https://gcc.gnu.org/g:5919488e18b558494133fdf9d918285a2fefa85f
commit 5919488e18b558494133fdf9d918285a2fefa85f Author: Michael Meissner <[email protected]> Date: Tue May 12 20:52:16 2026 -0400 PR target/117251: Improve vector fusion #4 See the following post for a complete explanation of what the patches for PR target/117251: * https://gcc.gnu.org/pipermail/gcc-patches/2025-June/686474.html This is patch #2 of 45 to generate the 'XXEVAL' instruction on power10 and power11 instead of using the Altivec 'VANDC' instruction feeding into 'VAND'. The 'XXEVAL' instruction can use all 64 vector registers, instead of the 32 registers that traditional Altivec vector instructions use. By allowing all of the vector registers to be used, it reduces the amount of spilling that a large benchmark generated. For vectors such as: vector int a, b, c, d; This patch will generate a call to xxeval if any of the registers are allocated to traditional floating point registers: Code: Old: New: ===== ==== ==== a = ~ ((~ (c ^ d)) | b); veqv t,c,d xxeval a,b,c,d,96 vnor a,t,b a = (c ^ d) ^ b; vxor t,c,d xxeval a,b,c,d,105 vxor a,t,b a = (c ^ d) | b; vxor t,c,d xxeval a,b,c,d,111 vor a,t,b a = ~ ((~ (c | d)) | b); vnor t,c,d xxeval a,b,c,d,112 vnor a,t,b a = (c | d) ^ b; vor t,c,d xxeval a,b,c,d,120 vxor a,t,b a = (c | d) | b; vor t,c,d xxeval a,b,c,d,127 vor a,t,b Since fusion using 2 Altivec instructions is slightly faster than using the 'XXEVAL' instruction we prefer to generate the Altivec instructions if we can. In addition, because 'XXEVAL' is a prefixed instruction, it possibly might generate an extra NOP instruction to align the 'XXEVAL' instruction. 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? 2026-05-12 Michael Meissner <[email protected]> gcc/ PR target/117251 * config/rs6000/fusion.md: Regenerate. * config/rs6000/genfusion.pl (gen_logical_addsubf): Add support to generate vector/vector fusion if XXEVAL is supported. Diff: --- gcc/config/rs6000/fusion.md | 90 +++++++++++++++++++++++++----------------- gcc/config/rs6000/genfusion.pl | 6 +++ 2 files changed, 60 insertions(+), 36 deletions(-) diff --git a/gcc/config/rs6000/fusion.md b/gcc/config/rs6000/fusion.md index 2a2a1d5de92c..78e846207313 100644 --- a/gcc/config/rs6000/fusion.md +++ b/gcc/config/rs6000/fusion.md @@ -2513,20 +2513,23 @@ ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector veqv -> vnor (define_insn "*fuse_veqv_vnor" - [(set (match_operand:VM 3 "altivec_register_operand" "=&0,&1,&v,v") - (and:VM (not:VM (not:VM (xor:VM (match_operand:VM 0 "altivec_register_operand" "v,v,v,v") - (match_operand:VM 1 "altivec_register_operand" "v,v,v,v")))) - (not:VM (match_operand:VM 2 "altivec_register_operand" "v,v,v,v")))) - (clobber (match_scratch:VM 4 "=X,X,X,&v"))] + [(set (match_operand:VM 3 "vector_fusion_operand" "=&0,&1,&v,wa,v") + (and:VM (not:VM (not:VM (xor:VM (match_operand:VM 0 "vector_fusion_operand" "v,v,v,wa,v") + (match_operand:VM 1 "vector_fusion_operand" "v,v,v,wa,v")))) + (not:VM (match_operand:VM 2 "vector_fusion_operand" "v,v,v,wa,v")))) + (clobber (match_scratch:VM 4 "=X,X,X,X,&v"))] "(TARGET_P10_FUSION)" "@ veqv %3,%1,%0\;vnor %3,%3,%2 veqv %3,%1,%0\;vnor %3,%3,%2 veqv %3,%1,%0\;vnor %3,%3,%2 + xxeval %x3,%x2,%x1,%x0,96 veqv %4,%1,%0\;vnor %3,%4,%2" [(set_attr "type" "fused_vector") (set_attr "cost" "6") - (set_attr "length" "8")]) + (set_attr "length" "8") + (set_attr "prefixed" "*,*,*,yes,*") + (set_attr "isa" "*,*,*,xxeval,*")]) ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vnand -> vnor @@ -2552,20 +2555,23 @@ ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vnor -> vnor (define_insn "*fuse_vnor_vnor" - [(set (match_operand:VM 3 "altivec_register_operand" "=&0,&1,&v,v") - (and:VM (not:VM (and:VM (not:VM (match_operand:VM 0 "altivec_register_operand" "v,v,v,v")) - (not:VM (match_operand:VM 1 "altivec_register_operand" "v,v,v,v")))) - (not:VM (match_operand:VM 2 "altivec_register_operand" "v,v,v,v")))) - (clobber (match_scratch:VM 4 "=X,X,X,&v"))] + [(set (match_operand:VM 3 "vector_fusion_operand" "=&0,&1,&v,wa,v") + (and:VM (not:VM (and:VM (not:VM (match_operand:VM 0 "vector_fusion_operand" "v,v,v,wa,v")) + (not:VM (match_operand:VM 1 "vector_fusion_operand" "v,v,v,wa,v")))) + (not:VM (match_operand:VM 2 "vector_fusion_operand" "v,v,v,wa,v")))) + (clobber (match_scratch:VM 4 "=X,X,X,X,&v"))] "(TARGET_P10_FUSION)" "@ vnor %3,%1,%0\;vnor %3,%3,%2 vnor %3,%1,%0\;vnor %3,%3,%2 vnor %3,%1,%0\;vnor %3,%3,%2 + xxeval %x3,%x2,%x1,%x0,112 vnor %4,%1,%0\;vnor %3,%4,%2" [(set_attr "type" "fused_vector") (set_attr "cost" "6") - (set_attr "length" "8")]) + (set_attr "length" "8") + (set_attr "prefixed" "*,*,*,yes,*") + (set_attr "isa" "*,*,*,xxeval,*")]) ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vor -> vnor @@ -2723,20 +2729,23 @@ ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vor -> vor (define_insn "*fuse_vor_vor" - [(set (match_operand:VM 3 "altivec_register_operand" "=&0,&1,&v,v") - (ior:VM (ior:VM (match_operand:VM 0 "altivec_register_operand" "v,v,v,v") - (match_operand:VM 1 "altivec_register_operand" "%v,v,v,v")) - (match_operand:VM 2 "altivec_register_operand" "v,v,v,v"))) - (clobber (match_scratch:VM 4 "=X,X,X,&v"))] + [(set (match_operand:VM 3 "vector_fusion_operand" "=&0,&1,&v,wa,v") + (ior:VM (ior:VM (match_operand:VM 0 "vector_fusion_operand" "v,v,v,wa,v") + (match_operand:VM 1 "vector_fusion_operand" "%v,v,v,wa,v")) + (match_operand:VM 2 "vector_fusion_operand" "v,v,v,wa,v"))) + (clobber (match_scratch:VM 4 "=X,X,X,X,&v"))] "(TARGET_P10_FUSION)" "@ vor %3,%1,%0\;vor %3,%3,%2 vor %3,%1,%0\;vor %3,%3,%2 vor %3,%1,%0\;vor %3,%3,%2 + xxeval %x3,%x2,%x1,%x0,127 vor %4,%1,%0\;vor %3,%4,%2" [(set_attr "type" "fused_vector") (set_attr "cost" "6") - (set_attr "length" "8")]) + (set_attr "length" "8") + (set_attr "prefixed" "*,*,*,yes,*") + (set_attr "isa" "*,*,*,xxeval,*")]) ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vorc -> vor @@ -2759,20 +2768,23 @@ ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vxor -> vor (define_insn "*fuse_vxor_vor" - [(set (match_operand:VM 3 "altivec_register_operand" "=&0,&1,&v,v") - (ior:VM (xor:VM (match_operand:VM 0 "altivec_register_operand" "v,v,v,v") - (match_operand:VM 1 "altivec_register_operand" "v,v,v,v")) - (match_operand:VM 2 "altivec_register_operand" "v,v,v,v"))) - (clobber (match_scratch:VM 4 "=X,X,X,&v"))] + [(set (match_operand:VM 3 "vector_fusion_operand" "=&0,&1,&v,wa,v") + (ior:VM (xor:VM (match_operand:VM 0 "vector_fusion_operand" "v,v,v,wa,v") + (match_operand:VM 1 "vector_fusion_operand" "v,v,v,wa,v")) + (match_operand:VM 2 "vector_fusion_operand" "v,v,v,wa,v"))) + (clobber (match_scratch:VM 4 "=X,X,X,X,&v"))] "(TARGET_P10_FUSION)" "@ vxor %3,%1,%0\;vor %3,%3,%2 vxor %3,%1,%0\;vor %3,%3,%2 vxor %3,%1,%0\;vor %3,%3,%2 + xxeval %x3,%x2,%x1,%x0,111 vxor %4,%1,%0\;vor %3,%4,%2" [(set_attr "type" "fused_vector") (set_attr "cost" "6") - (set_attr "length" "8")]) + (set_attr "length" "8") + (set_attr "prefixed" "*,*,*,yes,*") + (set_attr "isa" "*,*,*,xxeval,*")]) ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vand -> vorc @@ -3020,20 +3032,23 @@ ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vor -> vxor (define_insn "*fuse_vor_vxor" - [(set (match_operand:VM 3 "altivec_register_operand" "=&0,&1,&v,v") - (xor:VM (ior:VM (match_operand:VM 0 "altivec_register_operand" "v,v,v,v") - (match_operand:VM 1 "altivec_register_operand" "v,v,v,v")) - (match_operand:VM 2 "altivec_register_operand" "v,v,v,v"))) - (clobber (match_scratch:VM 4 "=X,X,X,&v"))] + [(set (match_operand:VM 3 "vector_fusion_operand" "=&0,&1,&v,wa,v") + (xor:VM (ior:VM (match_operand:VM 0 "vector_fusion_operand" "v,v,v,wa,v") + (match_operand:VM 1 "vector_fusion_operand" "v,v,v,wa,v")) + (match_operand:VM 2 "vector_fusion_operand" "v,v,v,wa,v"))) + (clobber (match_scratch:VM 4 "=X,X,X,X,&v"))] "(TARGET_P10_FUSION)" "@ vor %3,%1,%0\;vxor %3,%3,%2 vor %3,%1,%0\;vxor %3,%3,%2 vor %3,%1,%0\;vxor %3,%3,%2 + xxeval %x3,%x2,%x1,%x0,120 vor %4,%1,%0\;vxor %3,%4,%2" [(set_attr "type" "fused_vector") (set_attr "cost" "6") - (set_attr "length" "8")]) + (set_attr "length" "8") + (set_attr "prefixed" "*,*,*,yes,*") + (set_attr "isa" "*,*,*,xxeval,*")]) ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vorc -> vxor @@ -3056,20 +3071,23 @@ ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vxor -> vxor (define_insn "*fuse_vxor_vxor" - [(set (match_operand:VM 3 "altivec_register_operand" "=&0,&1,&v,v") - (xor:VM (xor:VM (match_operand:VM 0 "altivec_register_operand" "v,v,v,v") - (match_operand:VM 1 "altivec_register_operand" "%v,v,v,v")) - (match_operand:VM 2 "altivec_register_operand" "v,v,v,v"))) - (clobber (match_scratch:VM 4 "=X,X,X,&v"))] + [(set (match_operand:VM 3 "vector_fusion_operand" "=&0,&1,&v,wa,v") + (xor:VM (xor:VM (match_operand:VM 0 "vector_fusion_operand" "v,v,v,wa,v") + (match_operand:VM 1 "vector_fusion_operand" "%v,v,v,wa,v")) + (match_operand:VM 2 "vector_fusion_operand" "v,v,v,wa,v"))) + (clobber (match_scratch:VM 4 "=X,X,X,X,&v"))] "(TARGET_P10_FUSION)" "@ vxor %3,%1,%0\;vxor %3,%3,%2 vxor %3,%1,%0\;vxor %3,%3,%2 vxor %3,%1,%0\;vxor %3,%3,%2 + xxeval %x3,%x2,%x1,%x0,105 vxor %4,%1,%0\;vxor %3,%4,%2" [(set_attr "type" "fused_vector") (set_attr "cost" "6") - (set_attr "length" "8")]) + (set_attr "length" "8") + (set_attr "prefixed" "*,*,*,yes,*") + (set_attr "isa" "*,*,*,xxeval,*")]) ;; add-add fusion pattern generated by gen_addadd (define_insn "*fuse_add_add" diff --git a/gcc/config/rs6000/genfusion.pl b/gcc/config/rs6000/genfusion.pl index 15ce9d850b55..84503594edfe 100755 --- a/gcc/config/rs6000/genfusion.pl +++ b/gcc/config/rs6000/genfusion.pl @@ -232,6 +232,12 @@ sub gen_logical_addsubf "vorc_vnor" => 64, "vorc_veqv" => 75, "vorc_vorc" => 79, + "veqv_vnor" => 96, + "vxor_vxor" => 105, + "vxor_vor" => 111, + "vnor_vnor" => 112, + "vor_vxor" => 120, + "vor_vor" => 127, ); KIND: foreach $kind ('scalar','vector') {
