J
Joe
Guest
Hi,
I'm implementing routines in C which extensively use 16-bit
multiplication. My Nios core is configured to support the MUL
instruction which exploits the Stratix DSP-blocks . All my
multiplications are resolved by the compiler to call __mulhi3 to use
the MUL instruction. But this involves some calling overhead. I want
to get rid of this overhead and directly use the MUL in my code
without issuing a subroutine call (inlining). Does anybody know how to
configure the system to get inlined multiplications?
Thank you.
Joe
I'm implementing routines in C which extensively use 16-bit
multiplication. My Nios core is configured to support the MUL
instruction which exploits the Stratix DSP-blocks . All my
multiplications are resolved by the compiler to call __mulhi3 to use
the MUL instruction. But this involves some calling overhead. I want
to get rid of this overhead and directly use the MUL in my code
without issuing a subroutine call (inlining). Does anybody know how to
configure the system to get inlined multiplications?
Thank you.
Joe