);