CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
90 lines
4.8 KiB
LLVM
90 lines
4.8 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
|
|
; RUN: opt < %s -S -unroll-runtime -unroll-count=2 -passes=loop-unroll -unroll-runtime-epilog=true | FileCheck %s -check-prefix=EPILOG
|
|
; RUN: opt < %s -S -unroll-runtime -unroll-count=2 -passes=loop-unroll -unroll-runtime-epilog=false | FileCheck %s -check-prefix=PROLOG
|
|
target datalayout = "e-m:o-i64:64-f80:128-n8:16:32:64-S128"
|
|
|
|
; This test case documents how runtime loop unrolling handles the case
|
|
; when the backedge-count is -1.
|
|
|
|
; If %N, the backedge-taken count, is -1 then %0 unsigned-overflows
|
|
; and is 0. %xtraiter too is 0, signifying that the total trip-count
|
|
; is divisible by 2. The prologue then branches to the unrolled loop
|
|
; and executes the 2^32 iterations there, in groups of 2.
|
|
|
|
define i32 @foo(i32 %N) {
|
|
; EPILOG-LABEL: @foo(
|
|
; EPILOG-NEXT: entry:
|
|
; EPILOG-NEXT: [[TMP0:%.*]] = add i32 [[N:%.*]], 1
|
|
; EPILOG-NEXT: [[XTRAITER:%.*]] = and i32 [[TMP0]], 1
|
|
; EPILOG-NEXT: [[TMP1:%.*]] = icmp ult i32 [[N]], 1
|
|
; EPILOG-NEXT: br i1 [[TMP1]], label [[WHILE_BODY_EPIL_PREHEADER:%.*]], label [[ENTRY_NEW:%.*]]
|
|
; EPILOG: entry.new:
|
|
; EPILOG-NEXT: [[UNROLL_ITER:%.*]] = sub i32 [[TMP0]], [[XTRAITER]]
|
|
; EPILOG-NEXT: br label [[WHILE_BODY:%.*]]
|
|
; EPILOG: while.body:
|
|
; EPILOG-NEXT: [[I:%.*]] = phi i32 [ 0, [[ENTRY_NEW]] ], [ [[INC_1:%.*]], [[WHILE_BODY]] ]
|
|
; EPILOG-NEXT: [[NITER:%.*]] = phi i32 [ 0, [[ENTRY_NEW]] ], [ [[NITER_NEXT_1:%.*]], [[WHILE_BODY]] ]
|
|
; EPILOG-NEXT: [[INC:%.*]] = add nuw nsw i32 [[I]], 1
|
|
; EPILOG-NEXT: [[INC_1]] = add i32 [[I]], 2
|
|
; EPILOG-NEXT: [[NITER_NEXT_1]] = add i32 [[NITER]], 2
|
|
; EPILOG-NEXT: [[NITER_NCMP_1:%.*]] = icmp eq i32 [[NITER_NEXT_1]], [[UNROLL_ITER]]
|
|
; EPILOG-NEXT: br i1 [[NITER_NCMP_1]], label [[WHILE_END_UNR_LCSSA:%.*]], label [[WHILE_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
|
|
; EPILOG: while.end.unr-lcssa:
|
|
; EPILOG-NEXT: [[I_LCSSA_PH_PH:%.*]] = phi i32 [ [[INC]], [[WHILE_BODY]] ]
|
|
; EPILOG-NEXT: [[I_UNR_PH:%.*]] = phi i32 [ [[INC_1]], [[WHILE_BODY]] ]
|
|
; EPILOG-NEXT: [[LCMP_MOD:%.*]] = icmp ne i32 [[XTRAITER]], 0
|
|
; EPILOG-NEXT: br i1 [[LCMP_MOD]], label [[WHILE_BODY_EPIL_PREHEADER]], label [[WHILE_END:%.*]]
|
|
; EPILOG: while.body.epil.preheader:
|
|
; EPILOG-NEXT: [[I_EPIL_INIT:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[I_UNR_PH]], [[WHILE_END_UNR_LCSSA]] ]
|
|
; EPILOG-NEXT: [[LCMP_MOD2:%.*]] = icmp ne i32 [[XTRAITER]], 0
|
|
; EPILOG-NEXT: call void @llvm.assume(i1 [[LCMP_MOD2]])
|
|
; EPILOG-NEXT: br label [[WHILE_BODY_EPIL:%.*]]
|
|
; EPILOG: while.body.epil:
|
|
; EPILOG-NEXT: br label [[WHILE_END]]
|
|
; EPILOG: while.end:
|
|
; EPILOG-NEXT: [[I_LCSSA:%.*]] = phi i32 [ [[I_LCSSA_PH_PH]], [[WHILE_END_UNR_LCSSA]] ], [ [[I_EPIL_INIT]], [[WHILE_BODY_EPIL]] ]
|
|
; EPILOG-NEXT: ret i32 [[I_LCSSA]]
|
|
;
|
|
; PROLOG-LABEL: @foo(
|
|
; PROLOG-NEXT: entry:
|
|
; PROLOG-NEXT: [[TMP0:%.*]] = add i32 [[N:%.*]], 1
|
|
; PROLOG-NEXT: [[XTRAITER:%.*]] = and i32 [[TMP0]], 1
|
|
; PROLOG-NEXT: [[LCMP_MOD:%.*]] = icmp ne i32 [[XTRAITER]], 0
|
|
; PROLOG-NEXT: br i1 [[LCMP_MOD]], label [[WHILE_BODY_PROL_PREHEADER:%.*]], label [[WHILE_BODY_PROL_LOOPEXIT:%.*]]
|
|
; PROLOG: while.body.prol.preheader:
|
|
; PROLOG-NEXT: br label [[WHILE_BODY_PROL:%.*]]
|
|
; PROLOG: while.body.prol:
|
|
; PROLOG-NEXT: br label [[WHILE_BODY_PROL_LOOPEXIT]]
|
|
; PROLOG: while.body.prol.loopexit:
|
|
; PROLOG-NEXT: [[I_LCSSA_UNR:%.*]] = phi i32 [ poison, [[ENTRY:%.*]] ], [ 0, [[WHILE_BODY_PROL]] ]
|
|
; PROLOG-NEXT: [[I_UNR:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ 1, [[WHILE_BODY_PROL]] ]
|
|
; PROLOG-NEXT: [[TMP1:%.*]] = icmp ult i32 [[N]], 1
|
|
; PROLOG-NEXT: br i1 [[TMP1]], label [[WHILE_END:%.*]], label [[ENTRY_NEW:%.*]]
|
|
; PROLOG: entry.new:
|
|
; PROLOG-NEXT: br label [[WHILE_BODY:%.*]]
|
|
; PROLOG: while.body:
|
|
; PROLOG-NEXT: [[I:%.*]] = phi i32 [ [[I_UNR]], [[ENTRY_NEW]] ], [ [[INC_1:%.*]], [[WHILE_BODY]] ]
|
|
; PROLOG-NEXT: [[INC:%.*]] = add i32 [[I]], 1
|
|
; PROLOG-NEXT: [[CMP_1:%.*]] = icmp eq i32 [[INC]], [[N]]
|
|
; PROLOG-NEXT: [[INC_1]] = add i32 [[I]], 2
|
|
; PROLOG-NEXT: br i1 [[CMP_1]], label [[WHILE_END_UNR_LCSSA:%.*]], label [[WHILE_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
|
|
; PROLOG: while.end.unr-lcssa:
|
|
; PROLOG-NEXT: [[I_LCSSA_PH:%.*]] = phi i32 [ [[INC]], [[WHILE_BODY]] ]
|
|
; PROLOG-NEXT: br label [[WHILE_END]]
|
|
; PROLOG: while.end:
|
|
; PROLOG-NEXT: [[I_LCSSA:%.*]] = phi i32 [ [[I_LCSSA_UNR]], [[WHILE_BODY_PROL_LOOPEXIT]] ], [ [[I_LCSSA_PH]], [[WHILE_END_UNR_LCSSA]] ]
|
|
; PROLOG-NEXT: ret i32 [[I_LCSSA]]
|
|
;
|
|
entry:
|
|
br label %while.body
|
|
|
|
while.body: ; preds = %while.body, %entry
|
|
%i = phi i32 [ 0, %entry ], [ %inc, %while.body ]
|
|
%cmp = icmp eq i32 %i, %N
|
|
%inc = add i32 %i, 1
|
|
br i1 %cmp, label %while.end, label %while.body
|
|
|
|
while.end: ; preds = %while.body
|
|
ret i32 %i
|
|
}
|