New FFT initialization functions to make it easier for the linker
to optimize the code size when the library is used as a static
library and can't be rebuilt.
Compilation directives for code size optimizations have been removed.
In relation with issue #58
When HELIUM, MVEF or MVEI are used as cmake options, the option
LAXVECTORCONVERSIONS can be set to ON or OFF to enable / disable
lax vector conversions in GCC or ArmClang.
Currently, CMSIS-DSP won't build if lax vector conversions are disabled.
The option is ON by default.
Some CMSIS-DSP code does not explicitly convert between vectors of
different numbers and types of elements. While some other compilers
default to implicitly converting vectors, gcc does not do this unless
explicitly enabled.
Signed-off-by: Lingkai Dong <lingkai.dong@arm.com>
Signed-off-by: Lingkai Dong <lingkai.dong@arm.com>
It is to provide more control and to disable
intrinsics version of some kernels when the C code is simple enough to
benefit from vectorization.
Since the result is compiler dependent, this option is optional.
(And it is not yet added to the source files).
Compilation flags may have to be tuned (-O3 for instance ...)
(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.
Christophe Favergeon
Lingkai Dong
ClaudioMartino