llvmbot wrote:
<!--LLVM PR SUMMARY COMMENT--> @llvm/pr-subscribers-backend-amdgpu Author: Christudasan Devadasan (cdevadas) <details> <summary>Changes</summary> For targets that support xnack replay feature (gfx8+), the multi-dword scalar loads shouldn't clobber any register that holds the src address. The constraint version of the scalar loads have the early clobber flag attached to the dst operand to restrict RA from re-allocating any of the src regs for its dst operand. --- Patch is 7.42 MiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/96163.diff 265 Files Affected: - (modified) llvm/lib/Target/AMDGPU/SMInstructions.td (+99-17) - (modified) llvm/test/CodeGen/AMDGPU/GlobalISel/addsubu64.ll (+10-10) - (modified) llvm/test/CodeGen/AMDGPU/GlobalISel/bool-legalization.ll (+4-4) - (modified) llvm/test/CodeGen/AMDGPU/GlobalISel/cvt_f32_ubyte.ll (+5-5) - (modified) llvm/test/CodeGen/AMDGPU/GlobalISel/fp-atomics-gfx940.ll (+4-4) - (modified) llvm/test/CodeGen/AMDGPU/GlobalISel/fp64-atomics-gfx90a.ll (+122-122) - (modified) llvm/test/CodeGen/AMDGPU/GlobalISel/frem.ll (+117-117) - (modified) llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-fract.f64.mir (+6-6) - (modified) llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-load-constant.mir (+36-36) - (modified) llvm/test/CodeGen/AMDGPU/GlobalISel/lds-zero-initializer.ll (+1-1) - (modified) llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.div.scale.ll (+215-190) - (modified) llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.intersect_ray.ll (+75-128) - (modified) llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.mfma.gfx90a.ll (+2-2) - (modified) llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.mov.dpp.ll (+21-21) - (modified) llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.set.inactive.ll (+90-90) - (modified) llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.update.dpp.ll (+64-64) - (modified) llvm/test/CodeGen/AMDGPU/GlobalISel/load-constant.96.ll (+15-5) - (modified) llvm/test/CodeGen/AMDGPU/GlobalISel/mul-known-bits.i64.ll (+33-33) - (modified) llvm/test/CodeGen/AMDGPU/GlobalISel/mul.ll (+82-82) - (modified) llvm/test/CodeGen/AMDGPU/GlobalISel/sdivrem.ll (+139-139) - (modified) llvm/test/CodeGen/AMDGPU/GlobalISel/udivrem.ll (+263-264) - (modified) llvm/test/CodeGen/AMDGPU/GlobalISel/widen-i8-i16-scalar-loads.ll (+146-146) - (modified) llvm/test/CodeGen/AMDGPU/add.ll (+273-272) - (modified) llvm/test/CodeGen/AMDGPU/add.v2i16.ll (+134-134) - (modified) llvm/test/CodeGen/AMDGPU/amd.endpgm.ll (+8-8) - (modified) llvm/test/CodeGen/AMDGPU/amdgcn-load-offset-from-reg.ll (+2-2) - (modified) llvm/test/CodeGen/AMDGPU/amdgpu-codegenprepare-idiv.ll (+614-611) - (modified) llvm/test/CodeGen/AMDGPU/and.ll (+6-6) - (modified) llvm/test/CodeGen/AMDGPU/anyext.ll (+11-11) - (modified) llvm/test/CodeGen/AMDGPU/atomic_optimizations_buffer.ll (+271-253) - (modified) llvm/test/CodeGen/AMDGPU/atomic_optimizations_global_pointer.ll (+1000-1005) - (modified) llvm/test/CodeGen/AMDGPU/atomic_optimizations_local_pointer.ll (+1095-1060) - (modified) llvm/test/CodeGen/AMDGPU/atomic_optimizations_raw_buffer.ll (+236-220) - (modified) llvm/test/CodeGen/AMDGPU/atomic_optimizations_struct_buffer.ll (+272-254) - (modified) llvm/test/CodeGen/AMDGPU/atomics_cond_sub.ll (+28-28) - (modified) llvm/test/CodeGen/AMDGPU/bfe-combine.ll (+28-28) - (modified) llvm/test/CodeGen/AMDGPU/bfe-patterns.ll (+36-36) - (modified) llvm/test/CodeGen/AMDGPU/bfi_int.ll (+68-68) - (modified) llvm/test/CodeGen/AMDGPU/bfm.ll (+6-6) - (modified) llvm/test/CodeGen/AMDGPU/bitreverse.ll (+70-70) - (modified) llvm/test/CodeGen/AMDGPU/br_cc.f16.ll (+37-37) - (modified) llvm/test/CodeGen/AMDGPU/branch-folding-implicit-def-subreg.ll (+1-1) - (modified) llvm/test/CodeGen/AMDGPU/bswap.ll (+78-78) - (modified) llvm/test/CodeGen/AMDGPU/build_vector.ll (+22-22) - (modified) llvm/test/CodeGen/AMDGPU/calling-conventions.ll (+162-162) - (modified) llvm/test/CodeGen/AMDGPU/carryout-selection.ll (+426-424) - (modified) llvm/test/CodeGen/AMDGPU/clamp-modifier.ll (+209-209) - (modified) llvm/test/CodeGen/AMDGPU/clamp.ll (+667-667) - (modified) llvm/test/CodeGen/AMDGPU/combine-cond-add-sub.ll (+15-15) - (modified) llvm/test/CodeGen/AMDGPU/combine-vload-extract.ll (+11-11) - (modified) llvm/test/CodeGen/AMDGPU/copy-illegal-type.ll (+127-130) - (modified) llvm/test/CodeGen/AMDGPU/copy_to_scc.ll (+11-11) - (modified) llvm/test/CodeGen/AMDGPU/ctlz.ll (+27-27) - (modified) llvm/test/CodeGen/AMDGPU/ctlz_zero_undef.ll (+43-43) - (modified) llvm/test/CodeGen/AMDGPU/ctpop16.ll (+26-26) - (modified) llvm/test/CodeGen/AMDGPU/ctpop64.ll (+38-38) - (modified) llvm/test/CodeGen/AMDGPU/cttz.ll (+16-16) - (modified) llvm/test/CodeGen/AMDGPU/cttz_zero_undef.ll (+24-24) - (modified) llvm/test/CodeGen/AMDGPU/cvt_f32_ubyte.ll (+14-14) - (modified) llvm/test/CodeGen/AMDGPU/dag-divergence-atomic.ll (+154-151) - (modified) llvm/test/CodeGen/AMDGPU/divergence-driven-buildvector.ll (+41-41) - (modified) llvm/test/CodeGen/AMDGPU/ds-alignment.ll (+135-135) - (modified) llvm/test/CodeGen/AMDGPU/ds-combine-with-dependence.ll (+1-1) - (modified) llvm/test/CodeGen/AMDGPU/ds_write2.ll (+23-23) - (modified) llvm/test/CodeGen/AMDGPU/extract_vector_dynelt.ll (+8-8) - (modified) llvm/test/CodeGen/AMDGPU/extract_vector_elt-f16.ll (+60-62) - (modified) llvm/test/CodeGen/AMDGPU/fabs.ll (+12-12) - (modified) llvm/test/CodeGen/AMDGPU/fadd.f16.ll (+211-211) - (modified) llvm/test/CodeGen/AMDGPU/fcanonicalize.f16.ll (+361-361) - (modified) llvm/test/CodeGen/AMDGPU/fcmp.f16.ll (+319-319) - (modified) llvm/test/CodeGen/AMDGPU/fcopysign.f16.ll (+208-207) - (modified) llvm/test/CodeGen/AMDGPU/fcopysign.f32.ll (+178-178) - (modified) llvm/test/CodeGen/AMDGPU/fcopysign.f64.ll (+155-155) - (modified) llvm/test/CodeGen/AMDGPU/fdiv.f16.ll (+199-199) - (modified) llvm/test/CodeGen/AMDGPU/fdiv.ll (+55-52) - (modified) llvm/test/CodeGen/AMDGPU/fdiv32-to-rcp-folding.ll (+92-92) - (modified) llvm/test/CodeGen/AMDGPU/flat_atomics.ll (+168-168) - (modified) llvm/test/CodeGen/AMDGPU/flat_atomics_i32_system.ll (+60-60) - (modified) llvm/test/CodeGen/AMDGPU/flat_atomics_i64.ll (+1117-1117) - (modified) llvm/test/CodeGen/AMDGPU/flat_atomics_i64_system.ll (+162-162) - (modified) llvm/test/CodeGen/AMDGPU/fma-combine.ll (+81-81) - (modified) llvm/test/CodeGen/AMDGPU/fmax3.ll (+224-224) - (modified) llvm/test/CodeGen/AMDGPU/fmax_legacy.f64.ll (+20-20) - (modified) llvm/test/CodeGen/AMDGPU/fmaximum.ll (+4-4) - (modified) llvm/test/CodeGen/AMDGPU/fmed3.ll (+1395-1395) - (modified) llvm/test/CodeGen/AMDGPU/fmin3.ll (+332-332) - (modified) llvm/test/CodeGen/AMDGPU/fmin_legacy.f64.ll (+40-40) - (modified) llvm/test/CodeGen/AMDGPU/fminimum.ll (+4-4) - (modified) llvm/test/CodeGen/AMDGPU/fmul.f16.ll (+171-171) - (modified) llvm/test/CodeGen/AMDGPU/fmuladd.f16.ll (+376-376) - (modified) llvm/test/CodeGen/AMDGPU/fnearbyint.ll (+45-45) - (modified) llvm/test/CodeGen/AMDGPU/fneg-combines.new.ll (+26-26) - (modified) llvm/test/CodeGen/AMDGPU/fneg-fabs.f64.ll (+42-42) - (modified) llvm/test/CodeGen/AMDGPU/fneg-fabs.ll (+24-24) - (modified) llvm/test/CodeGen/AMDGPU/fneg.ll (+137-137) - (modified) llvm/test/CodeGen/AMDGPU/fp-atomics-gfx940.ll (+10-10) - (modified) llvm/test/CodeGen/AMDGPU/fp-classify.ll (+141-141) - (modified) llvm/test/CodeGen/AMDGPU/fp-min-max-buffer-atomics.ll (+50-50) - (modified) llvm/test/CodeGen/AMDGPU/fp-min-max-buffer-ptr-atomics.ll (+18-18) - (modified) llvm/test/CodeGen/AMDGPU/fp64-atomics-gfx90a.ll (+131-131) - (modified) llvm/test/CodeGen/AMDGPU/fp64-min-max-buffer-atomics.ll (+28-28) - (modified) llvm/test/CodeGen/AMDGPU/fp64-min-max-buffer-ptr-atomics.ll (+14-14) - (modified) llvm/test/CodeGen/AMDGPU/fp_to_sint.ll (+63-63) - (modified) llvm/test/CodeGen/AMDGPU/fp_to_uint.ll (+52-52) - (modified) llvm/test/CodeGen/AMDGPU/fpext.f16.ll (+289-297) - (modified) llvm/test/CodeGen/AMDGPU/fptosi.f16.ll (+133-133) - (modified) llvm/test/CodeGen/AMDGPU/fptoui.f16.ll (+133-133) - (modified) llvm/test/CodeGen/AMDGPU/fptrunc.f16.ll (+486-486) - (modified) llvm/test/CodeGen/AMDGPU/fptrunc.ll (+406-406) - (modified) llvm/test/CodeGen/AMDGPU/frem.ll (+98-98) - (modified) llvm/test/CodeGen/AMDGPU/fshl.ll (+60-60) - (modified) llvm/test/CodeGen/AMDGPU/fshr.ll (+24-24) - (modified) llvm/test/CodeGen/AMDGPU/fsub.f16.ll (+126-126) - (modified) llvm/test/CodeGen/AMDGPU/fused-bitlogic.ll (+21-21) - (modified) llvm/test/CodeGen/AMDGPU/global-atomics-fp-wrong-subtarget.ll (+6-5) - (modified) llvm/test/CodeGen/AMDGPU/global-i16-load-store.ll (+8-8) - (modified) llvm/test/CodeGen/AMDGPU/global-load-saddr-to-vaddr.ll (+10-10) - (modified) llvm/test/CodeGen/AMDGPU/global_atomics.ll (+325-325) - (modified) llvm/test/CodeGen/AMDGPU/global_atomics_i32_system.ll (+90-90) - (modified) llvm/test/CodeGen/AMDGPU/global_atomics_i64.ll (+1323-1323) - (modified) llvm/test/CodeGen/AMDGPU/global_atomics_i64_system.ll (+282-282) - (modified) llvm/test/CodeGen/AMDGPU/global_atomics_scan_fadd.ll (+1022-1022) - (modified) llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmax.ll (+594-594) - (modified) llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmin.ll (+594-594) - (modified) llvm/test/CodeGen/AMDGPU/global_atomics_scan_fsub.ll (+806-806) - (modified) llvm/test/CodeGen/AMDGPU/identical-subrange-spill-infloop.ll (+15-15) - (modified) llvm/test/CodeGen/AMDGPU/idiv-licm.ll (+257-250) - (modified) llvm/test/CodeGen/AMDGPU/idot2.ll (+193-174) - (modified) llvm/test/CodeGen/AMDGPU/idot4s.ll (+312-295) - (modified) llvm/test/CodeGen/AMDGPU/idot4u.ll (+450-415) - (modified) llvm/test/CodeGen/AMDGPU/idot8s.ll (+128-130) - (modified) llvm/test/CodeGen/AMDGPU/idot8u.ll (+130-129) - (modified) llvm/test/CodeGen/AMDGPU/imm.ll (+321-321) - (modified) llvm/test/CodeGen/AMDGPU/indirect-addressing-term.ll (+1-1) - (modified) llvm/test/CodeGen/AMDGPU/insert_vector_dynelt.ll (+222-220) - (modified) llvm/test/CodeGen/AMDGPU/insert_vector_elt.ll (+7-7) - (modified) llvm/test/CodeGen/AMDGPU/insert_waitcnt_for_precise_memory.ll (+48-48) - (modified) llvm/test/CodeGen/AMDGPU/kernel-args.ll (+282-282) - (modified) llvm/test/CodeGen/AMDGPU/lds-atomic-fmin-fmax.ll (+190-190) - (modified) llvm/test/CodeGen/AMDGPU/lds-zero-initializer.ll (+1-1) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.cvt.pkrtz.ll (+62-62) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.exp.ll (+2-2) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.fcmp.w32.ll (+536-536) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.fcmp.w64.ll (+930-930) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.fdot2.bf16.bf16.ll (+19-18) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.fdot2.f16.f16.ll (+16-16) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.fdot2.f32.bf16.ll (+14-14) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.global.atomic.ordered.add.b64.ll (+14-13) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.global.load.tr-w32.ll (+6-6) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.global.load.tr-w64.ll (+6-6) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.icmp.w32.ll (+368-368) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.icmp.w64.ll (+634-634) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.intersect_ray.ll (+36-30) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.permlane.ll (+112-112) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.permlane16.var.ll (+96-96) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.permlane64.ll (+10-10) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.raw.ptr.tbuffer.store.d16.ll (+50-50) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.raw.tbuffer.store.d16.ll (+75-75) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.umax.ll (+155-155) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.umin.ll (+155-155) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.s.barrier.wait.ll (+167-167) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.sendmsg.rtn.ll (+48-48) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.set.inactive.ll (+107-107) - (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.ubfe.ll (+298-298) - (modified) llvm/test/CodeGen/AMDGPU/llvm.ceil.f16.ll (+60-60) - (modified) llvm/test/CodeGen/AMDGPU/llvm.cos.f16.ll (+28-28) - (modified) llvm/test/CodeGen/AMDGPU/llvm.exp.ll (+107-102) - (modified) llvm/test/CodeGen/AMDGPU/llvm.exp10.ll (+107-102) - (modified) llvm/test/CodeGen/AMDGPU/llvm.exp2.ll (+21-20) - (modified) llvm/test/CodeGen/AMDGPU/llvm.floor.f16.ll (+60-60) - (modified) llvm/test/CodeGen/AMDGPU/llvm.fmuladd.f16.ll (+268-268) - (modified) llvm/test/CodeGen/AMDGPU/llvm.get.fpmode.ll (+21-21) - (modified) llvm/test/CodeGen/AMDGPU/llvm.is.fpclass.bf16.ll (+15-15) - (modified) llvm/test/CodeGen/AMDGPU/llvm.is.fpclass.f16.ll (+16-16) - (modified) llvm/test/CodeGen/AMDGPU/llvm.is.fpclass.ll (+41-49) - (modified) llvm/test/CodeGen/AMDGPU/llvm.log.ll (+62-62) - (modified) llvm/test/CodeGen/AMDGPU/llvm.log10.ll (+62-62) - (modified) llvm/test/CodeGen/AMDGPU/llvm.log2.ll (+78-78) - (modified) llvm/test/CodeGen/AMDGPU/llvm.maxnum.f16.ll (+262-262) - (modified) llvm/test/CodeGen/AMDGPU/llvm.minnum.f16.ll (+260-260) - (modified) llvm/test/CodeGen/AMDGPU/llvm.mulo.ll (+212-212) - (modified) llvm/test/CodeGen/AMDGPU/llvm.r600.read.local.size.ll (+19-19) - (modified) llvm/test/CodeGen/AMDGPU/llvm.rint.f16.ll (+50-50) - (modified) llvm/test/CodeGen/AMDGPU/llvm.round.ll (+175-251) - (modified) llvm/test/CodeGen/AMDGPU/llvm.sin.f16.ll (+28-28) - (modified) llvm/test/CodeGen/AMDGPU/llvm.sqrt.f16.ll (+40-40) - (modified) llvm/test/CodeGen/AMDGPU/llvm.trunc.f16.ll (+40-40) - (modified) llvm/test/CodeGen/AMDGPU/load-constant-f64.ll (+10-10) - (modified) llvm/test/CodeGen/AMDGPU/load-constant-i1.ll (+1747-1747) - (modified) llvm/test/CodeGen/AMDGPU/load-constant-i16.ll (+793-792) - (modified) llvm/test/CodeGen/AMDGPU/load-constant-i32.ll (+206-207) - (modified) llvm/test/CodeGen/AMDGPU/load-constant-i64.ll (+44-44) - (modified) llvm/test/CodeGen/AMDGPU/load-constant-i8.ll (+1267-1266) - (modified) llvm/test/CodeGen/AMDGPU/load-global-i16.ll (+291-291) - (modified) llvm/test/CodeGen/AMDGPU/load-global-i32.ll (+224-224) - (modified) llvm/test/CodeGen/AMDGPU/loop-prefetch-data.ll (+39-39) - (modified) llvm/test/CodeGen/AMDGPU/lower-lds-struct-aa-memcpy.ll (+4-4) - (modified) llvm/test/CodeGen/AMDGPU/lower-lds-struct-aa.ll (+5-4) - (modified) llvm/test/CodeGen/AMDGPU/lshl-add-u64.ll (+3-3) - (modified) llvm/test/CodeGen/AMDGPU/lshr.v2i16.ll (+70-70) - (modified) llvm/test/CodeGen/AMDGPU/mad.u16.ll (+24-24) - (modified) llvm/test/CodeGen/AMDGPU/mad_64_32.ll (+11-11) - (modified) llvm/test/CodeGen/AMDGPU/madak.ll (+38-38) - (modified) llvm/test/CodeGen/AMDGPU/max-hard-clause-length.ll (+210-210) - (modified) llvm/test/CodeGen/AMDGPU/max.i16.ll (+24-24) - (modified) llvm/test/CodeGen/AMDGPU/memory_clause.ll (+13-12) - (modified) llvm/test/CodeGen/AMDGPU/min.ll (+66-66) - (modified) llvm/test/CodeGen/AMDGPU/move-to-valu-addsubu64.ll (+17-17) - (modified) llvm/test/CodeGen/AMDGPU/move-to-valu-pseudo-scalar-trans.ll (+30-30) - (modified) llvm/test/CodeGen/AMDGPU/mul.ll (+757-753) - (modified) llvm/test/CodeGen/AMDGPU/mul_int24.ll (+47-47) - (modified) llvm/test/CodeGen/AMDGPU/mul_uint24-amdgcn.ll (+73-73) - (modified) llvm/test/CodeGen/AMDGPU/offset-split-flat.ll (+445-445) - (modified) llvm/test/CodeGen/AMDGPU/offset-split-global.ll (+361-361) - (modified) llvm/test/CodeGen/AMDGPU/omod.ll (+44-44) - (modified) llvm/test/CodeGen/AMDGPU/optimize-compare.ll (+4-4) - (modified) llvm/test/CodeGen/AMDGPU/or.ll (+135-135) - (modified) llvm/test/CodeGen/AMDGPU/packed-op-sel.ll (+32-32) - (modified) llvm/test/CodeGen/AMDGPU/post-ra-soft-clause-dbg-info.ll (+4-4) - (modified) llvm/test/CodeGen/AMDGPU/preload-kernargs.ll (+6-6) - (modified) llvm/test/CodeGen/AMDGPU/promote-vect3-load.ll (+2-2) - (modified) llvm/test/CodeGen/AMDGPU/ptr-buffer-alias-scheduling.ll (+26-26) - (modified) llvm/test/CodeGen/AMDGPU/rcp-pattern.ll (+61-61) - (modified) llvm/test/CodeGen/AMDGPU/rotl.ll (+26-26) - (modified) llvm/test/CodeGen/AMDGPU/rotr.ll (+10-10) - (modified) llvm/test/CodeGen/AMDGPU/saddo.ll (+129-129) - (modified) llvm/test/CodeGen/AMDGPU/scalar_to_vector.ll (+26-26) - (modified) llvm/test/CodeGen/AMDGPU/sdiv.ll (+139-139) - (modified) llvm/test/CodeGen/AMDGPU/sdwa-peephole.ll (+185-185) - (modified) llvm/test/CodeGen/AMDGPU/select.f16.ll (+315-315) - (modified) llvm/test/CodeGen/AMDGPU/shl.ll (+193-193) - (modified) llvm/test/CodeGen/AMDGPU/shl.v2i16.ll (+70-70) - (modified) llvm/test/CodeGen/AMDGPU/shrink-add-sub-constant.ll (+682-682) - (modified) llvm/test/CodeGen/AMDGPU/si-annotate-cf.ll (+4-4) - (modified) llvm/test/CodeGen/AMDGPU/sign_extend.ll (+92-92) - (modified) llvm/test/CodeGen/AMDGPU/simple-indirect-call.ll (+10-10) - (modified) llvm/test/CodeGen/AMDGPU/sint_to_fp.i64.ll (+177-177) - (modified) llvm/test/CodeGen/AMDGPU/sitofp.f16.ll (+91-91) - (modified) llvm/test/CodeGen/AMDGPU/spill-scavenge-offset.ll (+2585-2585) - (modified) llvm/test/CodeGen/AMDGPU/sra.ll (+137-137) - (modified) llvm/test/CodeGen/AMDGPU/srl.ll (+35-35) - (modified) llvm/test/CodeGen/AMDGPU/sub.ll (+118-118) - (modified) llvm/test/CodeGen/AMDGPU/sub.v2i16.ll (+152-152) - (modified) llvm/test/CodeGen/AMDGPU/trap-abis.ll (+11-11) - (modified) llvm/test/CodeGen/AMDGPU/trunc-combine.ll (+7-7) - (modified) llvm/test/CodeGen/AMDGPU/twoaddr-constrain.ll (+3-3) - (modified) llvm/test/CodeGen/AMDGPU/uaddo.ll (+89-89) - (modified) llvm/test/CodeGen/AMDGPU/udiv.ll (+115-115) - (modified) llvm/test/CodeGen/AMDGPU/udivrem.ll (+74-72) - (modified) llvm/test/CodeGen/AMDGPU/uint_to_fp.i64.ll (+142-142) - (modified) llvm/test/CodeGen/AMDGPU/uitofp.f16.ll (+91-91) - (modified) llvm/test/CodeGen/AMDGPU/uniform-cfg.ll (+75-75) - (modified) llvm/test/CodeGen/AMDGPU/usubo.ll (+89-89) - (modified) llvm/test/CodeGen/AMDGPU/v_add_u64_pseudo_sdwa.ll (+4-4) - (modified) llvm/test/CodeGen/AMDGPU/v_cndmask.ll (+125-125) - (modified) llvm/test/CodeGen/AMDGPU/v_madak_f16.ll (+29-27) - (modified) llvm/test/CodeGen/AMDGPU/v_pack.ll (+41-41) - (modified) llvm/test/CodeGen/AMDGPU/v_sat_pk_u8_i16.ll (+45-45) - (modified) llvm/test/CodeGen/AMDGPU/v_sub_u64_pseudo_sdwa.ll (+4-4) - (modified) llvm/test/CodeGen/AMDGPU/vgpr-liverange-ir.ll (+5-5) - (modified) llvm/test/CodeGen/AMDGPU/vni8-across-blocks.ll (+5-5) - (modified) llvm/test/CodeGen/AMDGPU/wave32.ll (+142-140) - (modified) llvm/test/CodeGen/AMDGPU/widen-smrd-loads.ll (+45-45) - (modified) llvm/test/CodeGen/AMDGPU/xor.ll (+87-87) - (modified) llvm/test/CodeGen/AMDGPU/zero_extend.ll (+1-1) ``````````diff diff --git a/llvm/lib/Target/AMDGPU/SMInstructions.td b/llvm/lib/Target/AMDGPU/SMInstructions.td index 4551a3a615b15..9fbedce554a53 100644 --- a/llvm/lib/Target/AMDGPU/SMInstructions.td +++ b/llvm/lib/Target/AMDGPU/SMInstructions.td @@ -867,13 +867,104 @@ def SMRDBufferImm : ComplexPattern<iPTR, 1, "SelectSMRDBufferImm">; def SMRDBufferImm32 : ComplexPattern<iPTR, 1, "SelectSMRDBufferImm32">; def SMRDBufferSgprImm : ComplexPattern<iPTR, 2, "SelectSMRDBufferSgprImm">; +class SMRDAlignedLoadPat<PatFrag Op> : PatFrag <(ops node:$ptr), (Op node:$ptr), [{ + // Returns true if it is a naturally aligned multi-dword load. + LoadSDNode *Ld = cast<LoadSDNode>(N); + unsigned Size = Ld->getMemoryVT().getStoreSize(); + return (Size <= 4) || (Ld->getAlign().value() >= PowerOf2Ceil(Size)); +}]> { + let GISelPredicateCode = [{ + auto &Ld = cast<GLoad>(MI); + TypeSize Size = Ld.getMMO().getSize().getValue(); + return (Size <= 4) || (Ld.getMMO().getAlign().value() >= PowerOf2Ceil(Size)); + }]; +} + +class SMRDUnalignedLoadPat<PatFrag Op> : PatFrag <(ops node:$ptr), (Op node:$ptr), [{ + // Returns true if it is an under aligned multi-dword load. + LoadSDNode *Ld = cast<LoadSDNode>(N); + unsigned Size = Ld->getMemoryVT().getStoreSize(); + return (Size > 4) && (Ld->getAlign().value() < PowerOf2Ceil(Size)); +}]> { + let GISelPredicateCode = [{ + auto &Ld = cast<GLoad>(MI); + TypeSize Size = Ld.getMMO().getSize().getValue(); + return (Size > 4) && (Ld.getMMO().getAlign().value() < PowerOf2Ceil(Size)); + }]; +} + +def alignedmultidwordload : SMRDAlignedLoadPat<smrd_load>; +def unalignedmultidwordload : SMRDUnalignedLoadPat<smrd_load>; + +multiclass SMRD_Align_Pattern <string Instr, ValueType vt> { + + // 1. IMM offset + def : GCNPat < + (alignedmultidwordload (SMRDImm i64:$sbase, i32:$offset)), + (vt (!cast<SM_Pseudo>(Instr#"_IMM") $sbase, $offset, 0))> { + let OtherPredicates = [isGFX8Plus]; + } + def : GCNPat < + (unalignedmultidwordload (SMRDImm i64:$sbase, i32:$offset)), + (vt (!cast<SM_Pseudo>(Instr#"_IMM_ec") $sbase, $offset, 0))> { + let OtherPredicates = [isGFX8Plus]; + } + + // 2. SGPR offset + def : GCNPat < + (alignedmultidwordload (SMRDSgpr i64:$sbase, i32:$soffset)), + (vt (!cast<SM_Pseudo>(Instr#"_SGPR") $sbase, $soffset, 0))> { + let OtherPredicates = [isGFX8Only]; + } + def : GCNPat < + (unalignedmultidwordload (SMRDSgpr i64:$sbase, i32:$soffset)), + (vt (!cast<SM_Pseudo>(Instr#"_SGPR_ec") $sbase, $soffset, 0))> { + let OtherPredicates = [isGFX8Only]; + } + def : GCNPat < + (alignedmultidwordload (SMRDSgpr i64:$sbase, i32:$soffset)), + (vt (!cast<SM_Pseudo>(Instr#"_SGPR_IMM") $sbase, $soffset, 0, 0))> { + let OtherPredicates = [isGFX9Plus]; + } + def : GCNPat < + (unalignedmultidwordload (SMRDSgpr i64:$sbase, i32:$soffset)), + (vt (!cast<SM_Pseudo>(Instr#"_SGPR_IMM_ec") $sbase, $soffset, 0, 0))> { + let OtherPredicates = [isGFX9Plus]; + } + + // 3. SGPR+IMM offset + def : GCNPat < + (alignedmultidwordload (SMRDSgprImm i64:$sbase, i32:$soffset, i32:$offset)), + (vt (!cast<SM_Pseudo>(Instr#"_SGPR_IMM") $sbase, $soffset, $offset, 0))> { + let OtherPredicates = [isGFX9Plus]; + } + def : GCNPat < + (unalignedmultidwordload (SMRDSgprImm i64:$sbase, i32:$soffset, i32:$offset)), + (vt (!cast<SM_Pseudo>(Instr#"_SGPR_IMM_ec") $sbase, $soffset, $offset, 0))> { + let OtherPredicates = [isGFX9Plus]; + } + + // 4. No offset + def : GCNPat < + (vt (alignedmultidwordload (i64 SReg_64:$sbase))), + (vt (!cast<SM_Pseudo>(Instr#"_IMM") i64:$sbase, 0, 0))> { + let OtherPredicates = [isGFX8Plus]; + } + def : GCNPat < + (vt (unalignedmultidwordload (i64 SReg_64:$sbase))), + (vt (!cast<SM_Pseudo>(Instr#"_IMM_ec") i64:$sbase, 0, 0))> { + let OtherPredicates = [isGFX8Plus]; + } +} + multiclass SMRD_Pattern <string Instr, ValueType vt, bit immci = true> { // 1. IMM offset def : GCNPat < (smrd_load (SMRDImm i64:$sbase, i32:$offset)), - (vt (!cast<SM_Pseudo>(Instr#"_IMM") $sbase, $offset, 0)) - >; + (vt (!cast<SM_Pseudo>(Instr#"_IMM") $sbase, $offset, 0))> { + let OtherPredicates = [isGFX6GFX7]; + } // 2. 32-bit IMM offset on CI if immci then def : GCNPat < @@ -886,26 +977,17 @@ multiclass SMRD_Pattern <string Instr, ValueType vt, bit immci = true> { def : GCNPat < (smrd_load (SMRDSgpr i64:$sbase, i32:$soffset)), (vt (!cast<SM_Pseudo>(Instr#"_SGPR") $sbase, $soffset, 0))> { - let OtherPredicates = [isNotGFX9Plus]; - } - def : GCNPat < - (smrd_load (SMRDSgpr i64:$sbase, i32:$soffset)), - (vt (!cast<SM_Pseudo>(Instr#"_SGPR_IMM") $sbase, $soffset, 0, 0))> { - let OtherPredicates = [isGFX9Plus]; + let OtherPredicates = [isGFX6GFX7]; } - // 4. SGPR+IMM offset + // 4. No offset def : GCNPat < - (smrd_load (SMRDSgprImm i64:$sbase, i32:$soffset, i32:$offset)), - (vt (!cast<SM_Pseudo>(Instr#"_SGPR_IMM") $sbase, $soffset, $offset, 0))> { - let OtherPredicates = [isGFX9Plus]; + (vt (smrd_load (i64 SReg_64:$sbase))), + (vt (!cast<SM_Pseudo>(Instr#"_IMM") i64:$sbase, 0, 0))> { + let OtherPredicates = [isGFX6GFX7]; } - // 5. No offset - def : GCNPat < - (vt (smrd_load (i64 SReg_64:$sbase))), - (vt (!cast<SM_Pseudo>(Instr#"_IMM") i64:$sbase, 0, 0)) - >; + defm : SMRD_Align_Pattern<Instr, vt>; } multiclass SMLoad_Pattern <string Instr, ValueType vt, bit immci = true> { diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/addsubu64.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/addsubu64.ll index a38b6e3263882..9a8672dba5357 100644 --- a/llvm/test/CodeGen/AMDGPU/GlobalISel/addsubu64.ll +++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/addsubu64.ll @@ -7,11 +7,11 @@ define amdgpu_kernel void @s_add_u64(ptr addrspace(1) %out, i64 %a, i64 %b) { ; GFX11: ; %bb.0: ; %entry ; GFX11-NEXT: s_clause 0x1 ; GFX11-NEXT: s_load_b128 s[4:7], s[0:1], 0x24 -; GFX11-NEXT: s_load_b64 s[0:1], s[0:1], 0x34 +; GFX11-NEXT: s_load_b64 s[2:3], s[0:1], 0x34 ; GFX11-NEXT: v_mov_b32_e32 v2, 0 ; GFX11-NEXT: s_waitcnt lgkmcnt(0) -; GFX11-NEXT: s_add_u32 s0, s6, s0 -; GFX11-NEXT: s_addc_u32 s1, s7, s1 +; GFX11-NEXT: s_add_u32 s0, s6, s2 +; GFX11-NEXT: s_addc_u32 s1, s7, s3 ; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX11-NEXT: v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, s1 ; GFX11-NEXT: global_store_b64 v2, v[0:1], s[4:5] @@ -23,10 +23,10 @@ define amdgpu_kernel void @s_add_u64(ptr addrspace(1) %out, i64 %a, i64 %b) { ; GFX12: ; %bb.0: ; %entry ; GFX12-NEXT: s_clause 0x1 ; GFX12-NEXT: s_load_b128 s[4:7], s[0:1], 0x24 -; GFX12-NEXT: s_load_b64 s[0:1], s[0:1], 0x34 +; GFX12-NEXT: s_load_b64 s[2:3], s[0:1], 0x34 ; GFX12-NEXT: v_mov_b32_e32 v2, 0 ; GFX12-NEXT: s_wait_kmcnt 0x0 -; GFX12-NEXT: s_add_nc_u64 s[0:1], s[6:7], s[0:1] +; GFX12-NEXT: s_add_nc_u64 s[0:1], s[6:7], s[2:3] ; GFX12-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX12-NEXT: v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, s1 ; GFX12-NEXT: global_store_b64 v2, v[0:1], s[4:5] @@ -59,11 +59,11 @@ define amdgpu_kernel void @s_sub_u64(ptr addrspace(1) %out, i64 %a, i64 %b) { ; GFX11: ; %bb.0: ; %entry ; GFX11-NEXT: s_clause 0x1 ; GFX11-NEXT: s_load_b128 s[4:7], s[0:1], 0x24 -; GFX11-NEXT: s_load_b64 s[0:1], s[0:1], 0x34 +; GFX11-NEXT: s_load_b64 s[2:3], s[0:1], 0x34 ; GFX11-NEXT: v_mov_b32_e32 v2, 0 ; GFX11-NEXT: s_waitcnt lgkmcnt(0) -; GFX11-NEXT: s_sub_u32 s0, s6, s0 -; GFX11-NEXT: s_subb_u32 s1, s7, s1 +; GFX11-NEXT: s_sub_u32 s0, s6, s2 +; GFX11-NEXT: s_subb_u32 s1, s7, s3 ; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX11-NEXT: v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, s1 ; GFX11-NEXT: global_store_b64 v2, v[0:1], s[4:5] @@ -75,10 +75,10 @@ define amdgpu_kernel void @s_sub_u64(ptr addrspace(1) %out, i64 %a, i64 %b) { ; GFX12: ; %bb.0: ; %entry ; GFX12-NEXT: s_clause 0x1 ; GFX12-NEXT: s_load_b128 s[4:7], s[0:1], 0x24 -; GFX12-NEXT: s_load_b64 s[0:1], s[0:1], 0x34 +; GFX12-NEXT: s_load_b64 s[2:3], s[0:1], 0x34 ; GFX12-NEXT: v_mov_b32_e32 v2, 0 ; GFX12-NEXT: s_wait_kmcnt 0x0 -; GFX12-NEXT: s_sub_nc_u64 s[0:1], s[6:7], s[0:1] +; GFX12-NEXT: s_sub_nc_u64 s[0:1], s[6:7], s[2:3] ; GFX12-NEXT: s_delay_alu instid0(SALU_CYCLE_1) ; GFX12-NEXT: v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, s1 ; GFX12-NEXT: global_store_b64 v2, v[0:1], s[4:5] diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/bool-legalization.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/bool-legalization.ll index bb5ccc3657dc4..57a8bbbb7d185 100644 --- a/llvm/test/CodeGen/AMDGPU/GlobalISel/bool-legalization.ll +++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/bool-legalization.ll @@ -113,9 +113,9 @@ bb1: define amdgpu_kernel void @brcond_sgpr_trunc_and(i32 %cond0, i32 %cond1) { ; WAVE64-LABEL: brcond_sgpr_trunc_and: ; WAVE64: ; %bb.0: ; %entry -; WAVE64-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24 +; WAVE64-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x24 ; WAVE64-NEXT: s_waitcnt lgkmcnt(0) -; WAVE64-NEXT: s_and_b32 s0, s0, s1 +; WAVE64-NEXT: s_and_b32 s0, s2, s3 ; WAVE64-NEXT: s_xor_b32 s0, s0, 1 ; WAVE64-NEXT: s_and_b32 s0, s0, 1 ; WAVE64-NEXT: s_cmp_lg_u32 s0, 0 @@ -131,9 +131,9 @@ define amdgpu_kernel void @brcond_sgpr_trunc_and(i32 %cond0, i32 %cond1) { ; ; WAVE32-LABEL: brcond_sgpr_trunc_and: ; WAVE32: ; %bb.0: ; %entry -; WAVE32-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24 +; WAVE32-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x24 ; WAVE32-NEXT: s_waitcnt lgkmcnt(0) -; WAVE32-NEXT: s_and_b32 s0, s0, s1 +; WAVE32-NEXT: s_and_b32 s0, s2, s3 ; WAVE32-NEXT: s_xor_b32 s0, s0, 1 ; WAVE32-NEXT: s_and_b32 s0, s0, 1 ; WAVE32-NEXT: s_cmp_lg_u32 s0, 0 diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/cvt_f32_ubyte.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/cvt_f32_ubyte.ll index 3f034eaca4997..9cabe0c0ae9de 100644 --- a/llvm/test/CodeGen/AMDGPU/GlobalISel/cvt_f32_ubyte.ll +++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/cvt_f32_ubyte.ll @@ -1400,11 +1400,11 @@ define amdgpu_kernel void @cvt_ubyte0_or_multiuse(ptr addrspace(1) %in, ptr addr ; ; VI-LABEL: cvt_ubyte0_or_multiuse: ; VI: ; %bb.0: ; %bb -; VI-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x24 +; VI-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24 ; VI-NEXT: v_lshlrev_b32_e32 v2, 2, v0 ; VI-NEXT: s_waitcnt lgkmcnt(0) -; VI-NEXT: v_mov_b32_e32 v0, s0 -; VI-NEXT: v_mov_b32_e32 v1, s1 +; VI-NEXT: v_mov_b32_e32 v0, s4 +; VI-NEXT: v_mov_b32_e32 v1, s5 ; VI-NEXT: v_add_u32_e32 v0, vcc, v0, v2 ; VI-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc ; VI-NEXT: flat_load_dword v0, v[0:1] @@ -1412,8 +1412,8 @@ define amdgpu_kernel void @cvt_ubyte0_or_multiuse(ptr addrspace(1) %in, ptr addr ; VI-NEXT: v_or_b32_e32 v0, 0x80000001, v0 ; VI-NEXT: v_cvt_f32_ubyte0_sdwa v1, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 ; VI-NEXT: v_add_f32_e32 v2, v0, v1 -; VI-NEXT: v_mov_b32_e32 v0, s2 -; VI-NEXT: v_mov_b32_e32 v1, s3 +; VI-NEXT: v_mov_b32_e32 v0, s6 +; VI-NEXT: v_mov_b32_e32 v1, s7 ; VI-NEXT: flat_store_dword v[0:1], v2 ; VI-NEXT: s_endpgm bb: diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/fp-atomics-gfx940.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/fp-atomics-gfx940.ll index a018ea5bf18f1..ce0d9c3c5365e 100644 --- a/llvm/test/CodeGen/AMDGPU/GlobalISel/fp-atomics-gfx940.ll +++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/fp-atomics-gfx940.ll @@ -27,10 +27,10 @@ define amdgpu_kernel void @flat_atomic_fadd_f32_noret(ptr %ptr, float %data) { define amdgpu_kernel void @flat_atomic_fadd_f32_noret_pat(ptr %ptr) { ; GFX940-LABEL: flat_atomic_fadd_f32_noret_pat: ; GFX940: ; %bb.0: -; GFX940-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24 +; GFX940-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x24 ; GFX940-NEXT: v_mov_b32_e32 v2, 4.0 ; GFX940-NEXT: s_waitcnt lgkmcnt(0) -; GFX940-NEXT: v_mov_b64_e32 v[0:1], s[0:1] +; GFX940-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX940-NEXT: buffer_wbl2 sc0 sc1 ; GFX940-NEXT: flat_atomic_add_f32 v[0:1], v2 sc1 ; GFX940-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0) @@ -43,10 +43,10 @@ define amdgpu_kernel void @flat_atomic_fadd_f32_noret_pat(ptr %ptr) { define amdgpu_kernel void @flat_atomic_fadd_f32_noret_pat_ieee(ptr %ptr) #0 { ; GFX940-LABEL: flat_atomic_fadd_f32_noret_pat_ieee: ; GFX940: ; %bb.0: -; GFX940-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24 +; GFX940-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x24 ; GFX940-NEXT: v_mov_b32_e32 v2, 4.0 ; GFX940-NEXT: s_waitcnt lgkmcnt(0) -; GFX940-NEXT: v_mov_b64_e32 v[0:1], s[0:1] +; GFX940-NEXT: v_mov_b64_e32 v[0:1], s[2:3] ; GFX940-NEXT: buffer_wbl2 sc0 sc1 ; GFX940-NEXT: flat_atomic_add_f32 v[0:1], v2 sc1 ; GFX940-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0) diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/fp64-atomics-gfx90a.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/fp64-atomics-gfx90a.ll index 4e94a646f6da5..081e25708c067 100644 --- a/llvm/test/CodeGen/AMDGPU/GlobalISel/fp64-atomics-gfx90a.ll +++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/fp64-atomics-gfx90a.ll @@ -1021,20 +1021,20 @@ main_body: define amdgpu_kernel void @global_atomic_fadd_f64_noret(ptr addrspace(1) %ptr, double %data) { ; GFX90A-LABEL: global_atomic_fadd_f64_noret: ; GFX90A: ; %bb.0: ; %main_body -; GFX90A-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x24 +; GFX90A-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24 ; GFX90A-NEXT: v_mov_b32_e32 v2, 0 ; GFX90A-NEXT: s_waitcnt lgkmcnt(0) -; GFX90A-NEXT: v_pk_mov_b32 v[0:1], s[2:3], s[2:3] op_sel:[0,1] -; GFX90A-NEXT: global_atomic_add_f64 v2, v[0:1], s[0:1] +; GFX90A-NEXT: v_pk_mov_b32 v[0:1], s[6:7], s[6:7] op_sel:[0,1] +; GFX90A-NEXT: global_atomic_add_f64 v2, v[0:1], s[4:5] ; GFX90A-NEXT: s_endpgm ; ; GFX940-LABEL: global_atomic_fadd_f64_noret: ; GFX940: ; %bb.0: ; %main_body -; GFX940-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x24 +; GFX940-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24 ; GFX940-NEXT: v_mov_b32_e32 v2, 0 ; GFX940-NEXT: s_waitcnt lgkmcnt(0) -; GFX940-NEXT: v_mov_b64_e32 v[0:1], s[2:3] -; GFX940-NEXT: global_atomic_add_f64 v2, v[0:1], s[0:1] +; GFX940-NEXT: v_mov_b64_e32 v[0:1], s[6:7] +; GFX940-NEXT: global_atomic_add_f64 v2, v[0:1], s[4:5] ; GFX940-NEXT: s_endpgm main_body: %ret = call double @llvm.amdgcn.global.atomic.fadd.f64.p1.f64(ptr addrspace(1) %ptr, double %data) @@ -1044,20 +1044,20 @@ main_body: define amdgpu_kernel void @global_atomic_fmin_f64_noret(ptr addrspace(1) %ptr, double %data) { ; GFX90A-LABEL: global_atomic_fmin_f64_noret: ; GFX90A: ; %bb.0: ; %main_body -; GFX90A-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x24 +; GFX90A-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24 ; GFX90A-NEXT: v_mov_b32_e32 v2, 0 ; GFX90A-NEXT: s_waitcnt lgkmcnt(0) -; GFX90A-NEXT: v_pk_mov_b32 v[0:1], s[2:3], s[2:3] op_sel:[0,1] -; GFX90A-NEXT: global_atomic_min_f64 v2, v[0:1], s[0:1] +; GFX90A-NEXT: v_pk_mov_b32 v[0:1], s[6:7], s[6:7] op_sel:[0,1] +; GFX90A-NEXT: global_atomic_min_f64 v2, v[0:1], s[4:5] ; GFX90A-NEXT: s_endpgm ; ; GFX940-LABEL: global_atomic_fmin_f64_noret: ; GFX940: ; %bb.0: ; %main_body -; GFX940-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x24 +; GFX940-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24 ; GFX940-NEXT: v_mov_b32_e32 v2, 0 ; GFX940-NEXT: s_waitcnt lgkmcnt(0) -; GFX940-NEXT: v_mov_b64_e32 v[0:1], s[2:3] -; GFX940-NEXT: global_atomic_min_f64 v2, v[0:1], s[0:1] +; GFX940-NEXT: v_mov_b64_e32 v[0:1], s[6:7] +; GFX940-NEXT: global_atomic_min_f64 v2, v[0:1], s[4:5] ; GFX940-NEXT: s_endpgm main_body: %ret = call double @llvm.amdgcn.global.atomic.fmin.f64.p1.f64(ptr addrspace(1) %ptr, double %data) @@ -1067,20 +1067,20 @@ main_body: define amdgpu_kernel void @global_atomic_fmax_f64_noret(ptr addrspace(1) %ptr, double %data) { ; GFX90A-LABEL: global_atomic_fmax_f64_noret: ; GFX90A: ; %bb.0: ; %main_body -; GFX90A-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x24 +; GFX90A-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24 ; GFX90A-NEXT: v_mov_b32_e32 v2, 0 ; GFX90A-NEXT: s_waitcnt lgkmcnt(0) -; GFX90A-NEXT: v_pk_mov_b32 v[0:1], s[2:3], s[2:3] op_sel:[0,1] -; GFX90A-NEXT: global_atomic_max_f64 v2, v[0:1], s[0:1] +; GFX90A-NEXT: v_pk_mov_b32 v[0:1], s[6:7], s[6:7] op_sel:[0,1] +; GFX90A-NEXT: global_atomic_max_f64 v2, v[0:1], s[4:5] ; GFX90A-NEXT: s_endpgm ; ; GFX940-LABEL: global_atomic_fmax_f64_noret: ; GFX940: ; %bb.0: ; %main_body -; GFX940-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x24 +; GFX940-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24 ; GFX940-NEXT: v_mov_b32_e32 v2, 0 ; GFX940-NEXT: s_waitcnt lgkmcnt(0) -; GFX940-NEXT: v_mov_b64_e32 v[0:1], s[2:3] -; GFX940-NEXT: global_atomic_max_f64 v2, v[0:1], s[0:1] +; GFX940-NEXT: v_mov_b64_e32 v[0:1], s[6:7] +; GFX940-NEXT: global_atomic_max_f64 v2, v[0:1], s[4:5] ; GFX940-NEXT: s_endpgm main_body: %ret = call double @llvm.amdgcn.global.atomic.fmax.f64.p1.f64(ptr addrspace(1) %ptr, double %data) @@ -1090,21 +1090,21 @@ main_body: define amdgpu_kernel void @global_atomic_fadd_f64_noret_pat(ptr addrspace(1) %ptr) #1 { ; GFX90A-LABEL: global_atomic_fadd_f64_noret_pat: ; GFX90A: ; %bb.0: ; %main_body -; GFX90A-NEXT: s_mov_b64 s[2:3], exec -; GFX90A-NEXT: s_mov_b32 s4, s3 -; GFX90A-NEXT: v_mbcnt_lo_u32_b32 v0, s2, 0 -; GFX90A-NEXT: v_mbcnt_hi_u32_b32 v0, s4, v0 +; GFX90A-NEXT: s_mov_b64 s[4:5], exec +; GFX90A-NEXT: s_mov_b32 s2, s5 +; GFX90A-NEXT: v_mbcnt_lo_u32_b32 v0, s4, 0 +; GFX90A-NEXT: v_mbcnt_hi_u32_b32 v0, s2, v0 ; GFX90A-NEXT: v_cmp_eq_u32_e32 vcc, 0, v0 -; GFX90A-NEXT: s_and_saveexec_b64 s[4:5], vcc +; GFX90A-NEXT: s_and_saveexec_b64 s[2:3], vcc ; GFX90A-NEXT: s_cbranch_execz .LBB39_3 ; GFX90A-NEXT: ; %bb.1: -; GFX90A-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24 -; GFX90A-NEXT: s_bcnt1_i32_b64 s2, s[2:3] -; GFX90A-NEXT: v_cvt_f64_u32_e32 v[0:1], s2 +; GFX90A-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x24 +; GFX90A-NEXT: s_bcnt1_i32_b64 s0, s[4:5] +; GFX90A-NEXT: v_cvt_f64_u32_e32 v[0:1], s0 ; GFX90A-NEXT: v_mul_f64 v[4:5], v[0:1], 4.0 -; GFX90A-NEXT: s_mov_b64 s[2:3], 0 +; GFX90A-NEXT: s_mov_b64 s[0:1], 0 ; GFX90A-NEXT: s_waitcnt lgkmcnt(0) -; GFX90A-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x0 +; GFX90A-NEXT: s_load_dwordx2 s[4:5], s[2:3], 0x0 ; GFX90A-NEXT: v_mov_b32_e32 v6, 0 ; GFX90A-NEXT: s_waitcnt lgkmcnt(0) ; GFX90A-NEXT: v_pk_mov_b32 v[2:3], s[4:5], s[4:5] op_sel:[0,1] @@ -1112,14 +1112,14 @@ define amdgpu_kernel void @global_atomic_fadd_f64_noret_pat(ptr addrspace(1) %pt ; GFX90A-NEXT: ; =>This Inner Loop Header: Depth=1 ; GFX90A-NEXT: v_add_f64 v[0:1], v[2:3], v[4:5] ; GFX90A-NEXT: buffer_wbl2 -; GFX90A-NEXT: global_atomic_cmpswap_x2 v[0:1], v6, v[0:3], s[0:1] glc +; GFX90A-NEXT: global_atomic_cmpswap_x2 v[0:1], v6, v[0:3], s[2:3] glc ; GFX90A-NEXT: s_waitcnt vmcnt(0) ; GFX90A-NEXT: buffer_invl2 ; GFX90A-NEXT: buffer_wbinvl1_vol ; GFX90A-NEXT: v_cmp_eq_u64_e32 vcc, v[0:1], v[2:3] -; GFX90A-NEXT: s_or_b64 s[2:3], vcc, s[2:3] +; GFX90A-NEXT: s_or_b64 s[0:1], vcc, s[0:1] ; GFX90A-NEXT: v_pk_mov_b32 v[2:3], v[0:1], v[0:1] op_sel:[0,1] -; GFX90A-NEXT: s_andn2_b64 exec, exec, s[2:3] +; GFX90A-NEXT: s_andn2_b64 exec, exec, s[0:1] ; GFX90A-NEXT: s_cbranch_execnz .LBB39_2 ; GFX90A-NEXT: .LBB39_3: ; GFX90A-NEXT: s_endpgm @@ -1134,14 +1134,14 @@ define amdgpu_kernel void @global_atomic_fadd_f64_noret_pat(ptr addrspace(1) %pt ; GFX940-NEXT: s_and_saveexec_b64 s[4:5], vcc ; GFX940-NEXT: s_cbranch_execz .LBB39_2 ; GFX940-NEXT: ; %bb.1: -; GFX940-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x24 -; GFX940-NEXT: s_bcnt1_i32_b64 s2, s[2:3] -; GFX940-NEXT: v_cvt_f64_u32_e32 v[0:1], s2 +; GFX940-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x24 +; GFX940-NEXT: s_bcnt1_i32_b64 s0, s[2:3] +; GFX940-NEXT: v_cvt_f64_u32_e32 v[0:1], s0 ; GFX940-NEXT: v_mul_f64 v[0:1], v[0:1], 4.0 ; GFX940-NEXT: v_mov_b32_e32 v2, 0 ; GFX940-NEXT: buffer_wbl2 sc0 sc1 ; GFX940-NEXT: s_waitcnt lgkmcnt(0) -; GFX940-NEXT: global_atomic_add_f64 v2, v[0:1], s[0:1] sc1 +; GFX940-NEXT: global_atomic_add_f64 v2, v[0:1], s[4:5] sc1 ; GFX940-NEXT: s_waitcnt vmcnt(0) ; G... [truncated] `````````` </details> https://github.com/llvm/llvm-project/pull/96163 _______________________________________________ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
