Ciao Niels,
Il 2021-10-05 20:56 ni...@lysator.liu.se ha scritto:
Marco Bodrato <bodr...@mail.dm.unipi.it> writes:
mov %r11, -16(rp)
mov %r10, %r11
jmp L(one)
I had hoped this jump and preceding instructions could be eliminated,
to
get a structure like
ja L(two)
jz L(one)
L(nul): (no jumps to this label left)
...
But might need other move instructions, to get the right data into the
right registers?
Maybe with a couple of move with computed address, and with the same
value in r10 and r11 we can get rid of the jump.
The difference, wrt the last code I sent, follows.
***************
*** 122,135 ****
cmp $1, R32(n)
ja L(two)
! jnz L(nul)
!
! mov -8(ap), %rax
! mov %r11, -16(rp)
mov %r10, %r11
! jmp L(one)
! L(nul): mov -16(ap), %rax
! mov %r11, -24(rp)
! mul %r8
add %rax, %r10
mov -16(bp), %rax
--- 122,131 ----
cmp $1, R32(n)
ja L(two)
! mov -16(ap,n,8), %rax
! mov %r11, -24(rp,n,8)
mov %r10, %r11
! jz L(one)
! L(nul): mul %r8
add %rax, %r10
mov -16(bp), %rax
So I think your version is an improvement as is, and perhaps not worth
the effort to try to eliminate a few more instructions if this rather
obscure function.
I agree. Let's choose between the two last versions, and I'll push it.
Ĝis,
m
_______________________________________________
gmp-devel mailing list
gmp-devel@gmplib.org
https://gmplib.org/mailman/listinfo/gmp-devel
