On Tue, Dec 06, 2016 at 03:10:50PM +0000, Wilco Dijkstra wrote:
>     
> 
> ping

OK.

This has been on the list since before Stage 1 closed and should be low
risk outside of code using the SHA1H intrinsics.

Though, given where we are in the release cycle, please give
Richard/Marcus 24 hours to object before pushing it.

Thanks,
James

> From: Wilco Dijkstra
> Sent: 25 October 2016 18:08
> To: GCC Patches
> Cc: nd
> Subject: [PATCH][AArch64] Improve SHA1 scheduling
>     
> SHA1H instructions may be scheduled after a SHA1C instruction
> that uses the same input register.  However SHA1C updates its input,
> so if SHA1H is scheduled after it, it requires an extra move.
> Increase the priority of SHA1H to ensure it gets scheduled
> earlier, avoiding the move.
> 
> Is this something the generic scheduler could do automatically for
> instructions with RMW operands?
> 
> Passes bootstrap & regress. OK for commit?
> 
> ChangeLog:
> 2016-10-25  Wilco Dijkstra  <wdijk...@arm.com>
> 
>         * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
>         New function.
>         (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
> --
> diff --git a/gcc/config/aarch64/aarch64.c b/gcc/config/aarch64/aarch64.c
> index 
> 9b2f9cb19343828dc39e9950ebbefe941521942a..2b25bd1bdd6f4e7737f8e04c3b3684cdff6c4b80
>  100644
> --- a/gcc/config/aarch64/aarch64.c
> +++ b/gcc/config/aarch64/aarch64.c
> @@ -13668,6 +13668,26 @@ aarch64_sched_fusion_priority (rtx_insn *insn, int 
> max_pri,
>    return;
>  }
>  
> +/* Implement the TARGET_SCHED_ADJUST_PRIORITY hook.
> +   Adjust priority of sha1h instructions so they are scheduled before
> +   other SHA1 instructions.  */
> +
> +static int
> +aarch64_sched_adjust_priority (rtx_insn *insn, int priority)
> +{
> +  rtx x = PATTERN (insn);
> +
> +  if (GET_CODE (x) == SET)
> +    {
> +      x = SET_SRC (x);
> +
> +      if (GET_CODE (x) == UNSPEC && XINT (x, 1) == UNSPEC_SHA1H)
> +       return priority + 10;
> +    }
> +
> +  return priority;
> +}
> +
>  /* Given OPERANDS of consecutive load/store, check if we can merge
>     them into ldp/stp.  LOAD is true if they are load instructions.
>     MODE is the mode of memory operands.  */
> @@ -14431,6 +14451,9 @@ aarch64_optab_supported_p (int op, machine_mode 
> mode1, machine_mode,
>  #undef TARGET_CAN_USE_DOLOOP_P
>  #define TARGET_CAN_USE_DOLOOP_P can_use_doloop_if_innermost
>  
> +#undef TARGET_SCHED_ADJUST_PRIORITY
> +#define TARGET_SCHED_ADJUST_PRIORITY aarch64_sched_adjust_priority
> +
>  #undef TARGET_SCHED_MACRO_FUSION_P
>  #define TARGET_SCHED_MACRO_FUSION_P aarch64_macro_fusion_p
>  
> 
>             

Reply via email to