On Mon, Feb 13, 2023 at 10:43 AM Jeff Law <[email protected]> wrote:
>
>
>
> On 2/13/23 10:32, Richard Sandiford via Gcc-patches wrote:
> > Andrew Pinski via Gcc-patches <[email protected]> writes:
> >> On Fri, Feb 10, 2023 at 2:47 PM Philipp Tomsich
> >> <[email protected]> wrote:
> >>>
> >>> Some architectures, as it the case on RISC-V with the proposed
> >>> ZiCondOps and the vendor-defined XVentanaCondOps, define a
> >>> conditional-zero instruction that is equivalent to:
> >>> - the positive form: rd = (rc != 0) ? rs : 0
> >>> - the negated form: rd = (rc == 0) ? rs : 0
> >>>
> >>> While noce_try_store_flag_mask will somewhat work for this case, it
> >>> will generate a number of atomic RTX that will misdirect the cost
> >>> calculation and may be too long (i.e., 4 RTX and more) to successfully
> >>> merge at combine-time.
> >>
> >> Can you expand on this? Especially when there are patterns that use
> >> (if_then_else) already.
> >>
> >>>
> >>> Instead, we add two new transforms that attempt to build up what we
> >>> define as the canonical form of a conditional-zero expression:
> >>>
> >>> (set (match_operand 0 "register_operand" "=r")
> >>> (and (neg (eq_or_ne (match_operand 1 "register_operand" "r")
> >>> (const_int 0)))
> >>> (match_operand 2 "register_operand" "r")))
> >>
> >> Again why are you not using:
> >> (set (reg) (if_then_else (eq_ne (reg) (const_int 0)) (reg) (const_int 0)))
> >> Form instead of the really bad "canonical" form of the above?
> >
> > I don't think one form is inherently better than the other if we think
> > about just this one case. But I agree that the if_then_else form is
> > currently the canonical form for the operation, and extends more
> > naturally to the general case. AArch64 already matches specifically
> > for it (with xzr providing the zero value).
> The more I think about it, the more I prefer the if-then-else form. My
> biggest hesitation with getting behind one form or the other is a lack
> of knowledge about which is likely better interpreted by simplify-rtx
> and friends -- though it may not matter much in practice.
In the case of IF_THEN_ELSE, combine tries a few things:
/* If this is a simple operation applied to an IF_THEN_ELSE, try
applying it to the arms of the IF_THEN_ELSE. This often simplifies
things. Check for cases where both arms are testing the same
condition.
Also see simplify_if_then_else inside combine.cc.
I don't think the first one could be done not using IF_THEN_ELSE.
Thanks,
Andrew Pinski
>
> Jeff
