I use Sharc's processors ADSP-2137x and I want to use 40-bit float point.
When I use 40-bit float point I will use assembler. Otherwise I use C/C++.
So I must write my interrupt handler routine because I must save Dregs.
Course, I can use standart interrupt dispatchers but stack must have 48-bit width.
(Why do manual for compiler mention that stack must have 48-bit width is absent?)
So I waste my memory space for saving another registers (non Dregs).
Hence I must write my self interrupt handler routine which will be used memory segment.
I want use one DAG register which point to 48-bit memory and deny compiler by used this register.
Also I will use standart interrupt dispatchers for superfast interrupt.
So how can I combine standart interrupt dispatchers and my.
I try use crt_hdr.asm (now I don't remember file name. it locate in next dir src\crt???) and exclude *hdr.doj file from ldr file, but compiler out mistake
When do compiler C/C++ of VDSP++ normally support main Sharc's advantage (40-bit float point)