Hello,
Currently I am developing on 2 custom boards with a identical layout. The difference is the used processor.
One board has a iMX6S and the other board a iMX6Q processor. Both boards have the same DDR configuration so the chips (MT41K128M16JT-125:K) and layout (32-bit) are identical. The DDR on the boards are independently calibrated.
However when running "bw_mem" (part of lmbench) I get very different results as the table below shows.
The 'bzero' test for instance is 6x slower on the Quadcore then on the Solo.
Command | iMX6 Solo | iMX6 Quadcore |
---|---|---|
bw_mem -P 1 -N 4 128M rd | 1417.69 | 1318.39 |
bw_mem -P 1 -N 4 128M wr | 467.34 | 373.44 |
bw_mem -P 1 -N 4 128M rdwr | 486.82 | 397.27 |
bw_mem -P 1 -N 4 128M cp | 460.50 | 342.82 |
bw_mem -P 1 -N 4 128M fwr | 2178.06 | 370.24 |
bw_mem -P 1 -N 4 128M frd | 1158.55 | 622.61 |
bw_mem -P 1 -N 4 128M fcp | 768.31 | 317.41 |
bw_mem -P 1 -N 4 128M bzero | 2175.32 | 339.56 |
bw_mem -P 1 -N 4 128M bcopy | 770.06 | 355.44 |
Does anyone know why there is such a big difference between the Solo and Quadcore with identical DDR configuration?
Hi Dennie
one can use profile tool for memory bandwidth analysis
About MMDC Profiling Tool for i.MX6
Also difference may be caused by different NIC-301 settings
described in Chapter 45 Network Interconnect Bus System (NIC-301)
i.MX6DQ Reference Manual.
Best regards
igor
-----------------------------------------------------------------------------------------------------------------------
Note: If this post answers your question, please click the Correct Answer button. Thank you!
-----------------------------------------------------------------------------------------------------------------------