diff --git a/DPDPU/NetworkEngine/Source/PEPOTLDKTCP.c b/DPDPU/NetworkEngine/Source/PEPOTLDKTCP.c
index e71d0d0..295be7b 100644
--- a/DPDPU/NetworkEngine/Source/PEPOTLDKTCP.c
+++ b/DPDPU/NetworkEngine/Source/PEPOTLDKTCP.c
@@ -2857,6 +2857,10 @@ NetfeRxTxProcess(
             offsetInMsg = 0;
             totalBytesOfDPUReqs = 0;
 
+            RequestIdT firstIndex = FeLcore->ReadOpsTailPointer;
+            DataPlaneRequestContext* firstCtxt = &(FeLcore->ReadOps[firstIndex].ReadCtxt);
+            printf("DPU offlaoding firstIndex: %hu\n", firstIndex);
+
             for (i = 0; i != FeLcore->NumReqsForDPU; i++) {
                 ReadOpDescriptorT* readOp = &FeLcore->ReadOps[FeLcore->ReadOpsTailPointer];
                 readOp->ReadResp.Result = DDS_ERROR_CODE_IO_PENDING;
@@ -2888,18 +2892,27 @@ NetfeRxTxProcess(
 #endif
                 readOp->StreamCtxt = (ContextT)FeStream;
 
-                SubmitDataPlaneRequest(
+                // now batching, don't submit one by one
+                /* SubmitDataPlaneRequest(
                     FS,
                     &readOp->ReadCtxt,
                     true,
                     DDS_DPU_IO_SLOT_NUMBER_BASE + FeLcore->ReadOpsTailPointer
-                );
+                ); */
+                // `SubmitDataPlaneRequest` signature has changed
+
+                readOp->ReadCtxt.IsRead = 1;
 
                 FeLcore->ReadOpsTailPointer++;
                 if (FeLcore->ReadOpsTailPointer == DDS_MAX_OUTSTANDING_IO) {
                     FeLcore->ReadOpsTailPointer = 0;
                 }
             }
+            //
+            // Batched SubmitDataPlaneRequest() call here
+            //
+            //
+            SubmitDataPlaneRequest(FS, firstCtxt, firstIndex, n);
 
             if (FeLcore->NumReqsForHost) {
                 //