1 ; In this test we check how heuristics for complete unrolling work. We have 2 ; three knobs: 3 ; 1) -unroll-threshold 4 ; 3) -unroll-percent-dynamic-cost-saved-threshold and 5 ; 2) -unroll-dynamic-cost-savings-discount 6 ; 7 ; They control loop-unrolling according to the following rules: 8 ; * If size of unrolled loop exceeds the absoulte threshold, we don't unroll 9 ; this loop under any circumstances. 10 ; * If size of unrolled loop is below the '-unroll-threshold', then we'll 11 ; consider this loop as a very small one, and completely unroll it. 12 ; * If a loop size is between these two tresholds, we only do complete unroll 13 ; it if estimated number of potentially optimized instructions is high (we 14 ; specify the minimal percent of such instructions). 15 16 ; In this particular test-case, complete unrolling will allow later 17 ; optimizations to remove ~55% of the instructions, the loop body size is 9, 18 ; and unrolled size is 65. 19 20 ; RUN: opt < %s -S -loop-unroll -unroll-max-iteration-count-to-analyze=1000 -unroll-threshold=10 -unroll-max-percent-threshold-boost=100 | FileCheck %s -check-prefix=TEST1 21 ; RUN: opt < %s -S -loop-unroll -unroll-max-iteration-count-to-analyze=1000 -unroll-threshold=20 -unroll-max-percent-threshold-boost=200 | FileCheck %s -check-prefix=TEST2 22 ; RUN: opt < %s -S -loop-unroll -unroll-max-iteration-count-to-analyze=1000 -unroll-threshold=20 -unroll-max-percent-threshold-boost=100 | FileCheck %s -check-prefix=TEST3 23 24 ; RUN: opt < %s -S -passes='require<opt-remark-emit>,loop(unroll-full)' -unroll-max-iteration-count-to-analyze=1000 -unroll-threshold=10 -unroll-max-percent-threshold-boost=100 | FileCheck %s -check-prefix=TEST1 25 ; RUN: opt < %s -S -passes='require<opt-remark-emit>,loop(unroll-full)' -unroll-max-iteration-count-to-analyze=1000 -unroll-threshold=20 -unroll-max-percent-threshold-boost=200 | FileCheck %s -check-prefix=TEST2 26 ; RUN: opt < %s -S -passes='require<opt-remark-emit>,loop(unroll-full)' -unroll-max-iteration-count-to-analyze=1000 -unroll-threshold=20 -unroll-max-percent-threshold-boost=100 | FileCheck %s -check-prefix=TEST3 27 28 ; Check that these work when the unroller has partial unrolling enabled too. 29 ; RUN: opt < %s -S -passes='require<opt-remark-emit>,unroll' -unroll-max-iteration-count-to-analyze=1000 -unroll-threshold=10 -unroll-max-percent-threshold-boost=100 | FileCheck %s -check-prefix=TEST1 30 ; RUN: opt < %s -S -passes='require<opt-remark-emit>,unroll' -unroll-max-iteration-count-to-analyze=1000 -unroll-threshold=20 -unroll-max-percent-threshold-boost=200 | FileCheck %s -check-prefix=TEST2 31 ; RUN: opt < %s -S -passes='require<opt-remark-emit>,unroll' -unroll-max-iteration-count-to-analyze=1000 -unroll-threshold=20 -unroll-max-percent-threshold-boost=100 | FileCheck %s -check-prefix=TEST3 32 33 ; If the absolute threshold is too low, we should not unroll: 34 ; TEST1: %array_const_idx = getelementptr inbounds [9 x i32], [9 x i32]* @known_constant, i64 0, i64 %iv 35 36 ; Otherwise, we should: 37 ; TEST2-NOT: %array_const_idx = getelementptr inbounds [9 x i32], [9 x i32]* @known_constant, i64 0, i64 %iv 38 39 ; If we do not boost threshold, the unroll will not happen: 40 ; TEST3: %array_const_idx = getelementptr inbounds [9 x i32], [9 x i32]* @known_constant, i64 0, i64 %iv 41 42 ; And check that we don't crash when we're not allowed to do any analysis. 43 ; RUN: opt < %s -loop-unroll -unroll-max-iteration-count-to-analyze=0 -disable-output 44 ; RUN: opt < %s -passes='require<opt-remark-emit>,loop(unroll-full)' -unroll-max-iteration-count-to-analyze=0 -disable-output 45 target datalayout = "e-m:o-i64:64-f80:128-n8:16:32:64-S128" 46 47 @known_constant = internal unnamed_addr constant [9 x i32] [i32 0, i32 -1, i32 0, i32 -1, i32 5, i32 -1, i32 0, i32 -1, i32 0], align 16 48 49 define i32 @foo(i32* noalias nocapture readonly %src) { 50 entry: 51 br label %loop 52 53 loop: ; preds = %loop, %entry 54 %iv = phi i64 [ 0, %entry ], [ %inc, %loop ] 55 %r = phi i32 [ 0, %entry ], [ %add, %loop ] 56 %arrayidx = getelementptr inbounds i32, i32* %src, i64 %iv 57 %src_element = load i32, i32* %arrayidx, align 4 58 %array_const_idx = getelementptr inbounds [9 x i32], [9 x i32]* @known_constant, i64 0, i64 %iv 59 %const_array_element = load i32, i32* %array_const_idx, align 4 60 %mul = mul nsw i32 %src_element, %const_array_element 61 %add = add nsw i32 %mul, %r 62 %inc = add nuw nsw i64 %iv, 1 63 %exitcond86.i = icmp eq i64 %inc, 9 64 br i1 %exitcond86.i, label %loop.end, label %loop 65 66 loop.end: ; preds = %loop 67 %r.lcssa = phi i32 [ %r, %loop ] 68 ret i32 %r.lcssa 69 } 70
