updated cfft_f32 with CM85 / slothy optimized radix-4 loops

ARM-software · Mar 3, 2023 · 817bb57 · 817bb57
1 parent 538ffca
commit 817bb57
Show file tree

Hide file tree

Showing 4 changed files with 413 additions and 199 deletions.
diff --git a/...RM-Optimized/DSP/Source/FilteringFunctions/arm_biquad_cascade_df1_f32/README.md b/...RM-Optimized/DSP/Source/FilteringFunctions/arm_biquad_cascade_df1_f32/README.md
@@ -3,14 +3,14 @@
 ![CMSIS](http://www.keil.com/pack/doc/CMSIS/General/html/CMSIS_Logo_Final.png)
 
 
-[CMSIS-DSP](https://github.com/ARM-software/CMSIS_5/tree/develop/CMSIS/DSP) with ARMv8.1M **Helium** assembly
+[CMSIS-DSP](https://github.com/ARM-software/CMSIS_5/tree/develop/CMSIS/DSP) with **Helium** and **Neon** assembly
 
 
 Single-precision DF1 Biquad processing function
 
-- This version allows boosting F32 DF1 Biquad performance when using compilers having suboptimal Helium intrinsic code generation and can be sustitued to CMSIS DSP intrinsic variant.
-- It also provide efficient Neon Implementations
-- Helium and Neon functions have respectively _mve and _neon prefix
+- This version allows boosting DF1 Biquad performance when using compilers having suboptimal Helium intrinsic code generation and can be sustitued to CMSIS DSP intrinsic variant.
+- Efficient Neon Implementations are also provided. In order cores should use ASM variants while out of order cores can rely on intrinsic variant.
+- Helium and Neon functions have respectively *_mve* and *_neon* prefix
 
 [see API details](https://www.keil.com/pack/doc/CMSIS/DSP/html/group__BiquadCascadeDF1.html)
 

diff --git a/Kernels/ARM-Optimized/DSP/Source/TransformFunctions/arm_cfft_f32/README.md b/Kernels/ARM-Optimized/DSP/Source/TransformFunctions/arm_cfft_f32/README.md
@@ -20,3 +20,8 @@ This version allows boosting CFFT F32 performance when using compilers having su
  arm-none-eabi-gcc -DUSE_ASM -mcpu=cortex-m55 -O3 -mfloat-abi=hard -flax-vector-conversions -c arm_cfft_f32.c -o arm_cfft_f32.o -I$(CMSIS5ROOT)/CMSIS/DSP/PrivateInclude/ -I $(CMSIS5ROOT)/CMSIS/DSP/Include/ -I $(CMSIS5ROOT)/CMSIS/Core/Include/
  ```
 
+ - Arm Compiler 6.19 or above example (CM85 build)
+
+```cpp
+armclang -target arm-arm-none-eabi -mthumb -std=gnu99 -mcpu=cortex-m85 -Ofast -mfloat-abi=hard -DUSE_ASM -DARM_CM85_OPT -c arm_cfft_f32.c -o arm_cfft_f32.o -I$(CMSIS5ROOT)/CMSIS/DSP/PrivateInclude/ -I $(CMSIS5ROOT)/CMSIS/DSP/Include/ -I $(CMSIS5ROOT)/CMSIS/Core/Include/
+ ```