Hi All, As the discussion in the PR pointed out the RTL we have for the REDUC_PLUS patterns are wrong. The UNSPECs are modelled as returning a vector and then in an expand pattern we emit a vec_select of the 0th element to get the scalar. This is incorrect as the instruction itself already only returns a single scalar and by declaring it returns a vector it allows combine to push in a subreg into the pattern, which causes reload to make duplicate moves. This patch corrects this by removing the weird indirection and making the RTL pattern model the correct semantics of the instruction immediately. Bootstrapped Regtested on aarch64-none-linux-gnu and no issues. Ok for master? Thanks, Tamar gcc/ChangeLog: PR target/104049 * config/aarch64/aarch64-simd.md (aarch64_reduc_plus_internal): Fix RTL and rename to... (reduc_plus_scal_): ... This. (reduc_plus_scal_v4sf): Moved. (aarch64_reduc_plus_internalv2si): Fix RTL and rename to... (reduc_plus_scal_v2si): ... This. gcc/testsuite/ChangeLog: PR target/104049 * gcc.target/aarch64/vadd_reduc-1.c: New test. * gcc.target/aarch64/vadd_reduc-2.c: New test. --- inline copy of patch -- diff --git a/gcc/config/aarch64/aarch64-simd.md b/gcc/config/aarch64/aarch64-simd.md index 18733428f3fb91d937346aa360f6d1fe13ca1eae..ad69a1c4eafe30a98e3d2273fe498694eeddc05a 100644 --- a/gcc/config/aarch64/aarch64-simd.md +++ b/gcc/config/aarch64/aarch64-simd.md @@ -3385,20 +3385,6 @@ (define_insn "3" ;; 'across lanes' add. -(define_expand "reduc_plus_scal_" - [(match_operand: 0 "register_operand") - (unspec:VDQ_I [(match_operand:VDQ_I 1 "register_operand")] - UNSPEC_ADDV)] - "TARGET_SIMD" - { - rtx elt = aarch64_endian_lane_rtx (mode, 0); - rtx scratch = gen_reg_rtx (mode); - emit_insn (gen_aarch64_reduc_plus_internal (scratch, operands[1])); - emit_insn (gen_aarch64_get_lane (operands[0], scratch, elt)); - DONE; - } -) - (define_insn "aarch64_faddp" [(set (match_operand:VHSDF 0 "register_operand" "=w") (unspec:VHSDF [(match_operand:VHSDF 1 "register_operand" "w") @@ -3409,15 +3395,47 @@ (define_insn "aarch64_faddp" [(set_attr "type" "neon_fp_reduc_add_")] ) -(define_insn "aarch64_reduc_plus_internal" - [(set (match_operand:VDQV 0 "register_operand" "=w") - (unspec:VDQV [(match_operand:VDQV 1 "register_operand" "w")] +(define_insn "reduc_plus_scal_" + [(set (match_operand: 0 "register_operand" "=w") + (unspec: [(match_operand:VDQV 1 "register_operand" "w")] UNSPEC_ADDV))] "TARGET_SIMD" "add\\t%0, %1." [(set_attr "type" "neon_reduc_add")] ) +(define_insn "reduc_plus_scal_v2si" + [(set (match_operand:SI 0 "register_operand" "=w") + (unspec:SI [(match_operand:V2SI 1 "register_operand" "w")] + UNSPEC_ADDV))] + "TARGET_SIMD" + "addp\\t%0.2s, %1.2s, %1.2s" + [(set_attr "type" "neon_reduc_add")] +) + +(define_insn "reduc_plus_scal_" + [(set (match_operand: 0 "register_operand" "=w") + (unspec: [(match_operand:V2F 1 "register_operand" "w")] + UNSPEC_FADDV))] + "TARGET_SIMD" + "faddp\\t%0, %1." + [(set_attr "type" "neon_fp_reduc_add_")] +) + +(define_expand "reduc_plus_scal_v4sf" + [(set (match_operand:SF 0 "register_operand") + (unspec:SF [(match_operand:V4SF 1 "register_operand")] + UNSPEC_FADDV))] + "TARGET_SIMD" +{ + rtx elt = aarch64_endian_lane_rtx (V4SFmode, 0); + rtx scratch = gen_reg_rtx (V4SFmode); + emit_insn (gen_aarch64_faddpv4sf (scratch, operands[1], operands[1])); + emit_insn (gen_aarch64_faddpv4sf (scratch, scratch, scratch)); + emit_insn (gen_aarch64_get_lanev4sf (operands[0], scratch, elt)); + DONE; +}) + (define_insn "aarch64_addlv" [(set (match_operand: 0 "register_operand" "=w") (unspec: [(match_operand:VDQV_L 1 "register_operand" "w")] @@ -3447,38 +3465,6 @@ (define_insn "aarch64_zero_extend_reduc_plus_" [(set_attr "type" "neon_reduc_add")] ) -(define_insn "aarch64_reduc_plus_internalv2si" - [(set (match_operand:V2SI 0 "register_operand" "=w") - (unspec:V2SI [(match_operand:V2SI 1 "register_operand" "w")] - UNSPEC_ADDV))] - "TARGET_SIMD" - "addp\\t%0.2s, %1.2s, %1.2s" - [(set_attr "type" "neon_reduc_add")] -) - -(define_insn "reduc_plus_scal_" - [(set (match_operand: 0 "register_operand" "=w") - (unspec: [(match_operand:V2F 1 "register_operand" "w")] - UNSPEC_FADDV))] - "TARGET_SIMD" - "faddp\\t%0, %1." - [(set_attr "type" "neon_fp_reduc_add_")] -) - -(define_expand "reduc_plus_scal_v4sf" - [(set (match_operand:SF 0 "register_operand") - (unspec:V4SF [(match_operand:V4SF 1 "register_operand")] - UNSPEC_FADDV))] - "TARGET_SIMD" -{ - rtx elt = aarch64_endian_lane_rtx (V4SFmode, 0); - rtx scratch = gen_reg_rtx (V4SFmode); - emit_insn (gen_aarch64_faddpv4sf (scratch, operands[1], operands[1])); - emit_insn (gen_aarch64_faddpv4sf (scratch, scratch, scratch)); - emit_insn (gen_aarch64_get_lanev4sf (operands[0], scratch, elt)); - DONE; -}) - (define_insn "clrsb2" [(set (match_operand:VDQ_BHSI 0 "register_operand" "=w") (clrsb:VDQ_BHSI (match_operand:VDQ_BHSI 1 "register_operand" "w")))] diff --git a/gcc/testsuite/gcc.target/aarch64/vadd_reduc-1.c b/gcc/testsuite/gcc.target/aarch64/vadd_reduc-1.c new file mode 100644 index 0000000000000000000000000000000000000000..271a1c3e8c319f7bdb8669e4b3199b6f88ef7f5d --- /dev/null +++ b/gcc/testsuite/gcc.target/aarch64/vadd_reduc-1.c @@ -0,0 +1,22 @@ +/* { dg-do compile } */ +/* { dg-additional-options "-O3" } */ +/* { dg-final { check-function-bodies "**" "" "" } } */ + +#include + +typedef int v4si __attribute__ ((vector_size (16))); + +/* +**bar: +** ... +** addv s0, v0.4s +** fmov w0, s0 +** lsr w1, w0, 16 +** add w0, w1, w0, uxth +** ret +*/ +int bar (v4si x) +{ + unsigned int sum = vaddvq_s32 (x); + return (((uint16_t)(sum & 0xffff)) + ((uint32_t)sum >> 16)); +} diff --git a/gcc/testsuite/gcc.target/aarch64/vadd_reduc-2.c b/gcc/testsuite/gcc.target/aarch64/vadd_reduc-2.c new file mode 100644 index 0000000000000000000000000000000000000000..5355cbcca34f40a686da837acf75780eb4286974 --- /dev/null +++ b/gcc/testsuite/gcc.target/aarch64/vadd_reduc-2.c @@ -0,0 +1,32 @@ +/* { dg-do compile } */ +/* { dg-additional-options "-O3 -std=c99" } */ +/* { dg-final { check-function-bodies "**" "" "" } } */ + +#include + +/* +**test: +** ... +** addv s0, v0.4s +** fmov w0, s0 +** and w1, w0, 65535 +** add w0, w1, w0, lsr 16 +** lsr w0, w0, 1 +** ret +*/ +int test (uint8_t *p, uint32_t t[1][1], int n) { + + int sum = 0; + uint32_t a0; + for (int i = 0; i < 4; i++, p++) + t[i][0] = p[0]; + + for (int i = 0; i < 4; i++) { + { + int t0 = t[0][i] + t[0][i]; + a0 = t0; + }; + sum += a0; + } + return (((uint16_t)sum) + ((uint32_t)sum >> 16)) >> 1; +} --