* RE: [PATCH][AArch64] Fix Cortex-A53 shift costs
@ 2015-04-27 14:05 Wilco Dijkstra
0 siblings, 0 replies; 4+ messages in thread
From: Wilco Dijkstra @ 2015-04-27 14:05 UTC (permalink / raw)
To: gcc-patches
ping
> -----Original Message-----
> From: Wilco Dijkstra [mailto:wdijkstr@arm.com]
> Sent: 05 March 2015 14:49
> To: gcc-patches@gcc.gnu.org
> Subject: [PATCH][AArch64] Fix Cortex-A53 shift costs
>
> This patch fixes the shift costs for Cortex-A53 so they are more accurate - immediate shifts
> use SBFM/UBFM which takes 2 cycles, register controlled shifts take 1 cycle. Bootstrap and
> regression OK.
>
> ChangeLog:
> 2015-03-05 Wilco Dijkstra <wdijkstr@arm.com>
>
> * gcc/config/arm/aarch-cost-tables.h (cortexa53_extra_costs):
> Make Cortex-A53 shift costs more accurate.
>
> ---
> gcc/config/arm/aarch-cost-tables.h | 8 ++++----
> 1 file changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/gcc/config/arm/aarch-cost-tables.h b/gcc/config/arm/aarch-cost-tables.h
> index 05e96a9..6bb8ede 100644
> --- a/gcc/config/arm/aarch-cost-tables.h
> +++ b/gcc/config/arm/aarch-cost-tables.h
> @@ -130,12 +130,12 @@ const struct cpu_cost_table cortexa53_extra_costs =
> 0, /* arith. */
> 0, /* logical. */
> COSTS_N_INSNS (1), /* shift. */
> - COSTS_N_INSNS (2), /* shift_reg. */
> + 0, /* shift_reg. */
> COSTS_N_INSNS (1), /* arith_shift. */
> - COSTS_N_INSNS (2), /* arith_shift_reg. */
> + COSTS_N_INSNS (1), /* arith_shift_reg. */
> COSTS_N_INSNS (1), /* log_shift. */
> - COSTS_N_INSNS (2), /* log_shift_reg. */
> - 0, /* extend. */
> + COSTS_N_INSNS (1), /* log_shift_reg. */
> + COSTS_N_INSNS (1), /* extend. */
> COSTS_N_INSNS (1), /* extend_arith. */
> COSTS_N_INSNS (1), /* bfi. */
> COSTS_N_INSNS (1), /* bfx. */
> --
> 1.9.1
^ permalink raw reply [flat|nested] 4+ messages in thread
* [PATCH][AArch64] Fix Cortex-A53 shift costs
@ 2015-03-05 14:49 Wilco Dijkstra
[not found] ` <CAFqB+Pzsi73v5nVJgwuTFwp4rKGw0=Vj7YDW34KeV-G3V8ayOQ@mail.gmail.com>
0 siblings, 1 reply; 4+ messages in thread
From: Wilco Dijkstra @ 2015-03-05 14:49 UTC (permalink / raw)
To: gcc-patches
This patch fixes the shift costs for Cortex-A53 so they are more accurate - immediate shifts use
SBFM/UBFM which takes 2 cycles, register controlled shifts take 1 cycle. Bootstrap and regression
OK.
ChangeLog:
2015-03-05 Wilco Dijkstra <wdijkstr@arm.com>
* gcc/config/arm/aarch-cost-tables.h (cortexa53_extra_costs):
Make Cortex-A53 shift costs more accurate.
---
gcc/config/arm/aarch-cost-tables.h | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/gcc/config/arm/aarch-cost-tables.h b/gcc/config/arm/aarch-cost-tables.h
index 05e96a9..6bb8ede 100644
--- a/gcc/config/arm/aarch-cost-tables.h
+++ b/gcc/config/arm/aarch-cost-tables.h
@@ -130,12 +130,12 @@ const struct cpu_cost_table cortexa53_extra_costs =
0, /* arith. */
0, /* logical. */
COSTS_N_INSNS (1), /* shift. */
- COSTS_N_INSNS (2), /* shift_reg. */
+ 0, /* shift_reg. */
COSTS_N_INSNS (1), /* arith_shift. */
- COSTS_N_INSNS (2), /* arith_shift_reg. */
+ COSTS_N_INSNS (1), /* arith_shift_reg. */
COSTS_N_INSNS (1), /* log_shift. */
- COSTS_N_INSNS (2), /* log_shift_reg. */
- 0, /* extend. */
+ COSTS_N_INSNS (1), /* log_shift_reg. */
+ COSTS_N_INSNS (1), /* extend. */
COSTS_N_INSNS (1), /* extend_arith. */
COSTS_N_INSNS (1), /* bfi. */
COSTS_N_INSNS (1), /* bfx. */
--
1.9.1
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2015-05-01 13:09 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2015-04-27 14:05 [PATCH][AArch64] Fix Cortex-A53 shift costs Wilco Dijkstra
-- strict thread matches above, loose matches on Subject: below --
2015-03-05 14:49 Wilco Dijkstra
[not found] ` <CAFqB+Pzsi73v5nVJgwuTFwp4rKGw0=Vj7YDW34KeV-G3V8ayOQ@mail.gmail.com>
2015-05-01 13:07 ` Wilco Dijkstra
2015-05-01 13:09 ` Kyrill Tkachov
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).