Relation between out of order command queue and blocking and non-blocking OpenCL Write

Question

I am trying to figure out the relationship between the clcreatecommandqueue and clenqueuewritebuffer API.

I am using a very simple example i.e. Vector addition and AMD CodeCL to gauge the performance.

Here are some numbers with different configurations for clcreatecommandqueue and clenqueuewritebuffer API:

Scenario      Command-Queue Properties      blocking_write       time taken

1             in-order                      non-blocking         46.83ms

2             in-order                      blocking             4.711ms

3             out-of-order                  non-blocking         46.55ms

4             out-of-order                  blocking             4.358ms

The recorded time taken is for 4MB data transfer from cpu-to-gpu (I got similar results for a 40MB data transfer).

From my understanding, the least time taken should be for out-of-order and non-blocking and the most time taken should be for in-order and blocking write.

But the above data shows that whatever the type of command queue, the blocking write outperforms everything else.

Can anybody help me out in understanding these stats?