diff options
| author | Andreas Krebbel <Andreas.Krebbel@de.ibm.com> | 2010-10-05 10:39:48 +0000 |
|---|---|---|
| committer | Andreas Krebbel <krebbel@gcc.gnu.org> | 2010-10-05 10:39:48 +0000 |
| commit | 65b1d8ea3ed53d03f92a614c93faf637fa708772 (patch) | |
| tree | f525e5ae1b824e7da9d37e83b5c4fb9c6e95b3c0 /gcc/config.gcc | |
| parent | 6e5723262255ea4d43d14d7c959e4dcc28cd01c3 (diff) | |
| download | gcc-65b1d8ea3ed53d03f92a614c93faf637fa708772.zip gcc-65b1d8ea3ed53d03f92a614c93faf637fa708772.tar.gz gcc-65b1d8ea3ed53d03f92a614c93faf637fa708772.tar.bz2 | |
s390.c (z196_cost): New.
2010-10-05 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
Christian Borntraeger <Christian.Borntraeger@de.ibm.com>
* gcc/config/s390/s390.c (z196_cost): New.
(s390_handle_arch_option): Add -march=z196.
(s390_option_override): Use the new cost function and use the z10
defaults also for z196.
(legitimate_reload_constant_p): Adjust comment.
(legitimate_reload_fp_constant_p): New function.
(s390_preferred_reload_class): Distingiush between FP and Int constants.
(preferred_la_operand_p): Avoid la with index on z196.
(s390_adjust_priority): Trigger also for z196.
(s390_issue_rate): Issue rate for z196 is 3.
(s390_z10_fix_long_loop_prediction): Rename to ...
(s390_fix_long_loop_prediction): ... this.
(s390_reorg): Apply the z10 adjustments also for z196.
(s390_loop_unroll_adjust): Do this also for z196.
* gcc/config/s390/s390.h (enum processor_type): Add PROCESSOR_2817_Z196.
(enum processor_flags): Add PF_Z196.
(TARGET_AVOID_CMP_AND_BRANCH): New macro.
(TARGET_CPU_Z196, TARGET_Z196): New macros.
* gcc/config.gcc: Enable z196 for --with-arch.
* gcc/config/s390/2817.md: New file.
* gcc/config/s390/2084.md: New type for multiply and add.
* gcc/config/s390/2097.md: Likewise.
* gcc/config/s390/s390.md (UNSPEC_POPCNT, UNSPEC_COPYSIGN)
(UNSPECV_ATOMIC_OP): New constants.
(fmadddf, fmaddsf): New values for type attribute.
(z196prop): New insn attribute.
(cpu, cpu_facility): Add z196.
(ATOMIC_Z196): New code iterator.
(noxa): New code attribute.
(gk): New mode attribute.
(*mov<mode>_64, *mov<mode>_31, *mov<mode>_64dfp, mov<mode>):
Support load zero for fp constants.
(fixuns_truncdddi2, fixuns_trunctddi2)
(fixuns_trunc<BFP:mode><GPR:mode>2): Use the standard rtx pattern
for z196.
(fixuns_trunc<mode>si2, mov<mode>cc, popcountdi2, popcountsi2)
(popcounthi2, popcountqi2): New expander.
(*fixuns_trunc<FP:mode><GPR:mode>2_z196, floatsi<mode>2)
(floatuns<GPR:mode><FP:mode>2, *mov<mode>cc, sync_<atomic><mode>)
(sync_old_<atomic><mode>, *popcount<mode>, copysign<mode>3): New
insn definition.
(add<mode>3, *add<mode>3_carry1_cc, *add<mode>3_carry1_cconly)
(*add<mode>3_carry2_cc, *add<mode>3_carry2_cconly, *add<mode>3_cc)
(*add<mode>3_cconly, *add<mode>3_cconly2, *add<mode>3_imm_cc, *sub<mode>3)
(*sub<mode>3_borrow_cc, *sub<mode>3_borrow_cconly, *sub<mode>3_cc)
(*sub<mode>3_cc2, *sub<mode>3_cconly, *sub<mode>3_cconly2)
(*anddi3_cc, *anddi3_cconly, *anddi3, *andsi3_cc, *andsi3_cconly)
(*andsi3_zarch, *andsi3_esa, *andhi3_zarch, *andqi3_zarch, *iordi3_cc)
(*iordi3_cconly, *iordi3, *iorsi3_cc, *iorsi3_cconly, *iorsi3_zarch)
(*iorhi3_zarch, *iorqi3_zarch, *xordi3_cc, *xordi3_cconly, *xordi3)
(*xorsi3_cc, *xorsi3_cconly, *xorsi3, *xorhi3, *xorqi3, *<shift><mode>3)
(*<shift><mode>3_and, *ashr<mode>3_cc, *ashr<mode>3_cconly, *ashr<mode>3)
(*ashr<mode>3_cc_and, *ashr<mode>3_cconly_and, *ashr<mode>3_and):
Support new z196 instructions.
Co-Authored-By: Christian Borntraeger <Christian.Borntraeger@de.ibm.com>
From-SVN: r164985
Diffstat (limited to 'gcc/config.gcc')
| -rw-r--r-- | gcc/config.gcc | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/gcc/config.gcc b/gcc/config.gcc index b1e86a0..b813ca3 100644 --- a/gcc/config.gcc +++ b/gcc/config.gcc @@ -3374,7 +3374,7 @@ case "${target}" in for which in arch tune; do eval "val=\$with_$which" case ${val} in - "" | g5 | g6 | z900 | z990 | z9-109 | z9-ec | z10) + "" | g5 | g6 | z900 | z990 | z9-109 | z9-ec | z10 | z196) # OK ;; *) |