RE: [PATCH v2] aarch64: disable LDP via tuning structure for -mcpu=ampere1/1a

[Kyrylo Tkachov <Kyrylo.Tkachov@arm.com>]

> -----Original Message-----
> From: Philipp Tomsich <philipp.tomsich@vrull.eu>
> Sent: Friday, April 14, 2023 7:06 PM
> To: gcc-patches@gcc.gnu.org
> Cc: Kyrylo Tkachov <Kyrylo.Tkachov@arm.com>; Philipp Tomsich
> <philipp.tomsich@vrull.eu>; Di Zhao <di.zhao@amperecomputing.com>
> Subject: [PATCH v2] aarch64: disable LDP via tuning structure for -
> mcpu=ampere1/1a
> 
> AmpereOne (-mcpu=ampere1) breaks LDP instructions into two uops.
> Given the chance that this causes instructions to slip into the next
> decoding cycle and the additional overheads when handling
> cacheline-crossing LDP instructions, we disable the generation of LDP
> isntructions through the tuning structure from instruction combining
> (such as in peephole2).
> 
> Given the code-density benefits in builtins and prologue/epilogue
> expansion, we allow LDPs there.
> 
> This commit:
>  * adds a new tuning option AARCH64_EXTRA_TUNE_NO_LDP_COMBINE
>  * allows -moverride=tune=... to override this
> 
> These changes are benchmark-driven, yielding the following changes
> (with a net-overall improvement):
>    503.bwaves_r.      -0.88%
>    507.cactuBSSN_r     0.35%
>    508.namd_r          3.09%
>    510.parest_r       -2.99%
>    511.povray_r        5.54%
>    519.lbm_r          15.83%
>    521.wrf_r           0.56%
>    526.blender_r       2.47%
>    527.cam4_r          0.70%
>    538.imagick_r       0.00%
>    544.nab_r          -0.33%
>    549.fotonik3d_r.   -0.42%
>    554.roms_r          0.00%
>    -------------------------
>    = total             1.79%
> 
> Signed-off-by: Philipp Tomsich <philipp.tomsich@vrull.eu>
> Co-Authored-By: Di Zhao <di.zhao@amperecomputing.com>

Ok.
Thanks,
Kyrill

> 
> gcc/ChangeLog:
> 
> 	* config/aarch64/aarch64-tuning-flags.def
> (AARCH64_EXTRA_TUNING_OPTION):
> 	Add AARCH64_EXTRA_TUNE_NO_LDP_COMBINE.
> 	* config/aarch64/aarch64.cc (aarch64_operands_ok_for_ldpstp):
> 	Check for the above tuning option when processing loads.
> 
> gcc/testsuite/ChangeLog:
> 
> 	* gcc.target/aarch64/ampere1-no_ldp_combine.c: New test.
> 
> ---
> 
> Changes in v2:
> - apply both to -mcpu=ampere1 and -mcpu=ampere1a
> - add TODO: tag, per discussions on the mailing list
> - add testcase
> 
>  gcc/config/aarch64/aarch64-tuning-flags.def    |  3 +++
>  gcc/config/aarch64/aarch64.cc                  | 18 ++++++++++++++++--
>  .../aarch64/ampere1-no_ldp_combine.c           | 11 +++++++++++
>  3 files changed, 30 insertions(+), 2 deletions(-)
>  create mode 100644 gcc/testsuite/gcc.target/aarch64/ampere1-
> no_ldp_combine.c
> 
> diff --git a/gcc/config/aarch64/aarch64-tuning-flags.def
> b/gcc/config/aarch64/aarch64-tuning-flags.def
> index 712895a5263..52112ba7c48 100644
> --- a/gcc/config/aarch64/aarch64-tuning-flags.def
> +++ b/gcc/config/aarch64/aarch64-tuning-flags.def
> @@ -44,6 +44,9 @@ AARCH64_EXTRA_TUNING_OPTION
> ("cheap_shift_extend", CHEAP_SHIFT_EXTEND)
>  /* Disallow load/store pair instructions on Q-registers.  */
>  AARCH64_EXTRA_TUNING_OPTION ("no_ldp_stp_qregs",
> NO_LDP_STP_QREGS)
> 
> +/* Disallow load-pair instructions to be formed in combine/peephole.  */
> +AARCH64_EXTRA_TUNING_OPTION ("no_ldp_combine",
> NO_LDP_COMBINE)
> +
>  AARCH64_EXTRA_TUNING_OPTION ("rename_load_regs",
> RENAME_LOAD_REGS)
> 
>  AARCH64_EXTRA_TUNING_OPTION ("cse_sve_vl_constants",
> CSE_SVE_VL_CONSTANTS)
> diff --git a/gcc/config/aarch64/aarch64.cc b/gcc/config/aarch64/aarch64.cc
> index f4ef22ce02f..0f04ab9fba0 100644
> --- a/gcc/config/aarch64/aarch64.cc
> +++ b/gcc/config/aarch64/aarch64.cc
> @@ -1933,7 +1933,7 @@ static const struct tune_params ampere1_tunings =
>    2,	/* min_div_recip_mul_df.  */
>    0,	/* max_case_values.  */
>    tune_params::AUTOPREFETCHER_WEAK,	/* autoprefetcher_model.  */
> -  (AARCH64_EXTRA_TUNE_NONE),		/* tune_flags.  */
> +  (AARCH64_EXTRA_TUNE_NO_LDP_COMBINE),	/* tune_flags.  */
>    &ampere1_prefetch_tune
>  };
> 
> @@ -1971,7 +1971,7 @@ static const struct tune_params ampere1a_tunings
> =
>    2,	/* min_div_recip_mul_df.  */
>    0,	/* max_case_values.  */
>    tune_params::AUTOPREFETCHER_WEAK,	/* autoprefetcher_model.  */
> -  (AARCH64_EXTRA_TUNE_NONE),		/* tune_flags.  */
> +  (AARCH64_EXTRA_TUNE_NO_LDP_COMBINE),	/* tune_flags.  */
>    &ampere1_prefetch_tune
>  };
> 
> @@ -26053,6 +26053,20 @@ aarch64_operands_ok_for_ldpstp (rtx
> *operands, bool load,
>    enum reg_class rclass_1, rclass_2;
>    rtx mem_1, mem_2, reg_1, reg_2;
> 
> +  /* Allow the tuning structure to disable LDP instruction formation
> +     from combining instructions (e.g., in peephole2).
> +     TODO: Implement fine-grained tuning control for LDP and STP:
> +	   1. control policies for load and store separately;
> +	   2. support the following policies:
> +	      - default (use what is in the tuning structure)
> +	      - always
> +	      - never
> +	      - aligned (only if the compiler can prove that the
> +		load will be aligned to 2 * element_size)  */
> +  if (load && (aarch64_tune_params.extra_tuning_flags
> +	       & AARCH64_EXTRA_TUNE_NO_LDP_COMBINE))
> +    return false;
> +
>    if (load)
>      {
>        mem_1 = operands[1];
> diff --git a/gcc/testsuite/gcc.target/aarch64/ampere1-no_ldp_combine.c
> b/gcc/testsuite/gcc.target/aarch64/ampere1-no_ldp_combine.c
> new file mode 100644
> index 00000000000..bc871f4481d
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/aarch64/ampere1-no_ldp_combine.c
> @@ -0,0 +1,11 @@
> +/* { dg-options "-O3 -mtune=ampere1" } */
> +
> +long
> +foo (long a[])
> +{
> +  return a[0] + a[1];
> +}
> +
> +/* We should see two ldrs instead of one ldp. */
> +/* { dg-final { scan-assembler {\tldr\t} } } */
> +/* { dg-final { scan-assembler-not {\tldp\t} } } */
> --
> 2.34.1