| Message ID | 52EA5830.8020006@arm.com |
|---|---|
| State | New |
| Headers | show |
On 30 January 2014 13:48, Kyrill Tkachov <kyrylo.tkachov@arm.com> wrote: > Hi all, > > This patch wires up the aarch64 backend to use the Cortex-A57 rtx costs > table that is proposed at > http://gcc.gnu.org/ml/gcc-patches/2014-01/msg01954.html OK if release manager agrees. /Marcus
On 10 February 2014 09:55, Marcus Shawcroft <marcus.shawcroft@gmail.com> wrote: > On 30 January 2014 13:48, Kyrill Tkachov <kyrylo.tkachov@arm.com> wrote: >> Hi all, >> >> This patch wires up the aarch64 backend to use the Cortex-A57 rtx costs >> table that is proposed at >> http://gcc.gnu.org/ml/gcc-patches/2014-01/msg01954.html > > OK if release manager agrees. > > /Marcus Jakub, are you OK with us taking this patch in stage4 ? This patch builds on the ARM patch http://gcc.gnu.org/ml/gcc-patches/2014-02/msg00403.html and like the ARM patch it is none default tuning. /Marcus
On Tue, Feb 11, 2014 at 03:03:32PM +0000, Marcus Shawcroft wrote: > On 10 February 2014 09:55, Marcus Shawcroft <marcus.shawcroft@gmail.com> wrote: > > On 30 January 2014 13:48, Kyrill Tkachov <kyrylo.tkachov@arm.com> wrote: > >> Hi all, > >> > >> This patch wires up the aarch64 backend to use the Cortex-A57 rtx costs > >> table that is proposed at > >> http://gcc.gnu.org/ml/gcc-patches/2014-01/msg01954.html > > > > OK if release manager agrees. > > > > /Marcus > > Jakub, are you OK with us taking this patch in stage4 ? > > This patch builds on the ARM patch > http://gcc.gnu.org/ml/gcc-patches/2014-02/msg00403.html and like the > ARM patch it is none default tuning. Ok. Jakub
commit 780da554b322ab1c54cc662849950adb1181cacb
Author: Kyrylo Tkachov <kyrylo.tkachov@arm.com>
Date: Fri Jan 24 11:36:02 2014 +0000
[AArch64] Wire up A57 costs
diff --git a/gcc/config/aarch64/aarch64-cores.def b/gcc/config/aarch64/aarch64-cores.def
index a41d7d9..4db63d6 100644
--- a/gcc/config/aarch64/aarch64-cores.def
+++ b/gcc/config/aarch64/aarch64-cores.def
@@ -35,8 +35,8 @@
/* V8 Architecture Processors. */
AARCH64_CORE("cortex-a53", cortexa53, cortexa53, 8, AARCH64_FL_FPSIMD, cortexa53)
-AARCH64_CORE("cortex-a57", cortexa15, cortexa15, 8, AARCH64_FL_FPSIMD, generic)
+AARCH64_CORE("cortex-a57", cortexa15, cortexa15, 8, AARCH64_FL_FPSIMD, cortexa57)
/* V8 big.LITTLE implementations. */
-AARCH64_CORE("cortex-a57.cortex-a53", cortexa57cortexa53, cortexa53, 8, AARCH64_FL_FPSIMD, generic)
+AARCH64_CORE("cortex-a57.cortex-a53", cortexa57cortexa53, cortexa53, 8, AARCH64_FL_FPSIMD, cortexa57)
diff --git a/gcc/config/aarch64/aarch64.c b/gcc/config/aarch64/aarch64.c
index 5867f75..aca4366 100644
--- a/gcc/config/aarch64/aarch64.c
+++ b/gcc/config/aarch64/aarch64.c
@@ -235,6 +235,16 @@ static const struct tune_params cortexa53_tunings =
NAMED_PARAM (issue_rate, 2)
};
+static const struct tune_params cortexa57_tunings =
+{
+ &cortexa57_extra_costs,
+ &generic_addrcost_table,
+ &generic_regmove_cost,
+ &generic_vector_cost,
+ NAMED_PARAM (memmov_cost, 4),
+ NAMED_PARAM (issue_rate, 3)
+};
+
/* A processor implementing AArch64. */
struct processor
{
Hi all, This patch wires up the aarch64 backend to use the Cortex-A57 rtx costs table that is proposed at http://gcc.gnu.org/ml/gcc-patches/2014-01/msg01954.html It also sets the generic tuning to use the Cortex-A57 costs since they are more sensible for aarch64 systems than the generic ones that were originally written with aarch32 code generation in mind. Tested aarch64-none-elf on a model. Ok for trunk once the prerequisite goes in? Thanks, Kyrill 2014-01-30 Kyrylo Tkachov <kyrylo.tkachov@arm.com> * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57 tuning struct. (cortex-a57.cortex-a53): Likewise. * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.