ln.md

ln

Table of Contents:

Functions

ln_lut

ln_cordic

Examples

Examples

Quality of Results

Error Graph

Resource Usage

Function ln_lut

template <unsigned int W_OUT, int IW_OUT, unsigned int W_IN, int IW_IN>
ap_fixpt<unsigned int W_OUT, int IW_OUT> ln_lut(ap_fixpt<unsigned int W_IN, int IW_IN> num, int error)

Lookup Table based implementation of ln. Uses the lookup implementation of log2. If the input is negative, then an error will occur.

Template Parameters:

• unsigned int W_OUT: width of the output
  
• int IW_OUT: width of integer portion of the output
  
• unsigned int W_IN: width of the input (automatically inferred)
  
• int IW_IN: width of integer portion of the input (automatically inferred)
  
  

Function Arguments:

• ap_fixpt<unsigned int W_IN, int IW_IN> num: input
  
• int error: variable to hold error code value if an error occurs
  

Limitations:

No limitations.

Returns:

• ap_fixpt<unsigned int W_OUT, int IW_OUT>: natural logarithm of the input value

Function ln_cordic

template <unsigned int W_OUT, int IW_OUT, int N_ITERATIONS, unsigned int W_IN, int IW_IN>
ap_fixpt<unsigned int W_OUT, int IW_OUT> ln_cordic(ap_fixpt<unsigned int W_IN, int IW_IN> num, int error)

CORDIC based implementation of ln. Uses the lookup implementation of log2. If the input is negative, then an error will occur.

Template Parameters:

• unsigned int W_OUT: width of the output
  
• int IW_OUT: width of integer portion of the output
  
• int N_ITERATIONS: number of CORDIC iterations
  
• unsigned int W_IN: width of the input (automatically inferred)
  
• int IW_IN: width of integer portion of the input (automatically inferred)
  
  

Function Arguments:

• ap_fixpt<unsigned int W_IN, int IW_IN> num: input
  
• int error: variable to hold error code value if an error occurs
  

Limitations:

No limitations.

Returns:

• ap_fixpt<unsigned int W_OUT, int IW_OUT>: natural logarithm of the input value

Examples

hls::ap_fixpt<10, 2> y = 2;
auto x = hls::math::ln_lut<10, 2>(y); //x will be an ap_fixpt w/ the value 0.69314718055995

hls::ap_fixpt<10, 2> y = 2;
auto x = hls::math::ln_cordic<10, 2, 16>(y); //x will be an ap_fixpt w/ the value 0.69314718055995

The example used to gather the following graph and resource report can be found here.

Error Graph

Resource Usage

Using MPF300

Input Plot Range: [0.01, 1.00]

Using W = 32, IW = 8, Q_M = AP_TRN, O_M = AP_WRAP

Name	Latency [cycles] (min/max/avg)	II [cycles] (min/max/avg)	Avg Error	Max Error	LUTs	DFFs	DSPs	LSRAM	uSRAM	Estimated Frequency
ln_lut	105 / 106 / 105.01	1 / 2 / 1.01	1.9e-05	3.5e-05	2931	5638	5	1	1	368.596 MHz
ln_cordic	58 / 59 / 58.01	1 / 2 / 1.01	2.9e-05	5.8e-05	3168	4141	11	0	4	339.905 MHz
ln_cmath	186 / 186 / 186.00	28 / 28 / 28.00	0	0	16248	22506	9	6	0	300.030 MHz

Input Plot Range: [1.00, 100.00]

Using W = 32, IW = 16, Q_M = AP_TRN, O_M = AP_WRAP

Name	Latency [cycles] (min/max/avg)	II [cycles] (min/max/avg)	Avg Error	Max Error	LUTs	DFFs	DSPs	LSRAM	uSRAM	Estimated Frequency
ln_lut	105 / 106 / 105.01	1 / 2 / 1.01	4e-05	0.000108	2639	4833	5	1	2	350.631 MHz
ln_cordic	55 / 56 / 55.01	1 / 2 / 1.01	0.000162	0.000286	2955	3636	11	0	4	352.237 MHz
ln_cmath	186 / 186 / 186.00	28 / 28 / 28.00	0	0	16248	22506	9	6	0	295.858 MHz

Notes:

• Targeted FMax was 400MHz.
• The standard C math library uses floating point numbers.

Back to top.