In the development for the power10 processor, GCC did not enable using the load vector pair and store vector pair instructions when optimizing things like memory copy. This patch enables using those instructions if -mcpu=future is used.
I have tested these patches on both big endian and little endian PowerPC servers, with no regressions. Can I check these patchs into the trunk? 2025-11-14 Michael Meissner <[email protected]> gcc/ * config/rs6000/rs6000-cpus.def (FUTURE_MASKS_SERVER): Enable using load vector pair and store vector pair instructions for memory copy operations. (POWERPC_MASKS): Make the option for enabling using load vector pair and store vector pair operations set and reset when the PowerPC processor is changed. * gcc/config/rs6000/rs6000.cc (rs6000_machine_from_flags): Disable -mblock-ops-vector-pair from influencing .machine selection. gcc/testsuite/ * gcc.target/powerpc/future-3.c: New test. --- gcc/config/rs6000/rs6000-cpus.def | 9 ++++++++- gcc/config/rs6000/rs6000.cc | 2 +- gcc/testsuite/gcc.target/powerpc/future-3.c | 22 +++++++++++++++++++++ 3 files changed, 31 insertions(+), 2 deletions(-) create mode 100644 gcc/testsuite/gcc.target/powerpc/future-3.c diff --git a/gcc/config/rs6000/rs6000-cpus.def b/gcc/config/rs6000/rs6000-cpus.def index 18547d82811..a0e6745495d 100644 --- a/gcc/config/rs6000/rs6000-cpus.def +++ b/gcc/config/rs6000/rs6000-cpus.def @@ -83,8 +83,14 @@ #define POWER11_MASKS_SERVER (ISA_3_1_MASKS_SERVER \ | OPTION_MASK_POWER11) -/* -mcpu=future flags. */ +/* -mcpu=future flags. + + During the development of the power10 support for GCC, using load/store + vector pair instructions for string operations was turned off by default, + because there was a use case that had really bad performance. Assume this + will be fixed in potential future machines. */ #define FUTURE_MASKS_SERVER (POWER11_MASKS_SERVER \ + | OPTION_MASK_BLOCK_OPS_VECTOR_PAIR \ | OPTION_MASK_FUTURE) /* Flags that need to be turned off if -mno-vsx. */ @@ -115,6 +121,7 @@ /* Mask of all options to set the default isa flags based on -mcpu=<xxx>. */ #define POWERPC_MASKS (OPTION_MASK_ALTIVEC \ + | OPTION_MASK_BLOCK_OPS_VECTOR_PAIR \ | OPTION_MASK_CMPB \ | OPTION_MASK_CRYPTO \ | OPTION_MASK_DFP \ diff --git a/gcc/config/rs6000/rs6000.cc b/gcc/config/rs6000/rs6000.cc index 61623df146e..ca71caff7e0 100644 --- a/gcc/config/rs6000/rs6000.cc +++ b/gcc/config/rs6000/rs6000.cc @@ -5911,7 +5911,7 @@ rs6000_machine_from_flags (void) /* Disable the flags that should never influence the .machine selection. */ flags &= ~(OPTION_MASK_PPC_GFXOPT | OPTION_MASK_PPC_GPOPT | OPTION_MASK_ISEL - | OPTION_MASK_ALTIVEC); + | OPTION_MASK_ALTIVEC | OPTION_MASK_BLOCK_OPS_VECTOR_PAIR); if ((flags & (FUTURE_MASKS_SERVER & ~POWER11_MASKS_SERVER)) != 0) return "future"; diff --git a/gcc/testsuite/gcc.target/powerpc/future-3.c b/gcc/testsuite/gcc.target/powerpc/future-3.c new file mode 100644 index 00000000000..afa22228b96 --- /dev/null +++ b/gcc/testsuite/gcc.target/powerpc/future-3.c @@ -0,0 +1,22 @@ +/* 32-bit doesn't generate vector pair instructions. */ +/* { dg-do compile { target lp64 } } */ +/* { dg-options "-mdejagnu-cpu=future -O2" } */ + +/* Test to see that memcpy will use load/store vector pair with + -mcpu=future. */ + +#ifndef SIZE +#define SIZE 4 +#endif + +extern vector double to[SIZE], from[SIZE]; + +void +copy (void) +{ + __builtin_memcpy (to, from, sizeof (to)); + return; +} + +/* { dg-final { scan-assembler {\mlxvpx?\M} } } */ +/* { dg-final { scan-assembler {\mstxvpx?\M} } } */ -- 2.51.1 -- Michael Meissner, IBM PO Box 98, Ayer, Massachusetts, USA, 01432 email: [email protected]
