diff --git a/src/plugins/intel_gpu/src/graph/impls/onednn/primitive_onednn_base.h b/src/plugins/intel_gpu/src/graph/impls/onednn/primitive_onednn_base.h
index 655b10933e5..653b24bfef5 100644
--- a/src/plugins/intel_gpu/src/graph/impls/onednn/primitive_onednn_base.h
+++ b/src/plugins/intel_gpu/src/graph/impls/onednn/primitive_onednn_base.h
@@ -156,13 +156,13 @@ protected:
 
         {
             auto& input = instance.input_memory(0);
-            auto offset = onednn::get_offset(_pd.dnnl::primitive_desc_base::src_desc(0));
+            auto offset = onednn::get_f_offset(instance.node.input().get_output_layout(), _pd.dnnl::primitive_desc_base::src_desc(0));
             args.insert({DNNL_ARG_SRC, input.get_onednn_memory(_pd.dnnl::primitive_desc_base::src_desc(0), offset)});
         }
 
         {
             auto& output = instance.output_memory();
-            auto offset = onednn::get_offset(_pd.dnnl::primitive_desc_base::dst_desc(0));
+            auto offset = onednn::get_f_offset(instance.node.get_output_layout(), _pd.dnnl::primitive_desc_base::dst_desc(0));
             args.insert({DNNL_ARG_DST, output.get_onednn_memory(_pd.dnnl::primitive_desc_base::dst_desc(0), offset)});
         }
 
diff --git a/src/plugins/intel_gpu/src/graph/impls/onednn/utils.cpp b/src/plugins/intel_gpu/src/graph/impls/onednn/utils.cpp
index ae4aa9320ab..1ad7a92aea9 100644
--- a/src/plugins/intel_gpu/src/graph/impls/onednn/utils.cpp
+++ b/src/plugins/intel_gpu/src/graph/impls/onednn/utils.cpp
@@ -131,23 +131,16 @@ void combine_bf_with_first_spatial_dim(cldnn::layout& l) {
     l.size.spatial[last_spatial_dim_idx] = 1;
 }
 
-int64_t get_offset(dnnl::memory::desc desc) {
+int64_t get_f_offset(cldnn::layout l, dnnl::memory::desc desc) {
     int64_t offset = 0;
-    int32_t padded_idx = -1;
-    for (int32_t i = 0; i < DNNL_MAX_NDIMS; ++i) {
-        if (desc.data.padded_offsets[i] > 0) {
-            padded_idx = i;
-            break;
-        }
-    }
-    if (padded_idx > -1) {
-        if (padded_idx != 1)
-            throw std::runtime_error(std::string("onednn only support feature padding. Unsupported padded_idx: ") + std::to_string(padded_idx));
-        offset = desc.data.padded_offsets[padded_idx];
-        for (int32_t i = padded_idx + 1; i < desc.data.ndims; ++i) {
-            offset *= desc.data.padded_dims[i];
+    auto f_padding = l.data_padding.lower_size().feature[0];
+    if (f_padding != 0) {
+        offset = f_padding;
+        for (size_t i = 0; i < l.size.spatial.size(); ++i) {
+            offset *= l.size.spatial[i];
         }
     }
+
     switch (desc.data.data_type) {
         case dnnl_data_type_t::dnnl_s8:
         case dnnl_data_type_t::dnnl_u8:
@@ -179,11 +172,8 @@ dnnl::memory::desc layout_to_memory_desc(cldnn::layout l, dnnl::memory::format_t
         padded_dims = dims;
     } else {
         auto rank = cldnn::format::dimension(l.format);
-        auto padded_size = l.size + l.data_padding.lower_size() + l.data_padding.upper_size();
-        auto offset = l.data_padding.lower_size();
         dims = convert_tensor(l.size, rank, cldnn::format::is_grouped(l.format));
-        padded_dims = convert_tensor(padded_size, rank);
-        padded_offset = convert_tensor(offset, rank);
+        padded_dims = dims;
     }
 
     pad_dims(padded_dims, l.format);
diff --git a/src/plugins/intel_gpu/src/graph/impls/onednn/utils.hpp b/src/plugins/intel_gpu/src/graph/impls/onednn/utils.hpp
index dcc3a2694f5..74dfe145362 100644
--- a/src/plugins/intel_gpu/src/graph/impls/onednn/utils.hpp
+++ b/src/plugins/intel_gpu/src/graph/impls/onednn/utils.hpp
@@ -33,7 +33,7 @@ dnnl::memory::desc layout_to_memory_desc(cldnn::layout l, dnnl::memory::format_t
 dnnl::algorithm convert_activation_func(cldnn::activation_func func);
 cldnn::format find_format(dnnl::memory::desc desc, bool is_grouped = false);
 
-int64_t get_offset(dnnl::memory::desc desc);
+int64_t get_f_offset(cldnn::layout l, dnnl::memory::desc desc);
 
 // If the values in the tensor are identical, make it as per-tensor value
 template <typename T>