0

Non-sequential address warning in dspparamreader

Edward Archibald

August 25, 2023 12:38

None

Follow

I am experimenting with different DSP profiles and I have been successful in creating working profiles thanks the the excellent instructions here.  I am hitting a problem, now, however.  I have two IIR filter banks, both with two channels each.  For some reason I am getting this 'non-sequential address' warning on all four banks.  In looking at the registers, and if I had to guess, it's probably because I changed the filter type from single-precision without Slew to double-precision with Slew.  So the Slew register at the top probably breaks the sequential nature of the registers?  I also suspect that I can work around this as long as I update the code that sets the registers, and which uses offsets, to take into account the slew register?  If this isn't the explanation I stand ready to be corrected.

Here's the output from one of the warnings: 

EQS300MultiDPHWSlewP1Alg4_slewmode
9874

 

EQS300MultiDPHWSlewP1Alg4Targ_B2_1

24734

EQS300MultiDPHWSlewP1Alg4Targ_B2_2

24739

EQS300MultiDPHWSlewP1Alg4Targ_B2_3

24744

EQS300MultiDPHWSlewP1Alg4Targ_B2_4

24749

EQS300MultiDPHWSlewP1Alg4Targ_B2_5

24754

EQS300MultiDPHWSlewP1Alg4Targ_B2_6

24759

EQS300MultiDPHWSlewP1Alg4Targ_B2_7

24764

EQS300MultiDPHWSlewP1Alg4Targ_B2_8

24769

EQS300MultiDPHWSlewP1Alg4Targ_B2_9

24774

EQS300MultiDPHWSlewP1Alg4Targ_B2_10

24779

EQS300MultiDPHWSlewP1Alg4Targ_B2_11

24784

EQS300MultiDPHWSlewP1Alg4Targ_B2_12

24789

EQS300MultiDPHWSlewP1Alg4Targ_B2_13

24794

EQS300MultiDPHWSlewP1Alg4Targ_B2_14

24799

 

EQS300MultiDPHWSlewP1Alg4Targ_B2_15

24804

• Facebook
• Twitter
• LinkedIn

2 comments

• 0
  
  HiFiBerry team August 25, 2023 12:54

  This would require some changes in DSP toolkit. It's simply not supported. If you want to directly manipulate register values, you don't have to put this into the XML file at all.

   

• 0
  
  Edward Archibald August 25, 2023 17:28

  Thanks for the feedback. As it turns out, it was pretty easy to get this to work.  The B2_ registers for the single-precision IIR filters were spaced by a factor of 5.  The single registers generated for the double-precision IIR filters were also spaced by a factor of 5 so I just needed to ensure that the dspreader.js 'knew' about the slew etc. and it generated the right metadata. After that all worked as expected.

  I found an issue with the balance control: at least in my install it always left some audio going to the left channel.  I resolved it by doing a simple rework of the balance control. Attached image has my changes. AB in CD out components were not needed and, according to my research, may have had some issues with reducing headroom.