Skip to content

Latest commit

 

History

History
90 lines (71 loc) · 2.64 KB

CHANGELOG.rst

File metadata and controls

90 lines (71 loc) · 2.64 KB

xCORE-200 DSP library change log

4.0.0

  • Removed synchronous sample rate conversion functions - now maintained in lib_src
  • Fixed bug in dsp_vector_mulv_addv()
  • Faster bit reverse and inverse FFT
  • Added real FFT
  • Added real reverse FFT
  • Added FFT with top half blanked
  • Logistics functions
  • Block floating point functions
  • Complex FIR
  • Complex vector arithmetic, with optional scaling
  • Added in-place complex vector scaling
  • Added complex vector magnitude
  • Added complex vector scaling with arithmetic shift
  • Added complex negative multiply and accumulate

3.1.0

  • Deprecated synchronous sample rate conversion functions - now maintained in lib_src
  • Added functions to compute a fast fixed point atan2 and hypotenuse
  • Added Q8 versions of the arc sine and arc cosine functions
  • Added 16384 point sine table
  • Improved performance of the forwards FFT function, with small reduction in memory footprint

3.0.0

  • Added exponential and natural logarithm functions
  • Added Hyperbolic sine and cosine
  • Fixed Matrix Multiplication and improved performance
  • Changed API prefix from lib_dsp_ to dsp_.
  • Changed lib_dsp_fft_complex_t to dsp_complex_t and lib_dsp_fft_complex_short_t to dsp_complex_short_t
  • Various fixes in API documentation
  • Added complex vector multiplication
  • Added synchronous sample rate conversion (downsample or upsample by factor 3)

2.0.0

  • FFT interface update. Consolidated interface and improved testing.
  • Halved the memory for processing two real signals with a single complex FFT.
  • Renamed *_transforms to *_fft to improve naming consistency
  • Improved performance and accuracy of dsp_math_sqrt. Error is <= 1. Worst case performance is 96 cycles.
  • int32_t and uint32_t now used more consistently.

1.0.4

  • Added fixed point sine and cosine functions. Performance: 62 cycles for dsp_math_sin, 64 cycles for dsp_math_cos.
  • Brute force testing of all input values proved accuracy to within one LSB (error is <= 1)
  • Added short int complex and tworeals FFT and iFFT
  • Improved Macros for converting from double to int and int to double.
  • Added optimised fixed point atan function dsp_math_atan
  • Most tests in math_app.xc are now self-checking. Improved error reporting.
  • Option for performance measurements in 10ns cycles.

1.0.3

  • Update to source code license and copyright

1.0.2

  • FFT and inverse FFT for two complex short int signals

1.0.1

  • FFT and inverse FFT for complex signals or two real signals.

1.0.0

  • Initial version