dsp - CMSIS DSP Library from CMSIS 2.0. See http://www.…

Users » simon » Code » dsp

CMSIS DSP Library from CMSIS 2.0. See http://www.onarm.com/cmsis/ for full details

Dependents: K22F_DSP_Matrix_least_square BNO055-ELEC3810 1BNO055 ECE4180Project--Slave2 ... more

src/Cortex-M4-M3/FilteringFunctions/arm_biquad_cascade_df1_fast_q31.c@0:1014af42efd9, 2011-03-10 (annotated)

Committer:: simon
Date:: Thu Mar 10 15:07:50 2011 +0000
Revision:: 0:1014af42efd9

Who changed what in which revision?

User	Revision	Line number	New contents of line
simon	0:1014af42efd9	1	/* ----------------------------------------------------------------------
simon	0:1014af42efd9	2	* Copyright (C) 2010 ARM Limited. All rights reserved.
simon	0:1014af42efd9	3	*
simon	0:1014af42efd9	4	* $Date: 29. November 2010
simon	0:1014af42efd9	5	* $Revision: V1.0.3
simon	0:1014af42efd9	6	*
simon	0:1014af42efd9	7	* Project: CMSIS DSP Library
simon	0:1014af42efd9	8	* Title: arm_biquad_cascade_df1_fast_q31.c
simon	0:1014af42efd9	9	*
simon	0:1014af42efd9	10	* Description: Processing function for the
simon	0:1014af42efd9	11	* Q31 Fast Biquad cascade DirectFormI(DF1) filter.
simon	0:1014af42efd9	12	*
simon	0:1014af42efd9	13	* Target Processor: Cortex-M4/Cortex-M3
simon	0:1014af42efd9	14	*
simon	0:1014af42efd9	15	* Version 1.0.3 2010/11/29
simon	0:1014af42efd9	16	* Re-organized the CMSIS folders and updated documentation.
simon	0:1014af42efd9	17	*
simon	0:1014af42efd9	18	* Version 1.0.2 2010/11/11
simon	0:1014af42efd9	19	* Documentation updated.
simon	0:1014af42efd9	20	*
simon	0:1014af42efd9	21	* Version 1.0.1 2010/10/05
simon	0:1014af42efd9	22	* Production release and review comments incorporated.
simon	0:1014af42efd9	23	*
simon	0:1014af42efd9	24	* Version 1.0.0 2010/09/20
simon	0:1014af42efd9	25	* Production release and review comments incorporated.
simon	0:1014af42efd9	26	*
simon	0:1014af42efd9	27	* Version 0.0.9 2010/08/27
simon	0:1014af42efd9	28	* Initial version
simon	0:1014af42efd9	29	*
simon	0:1014af42efd9	30	* -------------------------------------------------------------------- */
simon	0:1014af42efd9	31
simon	0:1014af42efd9	32	#include "arm_math.h"
simon	0:1014af42efd9	33
simon	0:1014af42efd9	34	/**
simon	0:1014af42efd9	35	* @ingroup groupFilters
simon	0:1014af42efd9	36	*/
simon	0:1014af42efd9	37
simon	0:1014af42efd9	38	/**
simon	0:1014af42efd9	39	* @addtogroup BiquadCascadeDF1
simon	0:1014af42efd9	40	* @{
simon	0:1014af42efd9	41	*/
simon	0:1014af42efd9	42
simon	0:1014af42efd9	43	/**
simon	0:1014af42efd9	44	* @details
simon	0:1014af42efd9	45	*
simon	0:1014af42efd9	46	* @param[in] *S points to an instance of the Q31 Biquad cascade structure.
simon	0:1014af42efd9	47	* @param[in] *pSrc points to the block of input data.
simon	0:1014af42efd9	48	* @param[out] *pDst points to the block of output data.
simon	0:1014af42efd9	49	* @param[in] blockSize number of samples to process per call.
simon	0:1014af42efd9	50	* @return none.
simon	0:1014af42efd9	51	*
simon	0:1014af42efd9	52	* <b>Scaling and Overflow Behavior:</b>
simon	0:1014af42efd9	53	* \par
simon	0:1014af42efd9	54	* This function is optimized for speed at the expense of fixed-point precision and overflow protection.
simon	0:1014af42efd9	55	* The result of each 1.31 x 1.31 multiplication is truncated to 2.30 format.
simon	0:1014af42efd9	56	* These intermediate results are added to a 2.30 accumulator.
simon	0:1014af42efd9	57	* Finally, the accumulator is saturated and converted to a 1.31 result.
simon	0:1014af42efd9	58	* The fast version has the same overflow behavior as the standard version and provides less precision since it discards the low 32 bits of each multiplication result.
simon	0:1014af42efd9	59	* In order to avoid overflows completely the input signal must be scaled down by two bits and lie in the range [-0.25 +0.25). Use the intialization function
simon	0:1014af42efd9	60	* arm_biquad_cascade_df1_init_q31() to initialize filter structure.
simon	0:1014af42efd9	61	*
simon	0:1014af42efd9	62	* \par
simon	0:1014af42efd9	63	* Refer to the function <code>arm_biquad_cascade_df1_q31()</code> for a slower implementation of this function which uses 64-bit accumulation to provide higher precision. Both the slow and the fast versions use the same instance structure.
simon	0:1014af42efd9	64	* Use the function <code>arm_biquad_cascade_df1_init_q31()</code> to initialize the filter structure.
simon	0:1014af42efd9	65	*/
simon	0:1014af42efd9	66
simon	0:1014af42efd9	67	void arm_biquad_cascade_df1_fast_q31(
simon	0:1014af42efd9	68	const arm_biquad_casd_df1_inst_q31 * S,
simon	0:1014af42efd9	69	q31_t * pSrc,
simon	0:1014af42efd9	70	q31_t * pDst,
simon	0:1014af42efd9	71	uint32_t blockSize)
simon	0:1014af42efd9	72	{
simon	0:1014af42efd9	73	q31_t pIn = pSrc; / input pointer initialization */
simon	0:1014af42efd9	74	q31_t pOut = pDst; / output pointer initialization */
simon	0:1014af42efd9	75	q31_t pState = S->pState; / pState pointer initialization */
simon	0:1014af42efd9	76	q31_t pCoeffs = S->pCoeffs; / coeff pointer initialization */
simon	0:1014af42efd9	77	q31_t acc; /* accumulator */
simon	0:1014af42efd9	78	q31_t Xn1, Xn2, Yn1, Yn2; /* Filter state variables */
simon	0:1014af42efd9	79	q31_t b0, b1, b2, a1, a2; /* Filter coefficients */
simon	0:1014af42efd9	80	q31_t Xn; /* temporary input */
simon	0:1014af42efd9	81	int32_t shift = (int32_t) S->postShift + 1; /* Shift to be applied to the output */
simon	0:1014af42efd9	82	uint32_t sample, stage = S->numStages; /* loop counters */
simon	0:1014af42efd9	83
simon	0:1014af42efd9	84
simon	0:1014af42efd9	85	do
simon	0:1014af42efd9	86	{
simon	0:1014af42efd9	87	/* Reading the coefficients */
simon	0:1014af42efd9	88	b0 = *pCoeffs++;
simon	0:1014af42efd9	89	b1 = *pCoeffs++;
simon	0:1014af42efd9	90	b2 = *pCoeffs++;
simon	0:1014af42efd9	91	a1 = *pCoeffs++;
simon	0:1014af42efd9	92	a2 = *pCoeffs++;
simon	0:1014af42efd9	93
simon	0:1014af42efd9	94	/* Reading the state values */
simon	0:1014af42efd9	95	Xn1 = pState[0];
simon	0:1014af42efd9	96	Xn2 = pState[1];
simon	0:1014af42efd9	97	Yn1 = pState[2];
simon	0:1014af42efd9	98	Yn2 = pState[3];
simon	0:1014af42efd9	99
simon	0:1014af42efd9	100	/* Apply loop unrolling and compute 4 output values simultaneously. */
simon	0:1014af42efd9	101	/* The variables acc ... acc3 hold output values that are being computed:
simon	0:1014af42efd9	102	*
simon	0:1014af42efd9	103	* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2]
simon	0:1014af42efd9	104	*/
simon	0:1014af42efd9	105
simon	0:1014af42efd9	106	sample = blockSize >> 2u;
simon	0:1014af42efd9	107
simon	0:1014af42efd9	108	/* First part of the processing with loop unrolling. Compute 4 outputs at a time.
simon	0:1014af42efd9	109	** a second loop below computes the remaining 1 to 3 samples. */
simon	0:1014af42efd9	110	while(sample > 0u)
simon	0:1014af42efd9	111	{
simon	0:1014af42efd9	112	/* Read the input */
simon	0:1014af42efd9	113	Xn = *pIn++;
simon	0:1014af42efd9	114
simon	0:1014af42efd9	115	/* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */
simon	0:1014af42efd9	116	/* acc = b0 * x[n] */
simon	0:1014af42efd9	117	acc = (q31_t) (((q63_t) b0 * Xn) >> 32);
simon	0:1014af42efd9	118	/* acc += b1 * x[n-1] */
simon	0:1014af42efd9	119	acc = (q31_t) ((((q63_t) acc << 32) + ((q63_t) b1 * (Xn1))) >> 32);
simon	0:1014af42efd9	120	/* acc += b[2] * x[n-2] */
simon	0:1014af42efd9	121	acc = (q31_t) ((((q63_t) acc << 32) + ((q63_t) b2 * (Xn2))) >> 32);
simon	0:1014af42efd9	122	/* acc += a1 * y[n-1] */
simon	0:1014af42efd9	123	acc = (q31_t) ((((q63_t) acc << 32) + ((q63_t) a1 * (Yn1))) >> 32);
simon	0:1014af42efd9	124	/* acc += a2 * y[n-2] */
simon	0:1014af42efd9	125	acc = (q31_t) ((((q63_t) acc << 32) + ((q63_t) a2 * (Yn2))) >> 32);
simon	0:1014af42efd9	126
simon	0:1014af42efd9	127	/* The result is converted to 1.31 , Yn2 variable is reused */
simon	0:1014af42efd9	128	Yn2 = acc << shift;
simon	0:1014af42efd9	129
simon	0:1014af42efd9	130	/* Store the output in the destination buffer. */
simon	0:1014af42efd9	131	*pOut++ = Yn2;
simon	0:1014af42efd9	132
simon	0:1014af42efd9	133	/* Read the second input */
simon	0:1014af42efd9	134	Xn2 = *pIn++;
simon	0:1014af42efd9	135
simon	0:1014af42efd9	136	/* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */
simon	0:1014af42efd9	137	/* acc = b0 * x[n] */
simon	0:1014af42efd9	138	acc = (q31_t) (((q63_t) b0 * (Xn2)) >> 32);
simon	0:1014af42efd9	139	/* acc += b1 * x[n-1] */
simon	0:1014af42efd9	140	acc = (q31_t) ((((q63_t) acc << 32) + ((q63_t) b1 * (Xn))) >> 32);
simon	0:1014af42efd9	141	/* acc += b[2] * x[n-2] */
simon	0:1014af42efd9	142	acc = (q31_t) ((((q63_t) acc << 32) + ((q63_t) b2 * (Xn1))) >> 32);
simon	0:1014af42efd9	143	/* acc += a1 * y[n-1] */
simon	0:1014af42efd9	144	acc = (q31_t) ((((q63_t) acc << 32) + ((q63_t) a1 * (Yn2))) >> 32);
simon	0:1014af42efd9	145	/* acc += a2 * y[n-2] */
simon	0:1014af42efd9	146	acc = (q31_t) ((((q63_t) acc << 32) + ((q63_t) a2 * (Yn1))) >> 32);
simon	0:1014af42efd9	147
simon	0:1014af42efd9	148	/* The result is converted to 1.31, Yn1 variable is reused */
simon	0:1014af42efd9	149	Yn1 = acc << shift;
simon	0:1014af42efd9	150
simon	0:1014af42efd9	151	/* Store the output in the destination buffer. */
simon	0:1014af42efd9	152	*pOut++ = Yn1;
simon	0:1014af42efd9	153
simon	0:1014af42efd9	154	/* Read the third input */
simon	0:1014af42efd9	155	Xn1 = *pIn++;
simon	0:1014af42efd9	156
simon	0:1014af42efd9	157	/* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */
simon	0:1014af42efd9	158	/* acc = b0 * x[n] */
simon	0:1014af42efd9	159	acc = (q31_t) (((q63_t) b0 * (Xn1)) >> 32);
simon	0:1014af42efd9	160	/* acc += b1 * x[n-1] */
simon	0:1014af42efd9	161	acc = (q31_t) ((((q63_t) acc << 32) + ((q63_t) b1 * (Xn2))) >> 32);
simon	0:1014af42efd9	162	/* acc += b[2] * x[n-2] */
simon	0:1014af42efd9	163	acc = (q31_t) ((((q63_t) acc << 32) + ((q63_t) b2 * (Xn))) >> 32);
simon	0:1014af42efd9	164	/* acc += a1 * y[n-1] */
simon	0:1014af42efd9	165	acc = (q31_t) ((((q63_t) acc << 32) + ((q63_t) a1 * (Yn1))) >> 32);
simon	0:1014af42efd9	166	/* acc += a2 * y[n-2] */
simon	0:1014af42efd9	167	acc = (q31_t) ((((q63_t) acc << 32) + ((q63_t) a2 * (Yn2))) >> 32);
simon	0:1014af42efd9	168
simon	0:1014af42efd9	169	/* The result is converted to 1.31, Yn2 variable is reused */
simon	0:1014af42efd9	170	Yn2 = acc << shift;
simon	0:1014af42efd9	171
simon	0:1014af42efd9	172	/* Store the output in the destination buffer. */
simon	0:1014af42efd9	173	*pOut++ = Yn2;
simon	0:1014af42efd9	174
simon	0:1014af42efd9	175	/* Read the forth input */
simon	0:1014af42efd9	176	Xn = *pIn++;
simon	0:1014af42efd9	177
simon	0:1014af42efd9	178	/* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */
simon	0:1014af42efd9	179	/* acc = b0 * x[n] */
simon	0:1014af42efd9	180	acc = (q31_t) (((q63_t) b0 * (Xn)) >> 32);
simon	0:1014af42efd9	181	/* acc += b1 * x[n-1] */
simon	0:1014af42efd9	182	acc = (q31_t) ((((q63_t) acc << 32) + ((q63_t) b1 * (Xn1))) >> 32);
simon	0:1014af42efd9	183	/* acc += b[2] * x[n-2] */
simon	0:1014af42efd9	184	acc = (q31_t) ((((q63_t) acc << 32) + ((q63_t) b2 * (Xn2))) >> 32);
simon	0:1014af42efd9	185	/* acc += a1 * y[n-1] */
simon	0:1014af42efd9	186	acc = (q31_t) ((((q63_t) acc << 32) + ((q63_t) a1 * (Yn2))) >> 32);
simon	0:1014af42efd9	187	/* acc += a2 * y[n-2] */
simon	0:1014af42efd9	188	acc = (q31_t) ((((q63_t) acc << 32) + ((q63_t) a2 * (Yn1))) >> 32);
simon	0:1014af42efd9	189
simon	0:1014af42efd9	190	/* The result is converted to 1.31, Yn1 variable is reused */
simon	0:1014af42efd9	191	Yn1 = acc << shift;
simon	0:1014af42efd9	192
simon	0:1014af42efd9	193	/* Every time after the output is computed state should be updated. */
simon	0:1014af42efd9	194	/* The states should be updated as: */
simon	0:1014af42efd9	195	/* Xn2 = Xn1 */
simon	0:1014af42efd9	196	/* Xn1 = Xn */
simon	0:1014af42efd9	197	/* Yn2 = Yn1 */
simon	0:1014af42efd9	198	/* Yn1 = acc */
simon	0:1014af42efd9	199	Xn2 = Xn1;
simon	0:1014af42efd9	200	Xn1 = Xn;
simon	0:1014af42efd9	201
simon	0:1014af42efd9	202	/* Store the output in the destination buffer. */
simon	0:1014af42efd9	203	*pOut++ = Yn1;
simon	0:1014af42efd9	204
simon	0:1014af42efd9	205	/* decrement the loop counter */
simon	0:1014af42efd9	206	sample--;
simon	0:1014af42efd9	207	}
simon	0:1014af42efd9	208
simon	0:1014af42efd9	209	/* If the blockSize is not a multiple of 4, compute any remaining output samples here.
simon	0:1014af42efd9	210	** No loop unrolling is used. */
simon	0:1014af42efd9	211	sample = (blockSize & 0x3u);
simon	0:1014af42efd9	212
simon	0:1014af42efd9	213	while(sample > 0u)
simon	0:1014af42efd9	214	{
simon	0:1014af42efd9	215	/* Read the input */
simon	0:1014af42efd9	216	Xn = *pIn++;
simon	0:1014af42efd9	217
simon	0:1014af42efd9	218	/* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */
simon	0:1014af42efd9	219	/* acc = b0 * x[n] */
simon	0:1014af42efd9	220	acc = (q31_t) (((q63_t) b0 * (Xn)) >> 32);
simon	0:1014af42efd9	221	/* acc += b1 * x[n-1] */
simon	0:1014af42efd9	222	acc = (q31_t) ((((q63_t) acc << 32) + ((q63_t) b1 * (Xn1))) >> 32);
simon	0:1014af42efd9	223	/* acc += b[2] * x[n-2] */
simon	0:1014af42efd9	224	acc = (q31_t) ((((q63_t) acc << 32) + ((q63_t) b2 * (Xn2))) >> 32);
simon	0:1014af42efd9	225	/* acc += a1 * y[n-1] */
simon	0:1014af42efd9	226	acc = (q31_t) ((((q63_t) acc << 32) + ((q63_t) a1 * (Yn1))) >> 32);
simon	0:1014af42efd9	227	/* acc += a2 * y[n-2] */
simon	0:1014af42efd9	228	acc = (q31_t) ((((q63_t) acc << 32) + ((q63_t) a2 * (Yn2))) >> 32);
simon	0:1014af42efd9	229	/* The result is converted to 1.31 */
simon	0:1014af42efd9	230	acc = acc << shift;
simon	0:1014af42efd9	231
simon	0:1014af42efd9	232	/* Every time after the output is computed state should be updated. */
simon	0:1014af42efd9	233	/* The states should be updated as: */
simon	0:1014af42efd9	234	/* Xn2 = Xn1 */
simon	0:1014af42efd9	235	/* Xn1 = Xn */
simon	0:1014af42efd9	236	/* Yn2 = Yn1 */
simon	0:1014af42efd9	237	/* Yn1 = acc */
simon	0:1014af42efd9	238	Xn2 = Xn1;
simon	0:1014af42efd9	239	Xn1 = Xn;
simon	0:1014af42efd9	240	Yn2 = Yn1;
simon	0:1014af42efd9	241	Yn1 = acc;
simon	0:1014af42efd9	242
simon	0:1014af42efd9	243	/* Store the output in the destination buffer. */
simon	0:1014af42efd9	244	*pOut++ = acc;
simon	0:1014af42efd9	245
simon	0:1014af42efd9	246	/* decrement the loop counter */
simon	0:1014af42efd9	247	sample--;
simon	0:1014af42efd9	248	}
simon	0:1014af42efd9	249
simon	0:1014af42efd9	250	/* The first stage goes from the input buffer to the output buffer. */
simon	0:1014af42efd9	251	/* Subsequent stages occur in-place in the output buffer */
simon	0:1014af42efd9	252	pIn = pDst;
simon	0:1014af42efd9	253
simon	0:1014af42efd9	254	/* Reset to destination pointer */
simon	0:1014af42efd9	255	pOut = pDst;
simon	0:1014af42efd9	256
simon	0:1014af42efd9	257	/* Store the updated state variables back into the pState array */
simon	0:1014af42efd9	258	*pState++ = Xn1;
simon	0:1014af42efd9	259	*pState++ = Xn2;
simon	0:1014af42efd9	260	*pState++ = Yn1;
simon	0:1014af42efd9	261	*pState++ = Yn2;
simon	0:1014af42efd9	262
simon	0:1014af42efd9	263	} while(--stage);
simon	0:1014af42efd9	264	}
simon	0:1014af42efd9	265
simon	0:1014af42efd9	266	/**
simon	0:1014af42efd9	267	* @} end of BiquadCascadeDF1 group
simon	0:1014af42efd9	268	*/

Repository toolbox

Export to desktop IDE

Repository details

Type:	Library
Created:	10 Mar 2011
Imports:	907
Forks:	1
Commits:	3
Dependents:	5
Dependencies:	0
Followers:	35

src/Cortex-M4-M3/FilteringFunctions/arm_biquad_cascade_df1_fast_q31.c@0:1014af42efd9, 2011-03-10 (annotated)

Who changed what in which revision?

Repository toolbox

Repository details

Important Information for this Arm website

Access Warning