Different output when choosing optimised or ansi function for FIR filtering. (DSP-100)

[herculanodavi]

Environment

• Development Kit: ESP32-S3-DevKitC-1
• Kit version (for WroverKit/PicoKit/DevKitC): v1.0
• Module or chip used: ESP32-32-WROOM-1U
• IDF version (run git describe --tags to find it): v4.4.3
• Build System: CMake
• Compiler version (run xtensa-esp32-elf-gcc --version to find it): 8.4.0
• Operating System: macOS
• Power Supply: USB

Problem Description

The dsps_fird_s16_aes3 function does not show the same results as the dsps_fird_s16_ansi given the same parameters.

Expected Behavior

This code

fir_s16_t handle;
alignas(16) int16_t coeffs[1] = {1};
alignas(16) int16_t delay[1] = {0};
int16_t coeffs_len = 1;
int16_t decimation = 2;
dsps_fird_init_s16(&handle, coeffs, delay, coeffs_len, decimation, 0, 15);

alignas(16) int16_t input[10] = {0, 1, 2, 3, 4, 5, 6, 7, 8, 9};
alignas(16) int16_t output[5] = {0, 0, 0, 0, 0};
dsps_fird_s16(&fir_handle.handle, input, output, 5);

should have the result {1, 3, 5, 7, 9} independently of the flag CONFIG_DSP_OPTIMIZED.

Actual Behavior

When CONFIG_DSP_OPTIMIZED is enabled, the function chosen when calling dsps_fird_s16 is dsps_fird_s16_aes3, and the resulting output is `

Steps to repropduce

Run the code below.

Code to reproduce this issue

int16_t coeffs_len = 1;
int16_t decimation = 2;

fir_s16_t handle_ansi;
alignas(16) int16_t delay_ansi[1] = {0};

fir_s16_t handle_s3;
alignas(16) int16_t delay_s3[1] = {0};

alignas(16) int16_t coeffs[1] = {1};

dsps_fird_init_s16(&handle_ansi, coeffs, delay, coeffs_len, decimation, 0, 15);
dsps_fird_init_s16(&handle_s3, coeffs, delay, coeffs_len, decimation, 0, 15);

alignas(16) int16_t input[10] = {0, 1, 2, 3, 4, 5, 6, 7, 8, 9};
alignas(16) int16_t output_ansi[5] = {0, 0, 0, 0, 0};
alignas(16) int16_t output_s3[5] = {0, 0, 0, 0, 0};

dsps_fird_s16_ansi(&fir_handle.handle, input, output_ansi, 5);
dsps_fird_s16_aes3(&fir_handle.handle, input, output_s3, 5);

for(int i = 0; i < 5; i++) {
  assert(output_ansi[i] == output_s3[i]);
}

[dmitry1945]

@herculanodavi thank you for the issue, we will look.

[dmitry1945]

Hi @herculanodavi,

We have fix the issue in the last releases. Please take a look.

Thank you, Dmitry