diff --git a/inference-engine/src/mkldnn_plugin/nodes/mkldnn_convert_node.cpp b/inference-engine/src/mkldnn_plugin/nodes/mkldnn_convert_node.cpp
index b34ca0761b5fab..fc7a034693283f 100644
--- a/inference-engine/src/mkldnn_plugin/nodes/mkldnn_convert_node.cpp
+++ b/inference-engine/src/mkldnn_plugin/nodes/mkldnn_convert_node.cpp
@@ -96,8 +96,7 @@ void MKLDNNConvertNode::createPrimitive() {
 }
 
 static inline uint8_t* getDataPtr(const MKLDNNMemory& memoryPtr) {
-    return reinterpret_cast<uint8_t*>(memoryPtr.GetData()) + memoryPtr.GetDescriptor().data.layout_desc.blocking.offset_padding *
-        MKLDNNExtensionUtils::sizeOfDataType(mkldnn::memory::data_type(memoryPtr.GetDescriptor().data.data_type));
+    return reinterpret_cast<uint8_t*>(memoryPtr.GetData()) + memoryPtr.GetDescriptor().data.layout_desc.blocking.offset_padding;
 }
 
 void MKLDNNConvertNode::execute(mkldnn::stream strm) {
diff --git a/inference-engine/src/mkldnn_plugin/nodes/mkldnn_split_node.cpp b/inference-engine/src/mkldnn_plugin/nodes/mkldnn_split_node.cpp
index 671223e813ba03..b90c467a5229ad 100644
--- a/inference-engine/src/mkldnn_plugin/nodes/mkldnn_split_node.cpp
+++ b/inference-engine/src/mkldnn_plugin/nodes/mkldnn_split_node.cpp
@@ -59,8 +59,7 @@ static TensorDesc makeChannelBlockedTensorDesc(const Precision& precision, const
 }
 
 static inline uint8_t* getDataPtr(const MKLDNNMemory& memoryPtr) {
-    return reinterpret_cast<uint8_t*>(memoryPtr.GetData()) + memoryPtr.GetDescriptor().data.layout_desc.blocking.offset_padding *
-        MKLDNNExtensionUtils::sizeOfDataType(mkldnn::memory::data_type(memoryPtr.GetDescriptor().data.data_type));
+    return reinterpret_cast<uint8_t*>(memoryPtr.GetData()) + memoryPtr.GetDescriptor().data.layout_desc.blocking.offset_padding;
 }
 
 MKLDNNSplitNode::MKLDNNSplitNode(const InferenceEngine::CNNLayerPtr& layer, const mkldnn::engine& eng, MKLDNNWeightsSharing::Ptr &cache) :