Renesas
Fork of mbed-dsp. CMSIS-DSP library of supporting NEON
Dependents: mbed-os-example-cmsis_dsp_neon
Fork of
mbed-dsp
by 
mbed official
Information
Japanese version is available in lower part of this page.
このページの後半に日本語版が用意されています．
CMSIS-DSP of supporting NEON
What is this ?
A library for CMSIS-DSP of supporting NEON.
We supported the NEON to CMSIS-DSP Ver1.4.3(CMSIS V4.1) that ARM supplied, has achieved the processing speed improvement.
If you use the mbed-dsp library, you can use to replace this library.
CMSIS-DSP of supporting NEON is provied as a library.
Library Creation environment
CMSIS-DSP library of supporting NEON was created by the following environment.
- Compiler
ARMCC Version 5.03 - Compile option switch[C Compiler]
-DARM_MATH_MATRIX_CHECK -DARM_MATH_ROUNDING -O3 -Otime --cpu=Cortex-A9 --littleend --arm --apcs=/interwork --no_unaligned_access --fpu=vfpv3_fp16 --fpmode=fast --apcs=/hardfp --vectorize --asm
- Compile option switch[Assembler]
--cpreproc --cpu=Cortex-A9 --littleend --arm --apcs=/interwork --no_unaligned_access --fpu=vfpv3_fp16 --fpmode=fast --apcs=/hardfp
Effects of NEON support
In the data which passes to each function, large size will be expected more effective than small size.
Also if the data is a multiple of 16, effect will be expected in every function in the CMSIS-DSP.
NEON対応CMSIS-DSP
概要
NEON対応したCMSIS-DSPのライブラリです。
ARM社提供のCMSIS-DSP Ver1.4.3(CMSIS V4.1)をターゲットにNEON対応を行ない、処理速度向上を実現しております。
mbed-dspライブラリを使用している場合は、本ライブラリに置き換えて使用することができます。
NEON対応したCMSIS-DSPはライブラリで提供します。
ライブラリ作成環境
NEON対応CMSIS-DSPライブラリは、以下の環境で作成しています。
- コンパイラ
ARMCC Version 5.03 - コンパイルオプションスイッチ[C Compiler]
-DARM_MATH_MATRIX_CHECK -DARM_MATH_ROUNDING -O3 -Otime --cpu=Cortex-A9 --littleend --arm --apcs=/interwork --no_unaligned_access --fpu=vfpv3_fp16 --fpmode=fast --apcs=/hardfp --vectorize --asm
- コンパイルオプションスイッチ[Assembler]
--cpreproc --cpu=Cortex-A9 --littleend --arm --apcs=/interwork --no_unaligned_access --fpu=vfpv3_fp16 --fpmode=fast --apcs=/hardfp
NEON対応による効果について
CMSIS-DSP内の各関数へ渡すデータは、小さいサイズよりも大きいサイズの方が効果が見込めます。
また、16の倍数のデータであれば、CMSIS-DSP内のどの関数でも効果が見込めます。
Diff: cmsis_dsp/arm_math.h
- Revision:
- 2:da51fb522205
- Parent:
- 1:fdd22bb7aa52
- Child:
- 3:7a284390b0ce
--- a/cmsis_dsp/arm_math.h Wed Nov 28 12:30:09 2012 +0000
+++ b/cmsis_dsp/arm_math.h Thu May 30 17:10:11 2013 +0100
@@ -2,12 +2,12 @@
* Copyright (C) 2010-2011 ARM Limited. All rights reserved.
*
* $Date: 15. February 2012
- * $Revision: V1.1.0
+ * $Revision: V1.1.0
*
- * Project: CMSIS DSP Library
- * Title: arm_math.h
+ * Project: CMSIS DSP Library
+ * Title: arm_math.h
*
- * Description: Public header file for CMSIS DSP Library
+ * Description: Public header file for CMSIS DSP Library
*
* Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
*
@@ -280,7 +280,7 @@
#undef __CMSIS_GENERIC /* enable NVIC and Systick functions */
#include "string.h"
#include "math.h"
-#ifdef __cplusplus
+#ifdef __cplusplus
extern "C"
{
#endif
@@ -290,27 +290,27 @@
* @brief Macros required for reciprocal calculation in Normalized LMS
*/
-#define DELTA_Q31 (0x100)
-#define DELTA_Q15 0x5
-#define INDEX_MASK 0x0000003F
+#define DELTA_Q31 (0x100)
+#define DELTA_Q15 0x5
+#define INDEX_MASK 0x0000003F
#ifndef PI
-#define PI 3.14159265358979f
+#define PI 3.14159265358979f
#endif
/**
* @brief Macros required for SINE and COSINE Fast math approximations
*/
-#define TABLE_SIZE 256
-#define TABLE_SPACING_Q31 0x800000
-#define TABLE_SPACING_Q15 0x80
+#define TABLE_SIZE 256
+#define TABLE_SPACING_Q31 0x800000
+#define TABLE_SPACING_Q15 0x80
/**
* @brief Macros required for SINE and COSINE Controller functions
*/
/* 1.31(q31) Fixed value of 2/360 */
/* -1 to +1 is divided into 360 values so total spacing is (2/360) */
-#define INPUT_SPACING 0xB60B61
+#define INPUT_SPACING 0xB60B61
/**
* @brief Macro for Unaligned Support
@@ -323,7 +323,7 @@
#else
#define ALIGN4 __align(4)
#endif
-#endif /* #ifndef UNALIGNED_SUPPORT_DISABLE */
+#endif /* #ifndef UNALIGNED_SUPPORT_DISABLE */
/**
* @brief Error status returned by some functions in the library.
@@ -400,16 +400,16 @@
*/
#ifndef ARM_MATH_BIG_ENDIAN
-#define __PACKq7(v0,v1,v2,v3) ( (((int32_t)(v0) << 0) & (int32_t)0x000000FF) | \
- (((int32_t)(v1) << 8) & (int32_t)0x0000FF00) | \
- (((int32_t)(v2) << 16) & (int32_t)0x00FF0000) | \
- (((int32_t)(v3) << 24) & (int32_t)0xFF000000) )
+#define __PACKq7(v0,v1,v2,v3) ( (((int32_t)(v0) << 0) & (int32_t)0x000000FF) | \
+ (((int32_t)(v1) << 8) & (int32_t)0x0000FF00) | \
+ (((int32_t)(v2) << 16) & (int32_t)0x00FF0000) | \
+ (((int32_t)(v3) << 24) & (int32_t)0xFF000000) )
#else
-#define __PACKq7(v0,v1,v2,v3) ( (((int32_t)(v3) << 0) & (int32_t)0x000000FF) | \
- (((int32_t)(v2) << 8) & (int32_t)0x0000FF00) | \
- (((int32_t)(v1) << 16) & (int32_t)0x00FF0000) | \
- (((int32_t)(v0) << 24) & (int32_t)0xFF000000) )
+#define __PACKq7(v0,v1,v2,v3) ( (((int32_t)(v3) << 0) & (int32_t)0x000000FF) | \
+ (((int32_t)(v2) << 8) & (int32_t)0x0000FF00) | \
+ (((int32_t)(v1) << 16) & (int32_t)0x00FF0000) | \
+ (((int32_t)(v0) << 24) & (int32_t)0xFF000000) )
#endif
@@ -1177,11 +1177,11 @@
/**
* @brief Initialization function for the Q31 FIR filter.
* @param[in,out] *S points to an instance of the Q31 FIR structure.
- * @param[in] numTaps Number of filter coefficients in the filter.
- * @param[in] *pCoeffs points to the filter coefficients.
- * @param[in] *pState points to the state buffer.
- * @param[in] blockSize number of samples that are processed at a time.
- * @return none.
+ * @param[in] numTaps Number of filter coefficients in the filter.
+ * @param[in] *pCoeffs points to the filter coefficients.
+ * @param[in] *pState points to the state buffer.
+ * @param[in] blockSize number of samples that are processed at a time.
+ * @return none.
*/
void arm_fir_init_q31(
arm_fir_instance_q31 * S,
@@ -1207,11 +1207,11 @@
/**
* @brief Initialization function for the floating-point FIR filter.
* @param[in,out] *S points to an instance of the floating-point FIR filter structure.
- * @param[in] numTaps Number of filter coefficients in the filter.
- * @param[in] *pCoeffs points to the filter coefficients.
- * @param[in] *pState points to the state buffer.
- * @param[in] blockSize number of samples that are processed at a time.
- * @return none.
+ * @param[in] numTaps Number of filter coefficients in the filter.
+ * @param[in] *pCoeffs points to the filter coefficients.
+ * @param[in] *pState points to the state buffer.
+ * @param[in] blockSize number of samples that are processed at a time.
+ * @return none.
*/
void arm_fir_init_f32(
arm_fir_instance_f32 * S,
@@ -1471,7 +1471,7 @@
* @brief Floating-point matrix transpose.
* @param[in] *pSrc points to the input matrix
* @param[out] *pDst points to the output matrix
- * @return The function returns either <code>ARM_MATH_SIZE_MISMATCH</code>
+ * @return The function returns either <code>ARM_MATH_SIZE_MISMATCH</code>
* or <code>ARM_MATH_SUCCESS</code> based on the outcome of size checking.
*/
@@ -1484,7 +1484,7 @@
* @brief Q15 matrix transpose.
* @param[in] *pSrc points to the input matrix
* @param[out] *pDst points to the output matrix
- * @return The function returns either <code>ARM_MATH_SIZE_MISMATCH</code>
+ * @return The function returns either <code>ARM_MATH_SIZE_MISMATCH</code>
* or <code>ARM_MATH_SUCCESS</code> based on the outcome of size checking.
*/
@@ -1496,7 +1496,7 @@
* @brief Q31 matrix transpose.
* @param[in] *pSrc points to the input matrix
* @param[out] *pDst points to the output matrix
- * @return The function returns either <code>ARM_MATH_SIZE_MISMATCH</code>
+ * @return The function returns either <code>ARM_MATH_SIZE_MISMATCH</code>
* or <code>ARM_MATH_SUCCESS</code> based on the outcome of size checking.
*/
@@ -1539,7 +1539,7 @@
* @param[in] *pSrcA points to the first input matrix structure
* @param[in] *pSrcB points to the second input matrix structure
* @param[out] *pDst points to output matrix structure
- * @param[in] *pState points to the array for storing intermediate results
+ * @param[in] *pState points to the array for storing intermediate results
* @return The function returns either
* <code>ARM_MATH_SIZE_MISMATCH</code> or <code>ARM_MATH_SUCCESS</code> based on the outcome of size checking.
*/
@@ -1673,7 +1673,7 @@
* @param[in,out] *S points to an instance of the floating-point matrix structure.
* @param[in] nRows number of rows in the matrix.
* @param[in] nColumns number of columns in the matrix.
- * @param[in] *pData points to the matrix data array.
+ * @param[in] *pData points to the matrix data array.
* @return none
*/
@@ -1688,7 +1688,7 @@
* @param[in,out] *S points to an instance of the floating-point matrix structure.
* @param[in] nRows number of rows in the matrix.
* @param[in] nColumns number of columns in the matrix.
- * @param[in] *pData points to the matrix data array.
+ * @param[in] *pData points to the matrix data array.
* @return none
*/
@@ -1703,7 +1703,7 @@
* @param[in,out] *S points to an instance of the floating-point matrix structure.
* @param[in] nRows number of rows in the matrix.
* @param[in] nColumns number of columns in the matrix.
- * @param[in] *pData points to the matrix data array.
+ * @param[in] *pData points to the matrix data array.
* @return none
*/
@@ -2195,7 +2195,7 @@
/*----------------------------------------------------------------------
- * Internal functions prototypes FFT function
+ * Internal functions prototypes FFT function
----------------------------------------------------------------------*/
/**
@@ -2264,7 +2264,7 @@
* @brief Core function for the f32 FFT butterfly process.
* @param[in, out] *pSrc points to the in-place buffer of f32 data type.
* @param[in] fftLen length of the FFT.
- * @param[in] *pCoef points to Twiddle coefficient buffer.
+ * @param[in] *pCoef points to Twiddle coefficient buffer.
* @param[in] twidCoefModifier twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table.
* @return none.
*/
@@ -2276,13 +2276,13 @@
uint16_t twidCoefModifier);
/**
- * @brief Core function for the Radix-2 Q31 CFFT butterfly process.
- * @param[in, out] *pSrc points to the in-place buffer of Q31 data type.
- * @param[in] fftLen length of the FFT.
- * @param[in] *pCoef points to Twiddle coefficient buffer.
- * @param[in] twidCoefModifier twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table.
- * @return none.
- */
+ * @brief Core function for the Radix-2 Q31 CFFT butterfly process.
+ * @param[in, out] *pSrc points to the in-place buffer of Q31 data type.
+ * @param[in] fftLen length of the FFT.
+ * @param[in] *pCoef points to Twiddle coefficient buffer.
+ * @param[in] twidCoefModifier twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table.
+ * @return none.
+ */
void arm_radix2_butterfly_q31(
q31_t * pSrc,
@@ -2291,13 +2291,13 @@
uint16_t twidCoefModifier);
/**
- * @brief Core function for the Radix-2 Q15 CFFT butterfly process.
- * @param[in, out] *pSrc points to the in-place buffer of Q15 data type.
- * @param[in] fftLen length of the FFT.
- * @param[in] *pCoef points to Twiddle coefficient buffer.
- * @param[in] twidCoefModifier twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table.
- * @return none.
- */
+ * @brief Core function for the Radix-2 Q15 CFFT butterfly process.
+ * @param[in, out] *pSrc points to the in-place buffer of Q15 data type.
+ * @param[in] fftLen length of the FFT.
+ * @param[in] *pCoef points to Twiddle coefficient buffer.
+ * @param[in] twidCoefModifier twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table.
+ * @return none.
+ */
void arm_radix2_butterfly_q15(
q15_t * pSrc,
@@ -2306,13 +2306,13 @@
uint16_t twidCoefModifier);
/**
- * @brief Core function for the Radix-2 Q15 CFFT Inverse butterfly process.
- * @param[in, out] *pSrc points to the in-place buffer of Q15 data type.
- * @param[in] fftLen length of the FFT.
- * @param[in] *pCoef points to Twiddle coefficient buffer.
- * @param[in] twidCoefModifier twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table.
- * @return none.
- */
+ * @brief Core function for the Radix-2 Q15 CFFT Inverse butterfly process.
+ * @param[in, out] *pSrc points to the in-place buffer of Q15 data type.
+ * @param[in] fftLen length of the FFT.
+ * @param[in] *pCoef points to Twiddle coefficient buffer.
+ * @param[in] twidCoefModifier twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table.
+ * @return none.
+ */
void arm_radix2_butterfly_inverse_q15(
q15_t * pSrc,
@@ -2321,13 +2321,13 @@
uint16_t twidCoefModifier);
/**
- * @brief Core function for the Radix-2 Q31 CFFT Inverse butterfly process.
- * @param[in, out] *pSrc points to the in-place buffer of Q31 data type.
- * @param[in] fftLen length of the FFT.
- * @param[in] *pCoef points to Twiddle coefficient buffer.
- * @param[in] twidCoefModifier twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table.
- * @return none.
- */
+ * @brief Core function for the Radix-2 Q31 CFFT Inverse butterfly process.
+ * @param[in, out] *pSrc points to the in-place buffer of Q31 data type.
+ * @param[in] fftLen length of the FFT.
+ * @param[in] *pCoef points to Twiddle coefficient buffer.
+ * @param[in] twidCoefModifier twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table.
+ * @return none.
+ */
void arm_radix2_butterfly_inverse_q31(
q31_t * pSrc,
@@ -2339,9 +2339,9 @@
* @brief Core function for the f32 IFFT butterfly process.
* @param[in, out] *pSrc points to the in-place buffer of f32 data type.
* @param[in] fftLen length of the FFT.
- * @param[in] *pCoef points to Twiddle coefficient buffer.
+ * @param[in] *pCoef points to Twiddle coefficient buffer.
* @param[in] twidCoefModifier twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table.
- * @param[in] onebyfftLen 1/fftLenfth
+ * @param[in] onebyfftLen 1/fftLenfth
* @return none.
*/
@@ -2497,7 +2497,7 @@
* @param[in] fftLenReal length of the FFT.
* @param[in] ifftFlagR flag that selects forward (ifftFlagR=0) or inverse (ifftFlagR=1) transform.
* @param[in] bitReverseFlag flag that enables (bitReverseFlag=1) or disables (bitReverseFlag=0) bit reversal of output.
- * @return The function returns ARM_MATH_SUCCESS if initialization is successful or ARM_MATH_ARGUMENT_ERROR if <code>fftLenReal</code> is not a supported value.
+ * @return The function returns ARM_MATH_SUCCESS if initialization is successful or ARM_MATH_ARGUMENT_ERROR if <code>fftLenReal</code> is not a supported value.
*/
arm_status arm_rfft_init_q15(
@@ -2527,7 +2527,7 @@
* @param[in] fftLenReal length of the FFT.
* @param[in] ifftFlagR flag that selects forward (ifftFlagR=0) or inverse (ifftFlagR=1) transform.
* @param[in] bitReverseFlag flag that enables (bitReverseFlag=1) or disables (bitReverseFlag=0) bit reversal of output.
- * @return The function returns ARM_MATH_SUCCESS if initialization is successful or ARM_MATH_ARGUMENT_ERROR if <code>fftLenReal</code> is not a supported value.
+ * @return The function returns ARM_MATH_SUCCESS if initialization is successful or ARM_MATH_ARGUMENT_ERROR if <code>fftLenReal</code> is not a supported value.
*/
arm_status arm_rfft_init_q31(
@@ -2544,7 +2544,7 @@
* @param[in] fftLenReal length of the FFT.
* @param[in] ifftFlagR flag that selects forward (ifftFlagR=0) or inverse (ifftFlagR=1) transform.
* @param[in] bitReverseFlag flag that enables (bitReverseFlag=1) or disables (bitReverseFlag=0) bit reversal of output.
- * @return The function returns ARM_MATH_SUCCESS if initialization is successful or ARM_MATH_ARGUMENT_ERROR if <code>fftLenReal</code> is not a supported value.
+ * @return The function returns ARM_MATH_SUCCESS if initialization is successful or ARM_MATH_ARGUMENT_ERROR if <code>fftLenReal</code> is not a supported value.
*/
arm_status arm_rfft_init_f32(
@@ -2590,7 +2590,7 @@
* @param[in] N length of the DCT4.
* @param[in] Nby2 half of the length of the DCT4.
* @param[in] normalize normalizing factor.
- * @return arm_status function returns ARM_MATH_SUCCESS if initialization is successful or ARM_MATH_ARGUMENT_ERROR if <code>fftLenReal</code> is not a supported transform length.
+ * @return arm_status function returns ARM_MATH_SUCCESS if initialization is successful or ARM_MATH_ARGUMENT_ERROR if <code>fftLenReal</code> is not a supported transform length.
*/
arm_status arm_dct4_init_f32(
@@ -2637,7 +2637,7 @@
* @param[in] N length of the DCT4.
* @param[in] Nby2 half of the length of the DCT4.
* @param[in] normalize normalizing factor.
- * @return arm_status function returns ARM_MATH_SUCCESS if initialization is successful or ARM_MATH_ARGUMENT_ERROR if <code>N</code> is not a supported transform length.
+ * @return arm_status function returns ARM_MATH_SUCCESS if initialization is successful or ARM_MATH_ARGUMENT_ERROR if <code>N</code> is not a supported transform length.
*/
arm_status arm_dct4_init_q31(
@@ -2684,7 +2684,7 @@
* @param[in] N length of the DCT4.
* @param[in] Nby2 half of the length of the DCT4.
* @param[in] normalize normalizing factor.
- * @return arm_status function returns ARM_MATH_SUCCESS if initialization is successful or ARM_MATH_ARGUMENT_ERROR if <code>N</code> is not a supported transform length.
+ * @return arm_status function returns ARM_MATH_SUCCESS if initialization is successful or ARM_MATH_ARGUMENT_ERROR if <code>N</code> is not a supported transform length.
*/
arm_status arm_dct4_init_q15(
@@ -3326,11 +3326,11 @@
*/
void arm_conv_fast_q15(
- q15_t * pSrcA,
- uint32_t srcALen,
- q15_t * pSrcB,
- uint32_t srcBLen,
- q15_t * pDst);
+ q15_t * pSrcA,
+ uint32_t srcALen,
+ q15_t * pSrcB,
+ uint32_t srcBLen,
+ q15_t * pDst);
/**
* @brief Convolution of Q15 sequences (fast version) for Cortex-M3 and Cortex-M4
@@ -3512,13 +3512,13 @@
*/
arm_status arm_conv_partial_fast_q15(
- q15_t * pSrcA,
- uint32_t srcALen,
- q15_t * pSrcB,
- uint32_t srcBLen,
- q15_t * pDst,
- uint32_t firstIndex,
- uint32_t numPoints);
+ q15_t * pSrcA,
+ uint32_t srcALen,
+ q15_t * pSrcB,
+ uint32_t srcBLen,
+ q15_t * pDst,
+ uint32_t firstIndex,
+ uint32_t numPoints);
/**
@@ -4709,11 +4709,11 @@
*/
void arm_correlate_fast_q15(
- q15_t * pSrcA,
- uint32_t srcALen,
- q15_t * pSrcB,
- uint32_t srcBLen,
- q15_t * pDst);
+ q15_t * pSrcA,
+ uint32_t srcALen,
+ q15_t * pSrcB,
+ uint32_t srcBLen,
+ q15_t * pDst);
@@ -5612,8 +5612,8 @@
* @brief Floating-point Park transform
* @param[in] Ialpha input two-phase vector coordinate alpha
* @param[in] Ibeta input two-phase vector coordinate beta
- * @param[out] *pId points to output rotor reference frame d
- * @param[out] *pIq points to output rotor reference frame q
+ * @param[out] *pId points to output rotor reference frame d
+ * @param[out] *pIq points to output rotor reference frame q
* @param[in] sinVal sine value of rotation angle theta
* @param[in] cosVal cosine value of rotation angle theta
* @return none.
@@ -5759,7 +5759,7 @@
/**
- * @brief Inverse Park transform for Q31 version
+ * @brief Inverse Park transform for Q31 version
* @param[in] Id input coordinate of rotor reference frame d
* @param[in] Iq input coordinate of rotor reference frame q
* @param[out] *pIalpha points to output two-phase orthogonal vector axis alpha
@@ -7551,7 +7551,7 @@
-#ifdef __cplusplus
+#ifdef __cplusplus
}
#endif