diff --git a/opencl/test/unit_test/command_stream/command_stream_receiver_flush_task_3_tests.cpp b/opencl/test/unit_test/command_stream/command_stream_receiver_flush_task_3_tests.cpp
index 9814368163..eca0c6c641 100644
--- a/opencl/test/unit_test/command_stream/command_stream_receiver_flush_task_3_tests.cpp
+++ b/opencl/test/unit_test/command_stream/command_stream_receiver_flush_task_3_tests.cpp
@@ -147,6 +147,19 @@ HWTEST_F(CommandStreamReceiverFlushTaskTests, givenCsrInBatchingModeAndTwoRecord
     EXPECT_EQ(mockCsr->recordedCommandBuffer->batchBuffer.endCmdPtr, lastbbEndPtr);
 }
 
+HWTEST_F(CommandStreamReceiverFlushTaskTests, whenFlushSmallTaskThenCommandStreamAlignedToCacheLine) {
+    using MI_BATCH_BUFFER_END = typename FamilyType::MI_BATCH_BUFFER_END;
+    auto &csr = pDevice->getUltCommandStreamReceiver<FamilyType>();
+
+    auto &stream = csr.getCS(2 * MemoryConstants::cacheLineSize);
+    stream.getSpace(MemoryConstants::cacheLineSize - sizeof(MI_BATCH_BUFFER_END) - 2);
+    csr.flushSmallTask(stream, stream.getUsed());
+
+    auto used = csr.commandStream.getUsed();
+    auto expected = 2 * MemoryConstants::cacheLineSize;
+    EXPECT_EQ(used, expected);
+}
+
 HWTEST_F(CommandStreamReceiverFlushTaskTests, givenCsrInBatchingModeAndThreeRecordedCommandBuffersWhenFlushTaskIsCalledThenBatchBuffersAreCombined) {
 
     typedef typename FamilyType::MI_BATCH_BUFFER_END MI_BATCH_BUFFER_END;
diff --git a/shared/source/command_stream/command_stream_receiver_hw_base.inl b/shared/source/command_stream/command_stream_receiver_hw_base.inl
index ba9eecb7da..7dbaee6fc2 100644
--- a/shared/source/command_stream/command_stream_receiver_hw_base.inl
+++ b/shared/source/command_stream/command_stream_receiver_hw_base.inl
@@ -1284,6 +1284,7 @@ void CommandStreamReceiverHw<GfxFamily>::flushSemaphoreWait(GraphicsAllocation *
 
 template <typename GfxFamily>
 void CommandStreamReceiverHw<GfxFamily>::flushSmallTask(LinearStream &commandStreamTask, size_t commandStreamStartTask) {
+    using MI_BATCH_BUFFER_START = typename GfxFamily::MI_BATCH_BUFFER_START;
     using MI_BATCH_BUFFER_END = typename GfxFamily::MI_BATCH_BUFFER_END;
 
     void *endingCmdPtr = nullptr;
@@ -1298,6 +1299,8 @@ void CommandStreamReceiverHw<GfxFamily>::flushSmallTask(LinearStream &commandStr
         *batchBufferEnd = GfxFamily::cmdInitBatchBufferEnd;
     }
 
+    auto bytesToPad = sizeof(MI_BATCH_BUFFER_START) - sizeof(MI_BATCH_BUFFER_END);
+    EncodeNoop<GfxFamily>::emitNoop(commandStreamTask, bytesToPad);
     EncodeNoop<GfxFamily>::alignToCacheLine(commandStreamTask);
 
     if (globalFenceAllocation) {