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); vor t,c,d xxeval a,b,c,d,128 vnor a,t,b a = (~ (c | d)) ^ b; vnor t,c,d xxeval a,b,c,d,135 vxor a,t,b a = (~ (c | d)) | b; vnor t,c,d xxeval a,b,c,d,143 vor a,t,b a = ~ ((c ^ d) | b); vxor t,c,d xxeval a,b,c,d,144 vnor a,t,b a = (~ (c ^ d)) ^ b; veqv t,c,d xxeval a,b,c,d,150 vxor a,t,b a = (~ (c ^ d)) | b; veqv t,c,d xxeval a,b,c,d,159 vor a,t,b a = (c | ~ d) ^ b; vorc t,c,d xxeval a,b,c,d,180 vxor a,t,b a = (c | ~ d) | b; vorc t,c,d xxeval a,b,c,d,191 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 patches into the trunk? I have committed all of the patches in my backlog (dense math registers, other -mcpu=future instructions, random bug fixes, support for _Float16 and __bfloat16, and optimizations for vector logical operations on power10/power11) into the IBM vendor branch: vendors/ibm/gcc-17-future 2026-07-01 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. --- gcc/config/rs6000/fusion.md | 120 ++++++++++++++++++++------------- gcc/config/rs6000/genfusion.pl | 8 +++ 2 files changed, 80 insertions(+), 48 deletions(-) diff --git a/gcc/config/rs6000/fusion.md b/gcc/config/rs6000/fusion.md index 84cf352b477..1491e871508 100644 --- a/gcc/config/rs6000/fusion.md +++ b/gcc/config/rs6000/fusion.md @@ -2576,20 +2576,23 @@ (define_insn "*fuse_vnor_vnor" ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vor -> vnor (define_insn "*fuse_vor_vnor" - [(set (match_operand:VM 3 "altivec_register_operand" "=&0,&1,&v,v") - (and:VM (not: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"))) - (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 (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"))) + (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)" "@ vor %3,%1,%0\;vnor %3,%3,%2 vor %3,%1,%0\;vnor %3,%3,%2 vor %3,%1,%0\;vnor %3,%3,%2 + xxeval %x3,%x2,%x1,%x0,128 vor %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" "*,*,*,p10p,*")]) ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vorc -> vnor @@ -2615,20 +2618,23 @@ (define_insn "*fuse_vorc_vnor" ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vxor -> vnor (define_insn "*fuse_vxor_vnor" - [(set (match_operand:VM 3 "altivec_register_operand" "=&0,&1,&v,v") - (and: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 (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)" "@ vxor %3,%1,%0\;vnor %3,%3,%2 vxor %3,%1,%0\;vnor %3,%3,%2 vxor %3,%1,%0\;vnor %3,%3,%2 + xxeval %x3,%x2,%x1,%x0,144 vxor %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" "*,*,*,p10p,*")]) ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vand -> vor @@ -2675,20 +2681,23 @@ (define_insn "*fuse_vandc_vor" ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector veqv -> vor (define_insn "*fuse_veqv_vor" - [(set (match_operand:VM 3 "altivec_register_operand" "=&0,&1,&v,v") - (ior: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"))) - (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 (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"))) + (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\;vor %3,%3,%2 veqv %3,%1,%0\;vor %3,%3,%2 veqv %3,%1,%0\;vor %3,%3,%2 + xxeval %x3,%x2,%x1,%x0,159 veqv %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" "*,*,*,p10p,*")]) ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vnand -> vor @@ -2711,20 +2720,23 @@ (define_insn "*fuse_vnand_vor" ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vnor -> vor (define_insn "*fuse_vnor_vor" - [(set (match_operand:VM 3 "altivec_register_operand" "=&0,&1,&v,v") - (ior: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"))) - (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 (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"))) + (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\;vor %3,%3,%2 vnor %3,%1,%0\;vor %3,%3,%2 vnor %3,%1,%0\;vor %3,%3,%2 + xxeval %x3,%x2,%x1,%x0,143 vnor %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" "*,*,*,p10p,*")]) ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vor -> vor @@ -2750,20 +2762,23 @@ (define_insn "*fuse_vor_vor" ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vorc -> vor (define_insn "*fuse_vorc_vor" - [(set (match_operand:VM 3 "altivec_register_operand" "=&0,&1,&v,v") - (ior:VM (ior:VM (not: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 (not: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)" "@ vorc %3,%1,%0\;vor %3,%3,%2 vorc %3,%1,%0\;vor %3,%3,%2 vorc %3,%1,%0\;vor %3,%3,%2 + xxeval %x3,%x2,%x1,%x0,191 vorc %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" "*,*,*,p10p,*")]) ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vxor -> vor @@ -2978,20 +2993,23 @@ (define_insn "*fuse_vandc_vxor" ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector veqv -> vxor (define_insn "*fuse_veqv_vxor" - [(set (match_operand:VM 3 "altivec_register_operand" "=&0,&1,&v,v") - (xor: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"))) - (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 (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"))) + (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\;vxor %3,%3,%2 veqv %3,%1,%0\;vxor %3,%3,%2 veqv %3,%1,%0\;vxor %3,%3,%2 + xxeval %x3,%x2,%x1,%x0,150 veqv %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" "*,*,*,p10p,*")]) ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vnand -> vxor @@ -3014,20 +3032,23 @@ (define_insn "*fuse_vnand_vxor" ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vnor -> vxor (define_insn "*fuse_vnor_vxor" - [(set (match_operand:VM 3 "altivec_register_operand" "=&0,&1,&v,v") - (xor: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"))) - (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 (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"))) + (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\;vxor %3,%3,%2 vnor %3,%1,%0\;vxor %3,%3,%2 vnor %3,%1,%0\;vxor %3,%3,%2 + xxeval %x3,%x2,%x1,%x0,135 vnor %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" "*,*,*,p10p,*")]) ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vor -> vxor @@ -3053,20 +3074,23 @@ (define_insn "*fuse_vor_vxor" ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vorc -> vxor (define_insn "*fuse_vorc_vxor" - [(set (match_operand:VM 3 "altivec_register_operand" "=&0,&1,&v,v") - (xor:VM (ior:VM (not: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 (not: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)" "@ vorc %3,%1,%0\;vxor %3,%3,%2 vorc %3,%1,%0\;vxor %3,%3,%2 vorc %3,%1,%0\;vxor %3,%3,%2 + xxeval %x3,%x2,%x1,%x0,180 vorc %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" "*,*,*,p10p,*")]) ;; logical-logical fusion pattern generated by gen_logical_addsubf ;; vector vxor -> vxor diff --git a/gcc/config/rs6000/genfusion.pl b/gcc/config/rs6000/genfusion.pl index 5fec485cdaa..1180551042b 100755 --- a/gcc/config/rs6000/genfusion.pl +++ b/gcc/config/rs6000/genfusion.pl @@ -238,6 +238,14 @@ sub gen_logical_addsubf "vnor_vnor" => 112, "vor_vxor" => 120, "vor_vor" => 127, + "vor_vnor" => 128, + "vnor_vxor" => 135, + "vnor_vor" => 143, + "vxor_vnor" => 144, + "veqv_vxor" => 150, + "veqv_vor" => 159, + "vorc_vxor" => 180, + "vorc_vor" => 191, ); KIND: foreach $kind ('scalar','vector') { -- 2.54.0 -- Michael Meissner, IBM PO Box 98, Ayer, Massachusetts, USA, 01432 email: [email protected]
