For integer vector comparisons without XOP before AVX512{F,VL} we are
constrained by only GT and EQ being supported in HW.
For GTU we play tricks to implement it using GT or unsigned saturating
subtraction, for LT/LTU we swap the operands and thus turn it into
GT/GTU. For LE/LEU we handle it by using GT/GTU and negating the
result and for GE/GEU by using GT/GTU on swapped operands and negating
the result.
If the second operand is a CONST_VECTOR, we can usually do better though,
we can avoid the negation. For LE/LEU cst by doing LT/LTU cst+1 (and
then cst+1 GT/GTU x) and for GE/GEU cst by doing GT/GTU cst-1, provided
there is no wrap-around on those cst+1 or cst-1.
GIMPLE canonicalizes x < cst to x <= cst-1 etc. (the rule is smaller
absolute value on constant), but only for scalars or uniform vectors,
so in some cases this undoes that canonicalization in order to avoid
the extra negation, but it handles also non-uniform constants.
E.g. with -mavx2 the testcase assembly difference is:
- movl $47, %eax
+ movl $48, %eax
vmovdqa %xmm0, %xmm1
vmovd %eax, %xmm0
vpbroadcastb %xmm0, %xmm0
- vpminsb %xmm0, %xmm1, %xmm0
- vpcmpeqb %xmm1, %xmm0, %xmm0
+ vpcmpgtb %xmm1, %xmm0, %xmm0
and
- vmovdqa %xmm0, %xmm1
- vmovdqa .LC1(%rip), %xmm0
- vpminsb %xmm1, %xmm0, %xmm1
- vpcmpeqb %xmm1, %xmm0, %xmm0
+ vpcmpgtb .LC1(%rip), %xmm0, %xmm0
while with just SSE2:
- pcmpgtb .LC0(%rip), %xmm0
- pxor %xmm1, %xmm1
- pcmpeqb %xmm1, %xmm0
+ movdqa %xmm0, %xmm1
+ movdqa .LC0(%rip), %xmm0
+ pcmpgtb %xmm1, %xmm0
and
- movdqa %xmm0, %xmm1
- movdqa .LC1(%rip), %xmm0
- pcmpgtb %xmm1, %xmm0
- pxor %xmm1, %xmm1
- pcmpeqb %xmm1, %xmm0
+ pcmpgtb .LC1(%rip), %xmm0
2022-11-08 Jakub Jelinek <jakub@redhat.com>
PR target/107546
* config/i386/predicates.md (vector_or_const_vector_operand): New
predicate.
* config/i386/sse.md (vec_cmp<mode><sseintvecmodelower>,
vec_cmpv2div2di, vec_cmpu<mode><sseintvecmodelower>,
vec_cmpuv2div2di): Use nonimmediate_or_const_vector_operand
predicate instead of nonimmediate_operand and
vector_or_const_vector_operand instead of vector_operand.
* config/i386/i386-expand.cc (ix86_expand_int_sse_cmp): For
LE/LEU or GE/GEU with CONST_VECTOR cop1 try to transform those
into LE/LEU or GT/GTU with larger or smaller by one cop1 if
there is no wrap-around. Force CONST_VECTOR cop0 or cop1 into
REG. Formatting fix.
* gcc.target/i386/pr107546.c: New test.
As r13-3609-g6d9dbdf51f9afe8 has been committed, we can now enable even the denorm_min test. 2022-11-08 Jakub Jelinek <jakub@redhat.com> * testsuite/20_util/to_chars/float128_c++23.cc (test): Uncomment denorm_min test.
gcc/ PR rtl-optimization/107482 * ira-color.cc (assign_hard_reg): Only call update_costs_from_copies when retry_p is false.
When the allocator is of an unconstrained array type and has an initializing expression, the copy of the initializing expression must be done separately from that of the bounds. gcc/ada/ * gcc-interface/utils2.cc (build_allocator): For unconstrained array types with a storage model and an initializing expression, copy the initialization expression separately from the bounds. In all cases with a storage model, pass the locally computed size for the store.
In the case of Some_Enumeration_Type'Image (<some static value>), the compiler will replace this expression in its internal program representation with a corresponding string literal. This is incorrect if the Put_Image aspect has been specified (directly or via inheritance) for the enumeration type. gcc/ada/ * sem_attr.adb (Eval_Attribute): Don't simplify 'Image call if Put_Image has been specified.
Code cleanup; semantics is unaffected. gcc/ada/ * exp_ch6.adb, exp_put_image.adb, sem_aggr.adb, sem_attr.adb, sem_ch5.adb, sem_type.adb, sem_util.adb: Replace "Present (Find_Aspect (...))" with "Has_Aspect".
Before this patch, a classwide contract expression was preanalyzed only when its primitive operation's type was frozen. It caused name resolution to be off in the cases where the freezing took place after the end of the declaration list the primitive operation was declared in. This patch makes it so that if the compiler gets to the end of the declaration list before the type is frozen, it preanalyzes the classwide contract expression, so that the names are resolved in the right context. gcc/ada/ * contracts.adb (Preanalyze_Class_Conditions): New procedure. (Preanalyze_Condition): Moved out from Merge_Class_Conditions in order to be spec-visible. * contracts.ads (Preanalyze_Class_Conditions): New procedure. * sem_prag.adb (Analyze_Pre_Post_Condition_In_Decl_Part): Call Preanalyze_Class_Conditions when necessary.
gcc/ada/ * libgnat/system-vxworks7-aarch64-rtp-smp.ads: Set Support_Atomic_Primitives to True. * libgnat/system-vxworks7-aarch64.ads: Set Support_Atomic_Primitives to True. * libgnat/system-vxworks7-arm-rtp-smp.ads: Set Support_Atomic_Primitives to True. * libgnat/system-vxworks7-arm.ads: Set Support_Atomic_Primitives to True. * libgnat/system-vxworks7-ppc-kernel.ads: Set Support_Atomic_Primitives to False. * libgnat/system-vxworks7-ppc-rtp-smp.ads: Set Support_Atomic_Primitives to False. * libgnat/system-vxworks7-ppc64-kernel.ads: Set Support_Atomic_Primitives to True. * libgnat/system-vxworks7-ppc64-rtp-smp.ads: Set Support_Atomic_Primitives to True. * libgnat/system-vxworks7-x86-kernel.ads: Set Support_Atomic_Primitives to True. * libgnat/system-vxworks7-x86-rtp-smp.ads: Set Support_Atomic_Primitives to True. * libgnat/system-vxworks7-x86_64-kernel.ads: Set Support_Atomic_Primitives to True. * libgnat/system-vxworks7-x86_64-rtp-smp.ads: Set Support_Atomic_Primitives to True.
gcc/ada/ * fe.h (Get_Warn_On_Questionable_Layout): Add void parameter.