Forum Rule: Always post complete source code & details to reproduce any issue!
Tab Content
  • Bob Larkin's Avatar
    05-19-2020, 07:11 PM
    Thanks, Frank B. That 3 cycles (T3.6 Cortex-M4) shows that the ARM FIR routine is getting about all there is! For the T4.0 some combination of hardware, programmer and compiler has caught the fixed point times up...
    15 replies | 248 view(s)
  • Bob Larkin's Avatar
    05-19-2020, 07:01 PM
    I left out T4.0 fixed point. It is 0.41 microseconds per FIR tap, essentially the same as for Floating Point. Dazzling for both...
    15 replies | 248 view(s)
  • Bob Larkin's Avatar
    05-19-2020, 06:44 PM
    Going back to the processor load question, I looked at the T4.0 F32 numbers and they are interesting. First though, one should not try to generalize from these measurements. These numbers are based on measuring the...
    15 replies | 248 view(s)
  • Bob Larkin's Avatar
    05-19-2020, 03:53 AM
    Hi Frank - Yes, I get it now! Fundamentally, the AudioInputI2S_F32 and AudioOutputI2S_F32 are not converted yet to handle T4.x. I had done all my testing on T3.6. I was able to get it to compile and run with a T4.0...
    15 replies | 248 view(s)
  • Bob Larkin's Avatar
    05-18-2020, 04:32 PM
    Hi Frank - I just re-checked with, I believe, the setup you describe. The AudioFilterEqualizer_F32 compiled and ran. I believe my OpenAudio_Arduino is per Chip's...
    15 replies | 248 view(s)
  • Bob Larkin's Avatar
    05-17-2020, 11:48 PM
    Right at the bottom of the test code, I asked for a type conversion of 0.0f to int. idout = 0.0f; should have been idout = 0; The compiler knew what I meant, and so the answers didn't change when I fixed it.
    15 replies | 248 view(s)
  • Bob Larkin's Avatar
    05-17-2020, 11:36 PM
    Frank B here are microseconds, for F32 and Q15, for a 128 coefficient FIR with 128 updates. I played with this (see INO below) and it seems to be consistent and probably correct. F32 ARM FIR microseconds = 326...
    15 replies | 248 view(s)
  • Bob Larkin's Avatar
    05-17-2020, 06:40 PM
    Thanks, Frank. And as always, great comments. Thanks for running it on the T4. It was on my list, but I wanted to be sure that 3.6 was OK. It would be very useful and interesting to have a convolution version. I...
    15 replies | 248 view(s)
  • Bob Larkin's Avatar
    05-17-2020, 04:50 PM
    Also, here is a frequency response for a octave band equalizer.
    15 replies | 248 view(s)
  • Bob Larkin's Avatar
    05-17-2020, 04:41 PM
    I have been putting together some communications object blocks with floating point data interfaces. This format of Audio objects has been discussed on this forum...
    15 replies | 248 view(s)
No More Results
About Bob Larkin

Basic Information

Statistics


Total Posts
Total Posts
113
Posts Per Day
0.09
Last Post
Audio Equalizer using FIR 05-19-2020 07:11 PM
General Information
Last Activity
Today 06:22 PM
Join Date
12-11-2016