Commit Graph

259 Commits (46231a68e5b12c7859ddc5ab68e8f0f03c56676b)

Author SHA1 Message Date
ClaudioMartino 033f5e17de Removed useless vld1q_f32 from arm_correlate_f32 6 years ago
Christophe Favergeon 799b7aa521 CMSIS-DSP: Added benchmarking code for Transform. 6 years ago
Christophe Favergeon f9ca44bc50 CMSIS-DSP: Cleaning of Doxygen comments for new functions.
New distance patterns.
6 years ago
Christophe Favergeon 7a8a02f976 CMSIS-DSP: Corrected Doxygen warnings. 6 years ago
Christophe Favergeon 1f87fd50c9 CMSIS-DSP: Change to cmake to more easily build wrappers 6 years ago
Christophe Favergeon b8fbe472b0 CMSIS-DSP:Some corrections to cmake build system. 6 years ago
Christophe Favergeon bd19f75d4f CMSIS-DSP: Minor cleaning and script for color syntax of test
descriptions
6 years ago
Christophe Favergeon 7a07cecef1 CMSIS-DSP: Added distance functions for clustering algorithms 6 years ago
Christophe Favergeon aa43cfb6dd CMSIS-DSP: Added some support functions.
entropy, Kullback-Leibler, LogSumExp, Dot product using LogSumExp
weighted sum, barycenter.
6 years ago
Christophe Favergeon 94b63664f2 CMSIS-DSP: Added Naive Gaussian Bayesian Estimator 6 years ago
Christophe Favergeon 302ada6633 CMSIS-DSP: Added SVM Functions and tests
Added more tests for BasicMathFunctions
Added script to postprocess result of benchmarks.
6 years ago
Christophe Favergeon d3170304e1 CMSIS-DSP:Added missing compilation flags to arm_rfft_fast_init_f32.c 7 years ago
Christophe Favergeon 8a4c63e4d8 CMSIS-DSP: New testing framework
(For our internal use. In short term, we won't give support about it).
CMSIS-DSP: Update to cmake build for the testing framework
CMSIS-NN:Implementation of arm_fully_connected_s8
Use API and quantization compatible with TF Lite.
7 years ago
ClaudioMartino ad726aa823 vmlaq in arm_dot_prod_f32 instead of vmulq and vadd 7 years ago
Christophe Favergeon bf1f324a6d CMSIS-DSP: DSP_Lib_TestSuite can be built with cmake.
DSP_Lib_TestSuite can be built with cmake and run on FVP.
Some issues with CMSIS-DSP cmake where discovered and corrected.

Comments added to arm_biqaud_cascade_df2T+f32 since the Neon version
must be initialized differently (and thus the corresponding test
in DSP_Lib_TestSuite will have to be updated to pass with Neon version).
7 years ago
Christophe Favergeon c0f96fe5d8 CMSIS-DSP: Corrected issue in Neon version of arm_correlate_f32.
Issue was introduced by a merge on our internal branch.
7 years ago
Christophe Favergeon 3bf2e2eda6 CMSIS-DSP: Enable build with MSVC compiler.
The goal of this commit is to allow the build of CMSIS-DSP on Windows to
enable the use of the library from Python or Matlab.

The Python wrapper has also been simplified and is now using the
CMSIS-DSP directly without requiring specific include files or source files.
7 years ago
Christophe Favergeon 080b994dd7 CMSIS-DSP: Customization of FFT tables.
Compilation symbols have been introduced to allow to include only the
FFT tables which are useful in a given application.

The README.md in DSP is giving more details.
7 years ago
Christophe Favergeon de626e2391 CMSIS DSP : Added a cmake build system
It is not replacing the current MDK and ArmDS projects.
It is a complementary way to build CMSIS-DSP.
Current version is supporting ARM AC6 compiler from ArmDS.

The result of the build can be run on a FVP for instance.

It could be extended to use different toolchain and build for different
targets.

The README.md in DSP folder is giving all the details.
7 years ago
stgoh 295292422c Rectified Description error on twiddeCoeff_rfft_32
Rectified error on formula that generates Twiddle coefficients table for RFFT (line 17749)
7 years ago
ClaudioMartino 7c1fca8052 Removed comments about loop unrolling in NEON codes to avoid confusion 7 years ago
ClaudioMartino 5792f80cf1 New NEON float32 algorithms for Cortex A5 (TransformFunctions excluded) with armDS project 7 years ago
Jonatan Antoni a8cf6e9bb1 Updated file versions, dates and copyright headers.
Change-Id: I2fe43d2ad56cff4347424929fd1f82eccd1cc770
7 years ago
GuentherMartin 5648cfd4af DSP_Lib changes:
- new version V1.5.5
 - reworked DSP library source files
     added macro ARM_MATH_LOOPUNROLL
     removed macro UNALIGNED_SUPPORT_DISABLE
     relpaced arm_bitreversal2.S with C version
     added const-correctness
     replaced SIMD pointer construct with memcpy solution
 - reworked DSP library documentation
 - moved DSP libraries to folder ./DSP/Lib
 - ARM DSP Libraries are built with ARMCLANG
 - Added DSP Libraries Source variant
7 years ago
Jonatan Antoni f20c424e92 Revert "Const-qualify all read-only pointers in CMSIS-DSP (#349)" 7 years ago
Torfinn Berset 37f6c581e8 Fix duplicate `const` declaration specifier 7 years ago
Torfinn Berset 61bd40e7a6 Fix incorrect pointer sign for arm_cfft_radix4_q15's bit reversal table (was signed, should be unsigned) 7 years ago
Torfinn Berset 294190cc2d Fix duplicate `const` declaration specifier 7 years ago
Torfinn Berset db3a3ae6a5 Fix const errors when compiling for Cortex M0 7 years ago
Torfinn Berset 23765aeff4 Fix const warnings 7 years ago
Torfinn Berset b54e1f9bf1 Converting C sources to use `const` where appropriate using regular expressions. Fix typos 7 years ago
GuentherMartin ddee94e8db DSP_Lib:
- updated arm_math.h
 - reduced ARM_MATH_CMx macros
7 years ago
Dave Flogeras 41a3f92259 Implement arm_rfft_fast_init_f32 with new separate initialize functions. 7 years ago
Dave Flogeras 0f5b33ee1b Split rfft_fast_init_f32 into separate initializer functions.
The rationale here is that if the caller knows the size FFT they require
at compile time, only the look-up tables required for that size are
placed in the data segment (ROM).  This saves between ~40k to ~80k of
ROM when only using a single FFT size.
7 years ago
GuentherMartin da38c27dd0 Merged PR #166 7 years ago
Andrew Leech 1b43c4f5d5 DSP: fix double-promotion warnings in arm_mat_inverse_f64 7 years ago
Oskar Weigl fd1a8d82f6 restore unrelated changes inadvertently reverted due to bad commit separation 7 years ago
Oskar Weigl c95aa7bde3 Revert "input range must be in [0:2PI]"
This reverts commit 6f9820bdd8.
7 years ago
Oskar Weigl dc65724d7c Revert "input range is [0..2Pi]"
This reverts commit 55cfa55b99.
7 years ago
Oskar Weigl dd4779520f fix float rounding issue in sincos index calculation 7 years ago
ARM IT 681c874896 Comment in DCT4 allowed length
https://github.com/ARM-software/CMSIS_5/issues/232
8 years ago
ARM IT 255011718a Pointer increment bug correction
https://github.com/ARM-software/CMSIS_5/issues/165
8 years ago
ARM IT 55cfa55b99 input range is [0..2Pi] 8 years ago
ARM IT 6f9820bdd8 input range must be in [0:2PI] 8 years ago
ARM IT 1c35a7a258 Cancel test
Cancel test
8 years ago
ARM IT ade09c22b7 test
just for test
8 years ago
Jonatan Antoni 2208df0e14 Global MISRA-C Rule 10.6 fix up: Unsigned constant values with U suffix, uppercase instead of lowercase. (Issue #227) 8 years ago
Martin Günther de2ce55a61 Updated CMSIS_DSP. Added ARMv8M support, changed copyrigth note. 9 years ago
Martin Günther adbc210886 updated DSP Lib to V1.5.0. 9 years ago
Martin Günther 7a41fd0818 updated DSP_Lib files arm_sin_cos_f32.c, arm_var_f32.c. 9 years ago
Martin Günther bc3b6d0e2f DSP ARMv8M preparation. 9 years ago
Martin Günther 147cb4e37a Updated CMSIS DSP because of wrong macro ARM_MATH_CM0_FAMILY_FAMILY. 9 years ago
Martin Günther 1c78832899 Updated DSP_Lib changelog and libraries. 9 years ago
Martin Günther 5d3e72e111 Merge pull request #81 from dpalframan/develop
performance optimizations for arm_mat_mult_fast_q15 and arm_mat_mult_fast_q31
9 years ago
Martin Günther 3a1cfe0e0b Corrected arm_conv_partial_q15(), arm_conv_q15(), arm_correlate_q15() for Cortex-M7 based cores. 9 years ago
David Palframan 365d910888 performance optimizations for arm_mat_mult_fast_q15 and arm_mat_mult_fast_q31 9 years ago
Martin Günther 740c6b2cb0 removed doxy warning 10 years ago
unknown 081039f609 Reduced -Wconversion warnings 10 years ago
Martin Günther 7a4fea9150 Renamed folder DSP_Lib to DSP to be consistent with documentation. 10 years ago