Other Parts Discussed in Thread: FFTLIB
Tool/software:
Hello Experts,
Please help to clear up some questions:
1. We have noticed only single precision support for FFT in the C7 accelerator. However, for matrix multiplication, there also seems to be support for double precision.
We are wondering if double precision can be supported by the accelerator, perhaps in a future version of the SDK? or is there an inherent issue in supporting FFT double precision in the accelerator?
2. We are wondering if there is any report by TI for FFT SQNR in the accelerator.
3. Is there additional documentation or app note for FFT of the accelerator? aside from <SDK folder>/source/fftlib/docs ?
4. We are wondering if there is a bit-exact simulation we could use in PC to test the FFT accelerator.
Thank you in advance,
Guy
