Bootstrapped and regtested on x86_64-pc-linux-gnu{-m32,}. Ready to push to trunk.
gcc/ChangeLog: * config/i386/sse.md (*<avx512>_eq<mode>3<mask_scalar_merge_name>_1): Extend to UNSPEC_PCMP_UNSIGNED. gcc/testsuite/ChangeLog: * gcc.target/i386/pr103774.c: New test. * gcc.target/i386/avx512bw-vpcmpequb-1.c: Adjust scan assembler from vpcmpub to (?:vpcmpub|vpcmpeqb). * gcc.target/i386/avx512bw-vpcmpequw-1.c: Ditto. * gcc.target/i386/avx512bw-vpcmpub-1.c: Ditto. * gcc.target/i386/avx512bw-vpcmpuw-1.c: Ditto. * gcc.target/i386/avx512f-vpcmpequd-1.c: Ditto. * gcc.target/i386/avx512f-vpcmpequq-1.c: Ditto. * gcc.target/i386/avx512f-vpcmpud-1.c: Ditto. * gcc.target/i386/avx512vl-vpcmpequd-1.c: Ditto. * gcc.target/i386/avx512vl-vpcmpequq-1.c: Ditto. * gcc.target/i386/avx512vl-vpcmpuq-1.c: Ditto. --- gcc/config/i386/sse.md | 64 +++++++++---------- .../gcc.target/i386/avx512bw-vpcmpequb-1.c | 12 ++-- .../gcc.target/i386/avx512bw-vpcmpequw-1.c | 12 ++-- .../gcc.target/i386/avx512bw-vpcmpub-1.c | 2 +- .../gcc.target/i386/avx512bw-vpcmpuw-1.c | 2 +- .../gcc.target/i386/avx512f-vpcmpequd-1.c | 4 +- .../gcc.target/i386/avx512f-vpcmpequq-1.c | 4 +- .../gcc.target/i386/avx512f-vpcmpud-1.c | 2 +- .../gcc.target/i386/avx512vl-vpcmpequd-1.c | 8 +-- .../gcc.target/i386/avx512vl-vpcmpequq-1.c | 8 +-- .../gcc.target/i386/avx512vl-vpcmpuq-1.c | 2 +- gcc/testsuite/gcc.target/i386/pr103774.c | 25 ++++++++ 12 files changed, 85 insertions(+), 60 deletions(-) create mode 100644 gcc/testsuite/gcc.target/i386/pr103774.c diff --git a/gcc/config/i386/sse.md b/gcc/config/i386/sse.md index fa1d56ae3e3..35360004818 100644 --- a/gcc/config/i386/sse.md +++ b/gcc/config/i386/sse.md @@ -3895,6 +3895,22 @@ (define_insn_and_split "*<avx512>_cmp<mode>3" UNSPEC_PCMP_ITER))] "operands[4] = GEN_INT (INTVAL (operands[3]) ^ 4);") +(define_insn "*<avx512>_eq<mode>3<mask_scalar_merge_name>_1" + [(set (match_operand:<avx512fmaskmode> 0 "register_operand" "=k,k") + (unspec:<avx512fmaskmode> + [(match_operand:VI12_AVX512VL 1 "nonimm_or_0_operand" "%v,v") + (match_operand:VI12_AVX512VL 2 "nonimm_or_0_operand" "vm,C") + (const_int 0)] + UNSPEC_PCMP_ITER))] + "TARGET_AVX512BW && !(MEM_P (operands[1]) && MEM_P (operands[2]))" + "@ + vpcmpeq<ssemodesuffix>\t{%2, %1, %0<mask_scalar_merge_operand3>|%0<mask_scalar_merge_operand3>, %1, %2} + vptestnm<ssemodesuffix>\t{%1, %1, %0<mask_scalar_merge_operand3>|%0<mask_scalar_merge_operand3>, %1, %1}" + [(set_attr "type" "ssecmp") + (set_attr "prefix_extra" "1") + (set_attr "prefix" "evex") + (set_attr "mode" "<sseinsnmode>")]) + (define_insn "<avx512>_ucmp<mode>3<mask_scalar_merge_name>" [(set (match_operand:<avx512fmaskmode> 0 "register_operand" "=k") (unspec:<avx512fmaskmode> @@ -3977,6 +3993,22 @@ (define_insn_and_split "*<avx512>_ucmp<VI12_AVX512VL:mode>3_zero_extend<SWI248x: (set_attr "prefix" "evex") (set_attr "mode" "<VI12_AVX512VL:sseinsnmode>")]) +(define_insn "*<avx512>_eq<mode>3<mask_scalar_merge_name>_1" + [(set (match_operand:<avx512fmaskmode> 0 "register_operand" "=k,k") + (unspec:<avx512fmaskmode> + [(match_operand:VI48_AVX512VL 1 "nonimm_or_0_operand" "%v,v") + (match_operand:VI48_AVX512VL 2 "nonimm_or_0_operand" "vm,C") + (const_int 0)] + UNSPEC_PCMP_ITER))] + "TARGET_AVX512F && !(MEM_P (operands[1]) && MEM_P (operands[2]))" + "@ + vpcmpeq<ssemodesuffix>\t{%2, %1, %0<mask_scalar_merge_operand3>|%0<mask_scalar_merge_operand3>, %1, %2} + vptestnm<ssemodesuffix>\t{%1, %1, %0<mask_scalar_merge_operand3>|%0<mask_scalar_merge_operand3>, %1, %1}" + [(set_attr "type" "ssecmp") + (set_attr "prefix_extra" "1") + (set_attr "prefix" "evex") + (set_attr "mode" "<sseinsnmode>")]) + (define_insn "<avx512>_ucmp<mode>3<mask_scalar_merge_name>" [(set (match_operand:<avx512fmaskmode> 0 "register_operand" "=k") (unspec:<avx512fmaskmode> @@ -16248,38 +16280,6 @@ (define_expand "<avx512>_eq<mode>3<mask_scalar_merge_name>" "TARGET_AVX512F" "ix86_fixup_binary_operands_no_copy (EQ, <MODE>mode, operands);") -(define_insn "*<avx512>_eq<mode>3<mask_scalar_merge_name>_1" - [(set (match_operand:<avx512fmaskmode> 0 "register_operand" "=k,k") - (unspec:<avx512fmaskmode> - [(match_operand:VI12_AVX512VL 1 "nonimm_or_0_operand" "%v,v") - (match_operand:VI12_AVX512VL 2 "nonimm_or_0_operand" "vm,C") - (const_int 0)] - UNSPEC_PCMP))] - "TARGET_AVX512BW && !(MEM_P (operands[1]) && MEM_P (operands[2]))" - "@ - vpcmpeq<ssemodesuffix>\t{%2, %1, %0<mask_scalar_merge_operand3>|%0<mask_scalar_merge_operand3>, %1, %2} - vptestnm<ssemodesuffix>\t{%1, %1, %0<mask_scalar_merge_operand3>|%0<mask_scalar_merge_operand3>, %1, %1}" - [(set_attr "type" "ssecmp") - (set_attr "prefix_extra" "1") - (set_attr "prefix" "evex") - (set_attr "mode" "<sseinsnmode>")]) - -(define_insn "*<avx512>_eq<mode>3<mask_scalar_merge_name>_1" - [(set (match_operand:<avx512fmaskmode> 0 "register_operand" "=k,k") - (unspec:<avx512fmaskmode> - [(match_operand:VI48_AVX512VL 1 "nonimm_or_0_operand" "%v,v") - (match_operand:VI48_AVX512VL 2 "nonimm_or_0_operand" "vm,C") - (const_int 0)] - UNSPEC_PCMP))] - "TARGET_AVX512F && !(MEM_P (operands[1]) && MEM_P (operands[2]))" - "@ - vpcmpeq<ssemodesuffix>\t{%2, %1, %0<mask_scalar_merge_operand3>|%0<mask_scalar_merge_operand3>, %1, %2} - vptestnm<ssemodesuffix>\t{%1, %1, %0<mask_scalar_merge_operand3>|%0<mask_scalar_merge_operand3>, %1, %1}" - [(set_attr "type" "ssecmp") - (set_attr "prefix_extra" "1") - (set_attr "prefix" "evex") - (set_attr "mode" "<sseinsnmode>")]) - (define_insn "*sse4_1_eqv2di3" [(set (match_operand:V2DI 0 "register_operand" "=Yr,*x,x") (eq:V2DI diff --git a/gcc/testsuite/gcc.target/i386/avx512bw-vpcmpequb-1.c b/gcc/testsuite/gcc.target/i386/avx512bw-vpcmpequb-1.c index 31a0afc2f17..9a85753c9fb 100644 --- a/gcc/testsuite/gcc.target/i386/avx512bw-vpcmpequb-1.c +++ b/gcc/testsuite/gcc.target/i386/avx512bw-vpcmpequb-1.c @@ -1,11 +1,11 @@ /* { dg-do compile } */ /* { dg-options "-mavx512bw -mavx512vl -O2" } */ -/* { dg-final { scan-assembler-times "vpcmpub\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ -/* { dg-final { scan-assembler-times "vpcmpub\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ -/* { dg-final { scan-assembler-times "vpcmpub\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ -/* { dg-final { scan-assembler-times "vpcmpub\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ -/* { dg-final { scan-assembler-times "vpcmpub\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ -/* { dg-final { scan-assembler-times "vpcmpub\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpub|vpcmpeqb)\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpub|vpcmpeqb)\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpub|vpcmpeqb)\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpub|vpcmpeqb)\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpub|vpcmpeqb)\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpub|vpcmpeqb)\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ #include <immintrin.h> diff --git a/gcc/testsuite/gcc.target/i386/avx512bw-vpcmpequw-1.c b/gcc/testsuite/gcc.target/i386/avx512bw-vpcmpequw-1.c index da5c4015797..a6c7776b52f 100644 --- a/gcc/testsuite/gcc.target/i386/avx512bw-vpcmpequw-1.c +++ b/gcc/testsuite/gcc.target/i386/avx512bw-vpcmpequw-1.c @@ -1,11 +1,11 @@ /* { dg-do compile } */ /* { dg-options "-mavx512bw -mavx512vl -O2" } */ -/* { dg-final { scan-assembler-times "vpcmpuw\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ -/* { dg-final { scan-assembler-times "vpcmpuw\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ -/* { dg-final { scan-assembler-times "vpcmpuw\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ -/* { dg-final { scan-assembler-times "vpcmpuw\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ -/* { dg-final { scan-assembler-times "vpcmpuw\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ -/* { dg-final { scan-assembler-times "vpcmpuw\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpuw|vpcmpeqw)\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpuw|vpcmpeqw)\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpuw|vpcmpeqw)\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpuw|vpcmpeqw)\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpuw|vpcmpeqw)\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpuw|vpcmpeqw)\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ #include <immintrin.h> diff --git a/gcc/testsuite/gcc.target/i386/avx512bw-vpcmpub-1.c b/gcc/testsuite/gcc.target/i386/avx512bw-vpcmpub-1.c index 120fe8fd19c..6d8a442d79f 100644 --- a/gcc/testsuite/gcc.target/i386/avx512bw-vpcmpub-1.c +++ b/gcc/testsuite/gcc.target/i386/avx512bw-vpcmpub-1.c @@ -1,6 +1,6 @@ /* { dg-do compile } */ /* { dg-options "-O2 -mavx512bw -mavx512vl" } */ -/* { dg-final { scan-assembler-times "vpcmpub\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n^k\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpub|vpcmpeqb)\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n^k\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ /* { dg-final { scan-assembler-times "vpcmpub\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n^k\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ /* { dg-final { scan-assembler-times "vpcmpub\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n^k\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ /* { dg-final { scan-assembler-times "vpcmpub\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n^k\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ diff --git a/gcc/testsuite/gcc.target/i386/avx512bw-vpcmpuw-1.c b/gcc/testsuite/gcc.target/i386/avx512bw-vpcmpuw-1.c index f17c73b05c1..c0d84ded11b 100644 --- a/gcc/testsuite/gcc.target/i386/avx512bw-vpcmpuw-1.c +++ b/gcc/testsuite/gcc.target/i386/avx512bw-vpcmpuw-1.c @@ -5,7 +5,7 @@ /* { dg-final { scan-assembler-times "vpcmpuw\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n^k\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ /* { dg-final { scan-assembler-times "vpcmpuw\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n^k\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ /* { dg-final { scan-assembler-times "vpcmpuw\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n^k\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ -/* { dg-final { scan-assembler-times "vpcmpuw\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n^k\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpuw|vpcmpeqw)\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n^k\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ #include <immintrin.h> diff --git a/gcc/testsuite/gcc.target/i386/avx512f-vpcmpequd-1.c b/gcc/testsuite/gcc.target/i386/avx512f-vpcmpequd-1.c index c02f2b1acef..3c567c648b6 100644 --- a/gcc/testsuite/gcc.target/i386/avx512f-vpcmpequd-1.c +++ b/gcc/testsuite/gcc.target/i386/avx512f-vpcmpequd-1.c @@ -1,7 +1,7 @@ /* { dg-do compile } */ /* { dg-options "-mavx512f -O2" } */ -/* { dg-final { scan-assembler-times "vpcmpud\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n^k\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ -/* { dg-final { scan-assembler-times "vpcmpud\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n^k\]*%k\[0-7\]\{%k\[0-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpud|vpcmpeqd)\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n^k\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpud|vpcmpeqd)\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n^k\]*%k\[0-7\]\{%k\[0-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ #include <immintrin.h> diff --git a/gcc/testsuite/gcc.target/i386/avx512f-vpcmpequq-1.c b/gcc/testsuite/gcc.target/i386/avx512f-vpcmpequq-1.c index 1a443d64cd6..f3f35ec62ec 100644 --- a/gcc/testsuite/gcc.target/i386/avx512f-vpcmpequq-1.c +++ b/gcc/testsuite/gcc.target/i386/avx512f-vpcmpequq-1.c @@ -1,7 +1,7 @@ /* { dg-do compile } */ /* { dg-options "-mavx512f -O2" } */ -/* { dg-final { scan-assembler-times "vpcmpuq\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n^k\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ -/* { dg-final { scan-assembler-times "vpcmpuq\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n^k\]*%k\[0-7\]\{%k\[0-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpuq|vpcmpeqq)\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n^k\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpuq|vpcmpeqq)\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n^k\]*%k\[0-7\]\{%k\[0-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ #include <immintrin.h> diff --git a/gcc/testsuite/gcc.target/i386/avx512f-vpcmpud-1.c b/gcc/testsuite/gcc.target/i386/avx512f-vpcmpud-1.c index 331f95c231d..be0a8feb025 100644 --- a/gcc/testsuite/gcc.target/i386/avx512f-vpcmpud-1.c +++ b/gcc/testsuite/gcc.target/i386/avx512f-vpcmpud-1.c @@ -1,6 +1,6 @@ /* { dg-do compile } */ /* { dg-options "-O2 -mavx512f" } */ -/* { dg-final { scan-assembler-times "vpcmpud\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n^k\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpud|vpcmpeqd)\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n^k\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ /* { dg-final { scan-assembler-times "vpcmpud\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n^k\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ #include <immintrin.h> diff --git a/gcc/testsuite/gcc.target/i386/avx512vl-vpcmpequd-1.c b/gcc/testsuite/gcc.target/i386/avx512vl-vpcmpequd-1.c index 7c87035afab..a0bee114c15 100644 --- a/gcc/testsuite/gcc.target/i386/avx512vl-vpcmpequd-1.c +++ b/gcc/testsuite/gcc.target/i386/avx512vl-vpcmpequd-1.c @@ -1,9 +1,9 @@ /* { dg-do compile } */ /* { dg-options "-mavx512vl -O2" } */ -/* { dg-final { scan-assembler-times "vpcmpud\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ -/* { dg-final { scan-assembler-times "vpcmpud\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ -/* { dg-final { scan-assembler-times "vpcmpud\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n^k\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ -/* { dg-final { scan-assembler-times "vpcmpud\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n^k\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpud|vpcmpeqd)\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpud|vpcmpeqd)\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpud|vpcmpeqd)\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n^k\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpud|vpcmpeqd)\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n^k\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ #include <immintrin.h> diff --git a/gcc/testsuite/gcc.target/i386/avx512vl-vpcmpequq-1.c b/gcc/testsuite/gcc.target/i386/avx512vl-vpcmpequq-1.c index 4d672a54698..c925d3226ec 100644 --- a/gcc/testsuite/gcc.target/i386/avx512vl-vpcmpequq-1.c +++ b/gcc/testsuite/gcc.target/i386/avx512vl-vpcmpequq-1.c @@ -1,9 +1,9 @@ /* { dg-do compile } */ /* { dg-options "-mavx512vl -O2" } */ -/* { dg-final { scan-assembler-times "vpcmpuq\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ -/* { dg-final { scan-assembler-times "vpcmpuq\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ -/* { dg-final { scan-assembler-times "vpcmpuq\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ -/* { dg-final { scan-assembler-times "vpcmpuq\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpuq|vpcmpeqq)\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpuq|vpcmpeqq)\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpuq|vpcmpeqq)\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpuq|vpcmpeqq)\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ #include <immintrin.h> diff --git a/gcc/testsuite/gcc.target/i386/avx512vl-vpcmpuq-1.c b/gcc/testsuite/gcc.target/i386/avx512vl-vpcmpuq-1.c index a5491c92b18..f93fca2b655 100644 --- a/gcc/testsuite/gcc.target/i386/avx512vl-vpcmpuq-1.c +++ b/gcc/testsuite/gcc.target/i386/avx512vl-vpcmpuq-1.c @@ -2,7 +2,7 @@ /* { dg-options "-O2 -mavx512vl" } */ /* { dg-final { scan-assembler-times "vpcmpuq\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n^k\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ /* { dg-final { scan-assembler-times "vpcmpuq\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\n^k\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ -/* { dg-final { scan-assembler-times "vpcmpuq\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n^k\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ +/* { dg-final { scan-assembler-times "(?:vpcmpuq|vpcmpeqq)\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n^k\]*%k\[0-7\](?:\n|\[ \\t\]+#)" 1 } } */ /* { dg-final { scan-assembler-times "vpcmpuq\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\n^k\]*%k\[0-7\]\{%k\[1-7\]\}(?:\n|\[ \\t\]+#)" 1 } } */ #include <immintrin.h> diff --git a/gcc/testsuite/gcc.target/i386/pr103774.c b/gcc/testsuite/gcc.target/i386/pr103774.c new file mode 100644 index 00000000000..0dafb1b0117 --- /dev/null +++ b/gcc/testsuite/gcc.target/i386/pr103774.c @@ -0,0 +1,25 @@ +/* { dg-do compile } */ +/* { dg-options "-march=cannonlake -O2" } */ +/* { dg-final { scan-assembler-not {(?n)vmovdq} } } */ +/* There should be no load + vpcmp, just vpcmp with memory operand. */ + +#include<immintrin.h> +__mmask16 +foo (__m256i* n, __m256i mch256) +{ + __m256i data1 = _mm256_loadu_si256(n); + __m256i data2 = _mm256_loadu_si256(n+1); + __mmask16 mask1 = _mm256_cmpeq_epu16_mask(data1, mch256); + __mmask16 mask2 = _mm256_cmpeq_epu16_mask(data2, mch256); + return mask1 + mask2; +} + +__mmask16 +foo1 (__m256i* n, __m256i mch256) +{ + __m256i data1 = _mm256_loadu_si256(n); + __m256i data2 = _mm256_loadu_si256(n+1); + __mmask16 mask1 = _mm256_cmpeq_epu16_mask(mch256, data1); + __mmask16 mask2 = _mm256_cmpeq_epu16_mask(mch256, data2); + return mask1 + mask2; +} -- 2.18.1