diff options
author | Stephen Hines <srhines@google.com> | 2014-05-29 02:49:00 -0700 |
---|---|---|
committer | Stephen Hines <srhines@google.com> | 2014-05-29 02:49:00 -0700 |
commit | dce4a407a24b04eebc6a376f8e62b41aaa7b071f (patch) | |
tree | dcebc53f2b182f145a2e659393bf9a0472cedf23 /test/Analysis/BlockFrequencyInfo | |
parent | 220b921aed042f9e520c26cffd8282a94c66c3d5 (diff) | |
download | external_llvm-dce4a407a24b04eebc6a376f8e62b41aaa7b071f.zip external_llvm-dce4a407a24b04eebc6a376f8e62b41aaa7b071f.tar.gz external_llvm-dce4a407a24b04eebc6a376f8e62b41aaa7b071f.tar.bz2 |
Update LLVM for 3.5 rebase (r209712).
Change-Id: I149556c940fb7dc92d075273c87ff584f400941f
Diffstat (limited to 'test/Analysis/BlockFrequencyInfo')
-rw-r--r-- | test/Analysis/BlockFrequencyInfo/bad_input.ll | 50 | ||||
-rw-r--r-- | test/Analysis/BlockFrequencyInfo/basic.ll | 55 | ||||
-rw-r--r-- | test/Analysis/BlockFrequencyInfo/double_backedge.ll | 27 | ||||
-rw-r--r-- | test/Analysis/BlockFrequencyInfo/double_exit.ll | 165 | ||||
-rw-r--r-- | test/Analysis/BlockFrequencyInfo/irreducible.ll | 421 | ||||
-rw-r--r-- | test/Analysis/BlockFrequencyInfo/loop_with_branch.ll | 44 | ||||
-rw-r--r-- | test/Analysis/BlockFrequencyInfo/nested_loop_with_branches.ll | 59 |
7 files changed, 797 insertions, 24 deletions
diff --git a/test/Analysis/BlockFrequencyInfo/bad_input.ll b/test/Analysis/BlockFrequencyInfo/bad_input.ll new file mode 100644 index 0000000..bcdc1e6 --- /dev/null +++ b/test/Analysis/BlockFrequencyInfo/bad_input.ll @@ -0,0 +1,50 @@ +; RUN: opt < %s -analyze -block-freq | FileCheck %s + +declare void @g(i32 %x) + +; CHECK-LABEL: Printing analysis {{.*}} for function 'branch_weight_0': +; CHECK-NEXT: block-frequency-info: branch_weight_0 +define void @branch_weight_0(i32 %a) { +; CHECK-NEXT: entry: float = 1.0, int = [[ENTRY:[0-9]+]] +entry: + br label %for.body + +; Check that we get 1,4 instead of 0,3. +; CHECK-NEXT: for.body: float = 4.0, +for.body: + %i = phi i32 [ 0, %entry ], [ %inc, %for.body ] + call void @g(i32 %i) + %inc = add i32 %i, 1 + %cmp = icmp ugt i32 %inc, %a + br i1 %cmp, label %for.end, label %for.body, !prof !0 + +; CHECK-NEXT: for.end: float = 1.0, int = [[ENTRY]] +for.end: + ret void +} + +!0 = metadata !{metadata !"branch_weights", i32 0, i32 3} + +; CHECK-LABEL: Printing analysis {{.*}} for function 'infinite_loop' +; CHECK-NEXT: block-frequency-info: infinite_loop +define void @infinite_loop(i1 %x) { +; CHECK-NEXT: entry: float = 1.0, int = [[ENTRY:[0-9]+]] +entry: + br i1 %x, label %for.body, label %for.end, !prof !1 + +; Check that the loop scale maxes out at 4096, giving 2048 here. +; CHECK-NEXT: for.body: float = 2048.0, +for.body: + %i = phi i32 [ 0, %entry ], [ %inc, %for.body ] + call void @g(i32 %i) + %inc = add i32 %i, 1 + br label %for.body + +; Check that the exit weight is half of entry, since half is lost in the +; infinite loop above. +; CHECK-NEXT: for.end: float = 0.5, +for.end: + ret void +} + +!1 = metadata !{metadata !"branch_weights", i32 1, i32 1} diff --git a/test/Analysis/BlockFrequencyInfo/basic.ll b/test/Analysis/BlockFrequencyInfo/basic.ll index ce29fb5..006e6ab 100644 --- a/test/Analysis/BlockFrequencyInfo/basic.ll +++ b/test/Analysis/BlockFrequencyInfo/basic.ll @@ -1,13 +1,14 @@ ; RUN: opt < %s -analyze -block-freq | FileCheck %s define i32 @test1(i32 %i, i32* %a) { -; CHECK: Printing analysis {{.*}} for function 'test1' -; CHECK: entry = 1.0 +; CHECK-LABEL: Printing analysis {{.*}} for function 'test1': +; CHECK-NEXT: block-frequency-info: test1 +; CHECK-NEXT: entry: float = 1.0, int = [[ENTRY:[0-9]+]] entry: br label %body ; Loop backedges are weighted and thus their bodies have a greater frequency. -; CHECK: body = 32.0 +; CHECK-NEXT: body: float = 32.0, body: %iv = phi i32 [ 0, %entry ], [ %next, %body ] %base = phi i32 [ 0, %entry ], [ %sum, %body ] @@ -18,29 +19,29 @@ body: %exitcond = icmp eq i32 %next, %i br i1 %exitcond, label %exit, label %body -; CHECK: exit = 1.0 +; CHECK-NEXT: exit: float = 1.0, int = [[ENTRY]] exit: ret i32 %sum } define i32 @test2(i32 %i, i32 %a, i32 %b) { -; CHECK: Printing analysis {{.*}} for function 'test2' -; CHECK: entry = 1.0 +; CHECK-LABEL: Printing analysis {{.*}} for function 'test2': +; CHECK-NEXT: block-frequency-info: test2 +; CHECK-NEXT: entry: float = 1.0, int = [[ENTRY:[0-9]+]] entry: %cond = icmp ult i32 %i, 42 br i1 %cond, label %then, label %else, !prof !0 ; The 'then' branch is predicted more likely via branch weight metadata. -; CHECK: then = 0.94116 +; CHECK-NEXT: then: float = 0.9411{{[0-9]*}}, then: br label %exit -; CHECK: else = 0.05877 +; CHECK-NEXT: else: float = 0.05882{{[0-9]*}}, else: br label %exit -; FIXME: It may be a bug that we don't sum back to 1.0. -; CHECK: exit = 0.99993 +; CHECK-NEXT: exit: float = 1.0, int = [[ENTRY]] exit: %result = phi i32 [ %a, %then ], [ %b, %else ] ret i32 %result @@ -49,37 +50,37 @@ exit: !0 = metadata !{metadata !"branch_weights", i32 64, i32 4} define i32 @test3(i32 %i, i32 %a, i32 %b, i32 %c, i32 %d, i32 %e) { -; CHECK: Printing analysis {{.*}} for function 'test3' -; CHECK: entry = 1.0 +; CHECK-LABEL: Printing analysis {{.*}} for function 'test3': +; CHECK-NEXT: block-frequency-info: test3 +; CHECK-NEXT: entry: float = 1.0, int = [[ENTRY:[0-9]+]] entry: switch i32 %i, label %case_a [ i32 1, label %case_b i32 2, label %case_c i32 3, label %case_d i32 4, label %case_e ], !prof !1 -; CHECK: case_a = 0.04998 +; CHECK-NEXT: case_a: float = 0.05, case_a: br label %exit -; CHECK: case_b = 0.04998 +; CHECK-NEXT: case_b: float = 0.05, case_b: br label %exit ; The 'case_c' branch is predicted more likely via branch weight metadata. -; CHECK: case_c = 0.79998 +; CHECK-NEXT: case_c: float = 0.8, case_c: br label %exit -; CHECK: case_d = 0.04998 +; CHECK-NEXT: case_d: float = 0.05, case_d: br label %exit -; CHECK: case_e = 0.04998 +; CHECK-NEXT: case_e: float = 0.05, case_e: br label %exit -; FIXME: It may be a bug that we don't sum back to 1.0. -; CHECK: exit = 0.99993 +; CHECK-NEXT: exit: float = 1.0, int = [[ENTRY]] exit: %result = phi i32 [ %a, %case_a ], [ %b, %case_b ], @@ -91,44 +92,50 @@ exit: !1 = metadata !{metadata !"branch_weights", i32 4, i32 4, i32 64, i32 4, i32 4} -; CHECK: Printing analysis {{.*}} for function 'nested_loops' -; CHECK: entry = 1.0 -; This test doesn't seem to be assigning sensible frequencies to nested loops. define void @nested_loops(i32 %a) { +; CHECK-LABEL: Printing analysis {{.*}} for function 'nested_loops': +; CHECK-NEXT: block-frequency-info: nested_loops +; CHECK-NEXT: entry: float = 1.0, int = [[ENTRY:[0-9]+]] entry: br label %for.cond1.preheader +; CHECK-NEXT: for.cond1.preheader: float = 4001.0, for.cond1.preheader: %x.024 = phi i32 [ 0, %entry ], [ %inc12, %for.inc11 ] br label %for.cond4.preheader +; CHECK-NEXT: for.cond4.preheader: float = 16008001.0, for.cond4.preheader: %y.023 = phi i32 [ 0, %for.cond1.preheader ], [ %inc9, %for.inc8 ] %add = add i32 %y.023, %x.024 br label %for.body6 +; CHECK-NEXT: for.body6: float = 64048012001.0, for.body6: %z.022 = phi i32 [ 0, %for.cond4.preheader ], [ %inc, %for.body6 ] %add7 = add i32 %add, %z.022 - tail call void @g(i32 %add7) #2 + tail call void @g(i32 %add7) %inc = add i32 %z.022, 1 %cmp5 = icmp ugt i32 %inc, %a br i1 %cmp5, label %for.inc8, label %for.body6, !prof !2 +; CHECK-NEXT: for.inc8: float = 16008001.0, for.inc8: %inc9 = add i32 %y.023, 1 %cmp2 = icmp ugt i32 %inc9, %a br i1 %cmp2, label %for.inc11, label %for.cond4.preheader, !prof !2 +; CHECK-NEXT: for.inc11: float = 4001.0, for.inc11: %inc12 = add i32 %x.024, 1 %cmp = icmp ugt i32 %inc12, %a br i1 %cmp, label %for.end13, label %for.cond1.preheader, !prof !2 +; CHECK-NEXT: for.end13: float = 1.0, int = [[ENTRY]] for.end13: ret void } -declare void @g(i32) #1 +declare void @g(i32) !2 = metadata !{metadata !"branch_weights", i32 1, i32 4000} diff --git a/test/Analysis/BlockFrequencyInfo/double_backedge.ll b/test/Analysis/BlockFrequencyInfo/double_backedge.ll new file mode 100644 index 0000000..df8217c --- /dev/null +++ b/test/Analysis/BlockFrequencyInfo/double_backedge.ll @@ -0,0 +1,27 @@ +; RUN: opt < %s -analyze -block-freq | FileCheck %s + +define void @double_backedge(i1 %x) { +; CHECK-LABEL: Printing analysis {{.*}} for function 'double_backedge': +; CHECK-NEXT: block-frequency-info: double_backedge +entry: +; CHECK-NEXT: entry: float = 1.0, int = [[ENTRY:[0-9]+]] + br label %loop + +loop: +; CHECK-NEXT: loop: float = 10.0, + br i1 %x, label %exit, label %loop.1, !prof !0 + +loop.1: +; CHECK-NEXT: loop.1: float = 9.0, + br i1 %x, label %loop, label %loop.2, !prof !1 + +loop.2: +; CHECK-NEXT: loop.2: float = 5.0, + br label %loop + +exit: +; CHECK-NEXT: exit: float = 1.0, int = [[ENTRY]] + ret void +} +!0 = metadata !{metadata !"branch_weights", i32 1, i32 9} +!1 = metadata !{metadata !"branch_weights", i32 4, i32 5} diff --git a/test/Analysis/BlockFrequencyInfo/double_exit.ll b/test/Analysis/BlockFrequencyInfo/double_exit.ll new file mode 100644 index 0000000..75f664d --- /dev/null +++ b/test/Analysis/BlockFrequencyInfo/double_exit.ll @@ -0,0 +1,165 @@ +; RUN: opt < %s -analyze -block-freq | FileCheck %s + +; CHECK-LABEL: Printing analysis {{.*}} for function 'double_exit': +; CHECK-NEXT: block-frequency-info: double_exit +define i32 @double_exit(i32 %N) { +; Mass = 1 +; Frequency = 1 +; CHECK-NEXT: entry: float = 1.0, int = [[ENTRY:[0-9]+]] +entry: + br label %outer + +; Mass = 1 +; Backedge mass = 1/3, exit mass = 2/3 +; Loop scale = 3/2 +; Pseudo-edges = exit +; Pseudo-mass = 1 +; Frequency = 1*3/2*1 = 3/2 +; CHECK-NEXT: outer: float = 1.5, +outer: + %I.0 = phi i32 [ 0, %entry ], [ %inc6, %outer.inc ] + %Return.0 = phi i32 [ 0, %entry ], [ %Return.1, %outer.inc ] + %cmp = icmp slt i32 %I.0, %N + br i1 %cmp, label %inner, label %exit, !prof !2 ; 2:1 + +; Mass = 1 +; Backedge mass = 3/5, exit mass = 2/5 +; Loop scale = 5/2 +; Pseudo-edges = outer.inc @ 1/5, exit @ 1/5 +; Pseudo-mass = 2/3 +; Frequency = 3/2*1*5/2*2/3 = 5/2 +; CHECK-NEXT: inner: float = 2.5, +inner: + %Return.1 = phi i32 [ %Return.0, %outer ], [ %call4, %inner.inc ] + %J.0 = phi i32 [ %I.0, %outer ], [ %inc, %inner.inc ] + %cmp2 = icmp slt i32 %J.0, %N + br i1 %cmp2, label %inner.body, label %outer.inc, !prof !1 ; 4:1 + +; Mass = 4/5 +; Frequency = 5/2*4/5 = 2 +; CHECK-NEXT: inner.body: float = 2.0, +inner.body: + %call = call i32 @c2(i32 %I.0, i32 %J.0) + %tobool = icmp ne i32 %call, 0 + br i1 %tobool, label %exit, label %inner.inc, !prof !0 ; 3:1 + +; Mass = 3/5 +; Frequency = 5/2*3/5 = 3/2 +; CHECK-NEXT: inner.inc: float = 1.5, +inner.inc: + %call4 = call i32 @logic2(i32 %Return.1, i32 %I.0, i32 %J.0) + %inc = add nsw i32 %J.0, 1 + br label %inner + +; Mass = 1/3 +; Frequency = 3/2*1/3 = 1/2 +; CHECK-NEXT: outer.inc: float = 0.5, +outer.inc: + %inc6 = add nsw i32 %I.0, 1 + br label %outer + +; Mass = 1 +; Frequency = 1 +; CHECK-NEXT: exit: float = 1.0, int = [[ENTRY]] +exit: + %Return.2 = phi i32 [ %Return.1, %inner.body ], [ %Return.0, %outer ] + ret i32 %Return.2 +} + +!0 = metadata !{metadata !"branch_weights", i32 1, i32 3} +!1 = metadata !{metadata !"branch_weights", i32 4, i32 1} +!2 = metadata !{metadata !"branch_weights", i32 2, i32 1} + +declare i32 @c2(i32, i32) +declare i32 @logic2(i32, i32, i32) + +; CHECK-LABEL: Printing analysis {{.*}} for function 'double_exit_in_loop': +; CHECK-NEXT: block-frequency-info: double_exit_in_loop +define i32 @double_exit_in_loop(i32 %N) { +; Mass = 1 +; Frequency = 1 +; CHECK-NEXT: entry: float = 1.0, int = [[ENTRY:[0-9]+]] +entry: + br label %outer + +; Mass = 1 +; Backedge mass = 1/2, exit mass = 1/2 +; Loop scale = 2 +; Pseudo-edges = exit +; Pseudo-mass = 1 +; Frequency = 1*2*1 = 2 +; CHECK-NEXT: outer: float = 2.0, +outer: + %I.0 = phi i32 [ 0, %entry ], [ %inc12, %outer.inc ] + %Return.0 = phi i32 [ 0, %entry ], [ %Return.3, %outer.inc ] + %cmp = icmp slt i32 %I.0, %N + br i1 %cmp, label %middle, label %exit, !prof !3 ; 1:1 + +; Mass = 1 +; Backedge mass = 1/3, exit mass = 2/3 +; Loop scale = 3/2 +; Pseudo-edges = outer.inc +; Pseudo-mass = 1/2 +; Frequency = 2*1*3/2*1/2 = 3/2 +; CHECK-NEXT: middle: float = 1.5, +middle: + %J.0 = phi i32 [ %I.0, %outer ], [ %inc9, %middle.inc ] + %Return.1 = phi i32 [ %Return.0, %outer ], [ %Return.2, %middle.inc ] + %cmp2 = icmp slt i32 %J.0, %N + br i1 %cmp2, label %inner, label %outer.inc, !prof !2 ; 2:1 + +; Mass = 1 +; Backedge mass = 3/5, exit mass = 2/5 +; Loop scale = 5/2 +; Pseudo-edges = middle.inc @ 1/5, outer.inc @ 1/5 +; Pseudo-mass = 2/3 +; Frequency = 3/2*1*5/2*2/3 = 5/2 +; CHECK-NEXT: inner: float = 2.5, +inner: + %Return.2 = phi i32 [ %Return.1, %middle ], [ %call7, %inner.inc ] + %K.0 = phi i32 [ %J.0, %middle ], [ %inc, %inner.inc ] + %cmp5 = icmp slt i32 %K.0, %N + br i1 %cmp5, label %inner.body, label %middle.inc, !prof !1 ; 4:1 + +; Mass = 4/5 +; Frequency = 5/2*4/5 = 2 +; CHECK-NEXT: inner.body: float = 2.0, +inner.body: + %call = call i32 @c3(i32 %I.0, i32 %J.0, i32 %K.0) + %tobool = icmp ne i32 %call, 0 + br i1 %tobool, label %outer.inc, label %inner.inc, !prof !0 ; 3:1 + +; Mass = 3/5 +; Frequency = 5/2*3/5 = 3/2 +; CHECK-NEXT: inner.inc: float = 1.5, +inner.inc: + %call7 = call i32 @logic3(i32 %Return.2, i32 %I.0, i32 %J.0, i32 %K.0) + %inc = add nsw i32 %K.0, 1 + br label %inner + +; Mass = 1/3 +; Frequency = 3/2*1/3 = 1/2 +; CHECK-NEXT: middle.inc: float = 0.5, +middle.inc: + %inc9 = add nsw i32 %J.0, 1 + br label %middle + +; Mass = 1/2 +; Frequency = 2*1/2 = 1 +; CHECK-NEXT: outer.inc: float = 1.0, +outer.inc: + %Return.3 = phi i32 [ %Return.2, %inner.body ], [ %Return.1, %middle ] + %inc12 = add nsw i32 %I.0, 1 + br label %outer + +; Mass = 1 +; Frequency = 1 +; CHECK-NEXT: exit: float = 1.0, int = [[ENTRY]] +exit: + ret i32 %Return.0 +} + +!3 = metadata !{metadata !"branch_weights", i32 1, i32 1} + +declare i32 @c3(i32, i32, i32) +declare i32 @logic3(i32, i32, i32, i32) diff --git a/test/Analysis/BlockFrequencyInfo/irreducible.ll b/test/Analysis/BlockFrequencyInfo/irreducible.ll new file mode 100644 index 0000000..af4ad15 --- /dev/null +++ b/test/Analysis/BlockFrequencyInfo/irreducible.ll @@ -0,0 +1,421 @@ +; RUN: opt < %s -analyze -block-freq | FileCheck %s + +; A loop with multiple exits isn't irreducible. It should be handled +; correctly. +; +; CHECK-LABEL: Printing analysis {{.*}} for function 'multiexit': +; CHECK-NEXT: block-frequency-info: multiexit +define void @multiexit(i1 %x) { +; CHECK-NEXT: entry: float = 1.0, int = [[ENTRY:[0-9]+]] +entry: + br label %loop.1 + +; CHECK-NEXT: loop.1: float = 2.0, +loop.1: + br i1 %x, label %exit.1, label %loop.2, !prof !0 + +; CHECK-NEXT: loop.2: float = 1.75, +loop.2: + br i1 %x, label %exit.2, label %loop.1, !prof !1 + +; CHECK-NEXT: exit.1: float = 0.25, +exit.1: + br label %return + +; CHECK-NEXT: exit.2: float = 0.75, +exit.2: + br label %return + +; CHECK-NEXT: return: float = 1.0, int = [[ENTRY]] +return: + ret void +} + +!0 = metadata !{metadata !"branch_weights", i32 1, i32 7} +!1 = metadata !{metadata !"branch_weights", i32 3, i32 4} + +; Irreducible control flow +; ======================== +; +; LoopInfo defines a loop as a non-trivial SCC dominated by a single block, +; called the header. A given loop, L, can have sub-loops, which are loops +; within the subgraph of L that excludes the header. +; +; In addition to loops, -block-freq has limited support for irreducible SCCs, +; which are SCCs with multiple entry blocks. Irreducible SCCs are discovered +; on they fly, and modelled as loops with multiple headers. +; +; The headers of irreducible sub-SCCs consist of its entry blocks and all nodes +; that are targets of a backedge within it (excluding backedges within true +; sub-loops). +; +; -block-freq is currently designed to act like a block is inserted that +; intercepts all the edges to the headers. All backedges and entries point to +; this block. Its successors are the headers, which split the frequency +; evenly. +; +; There are a number of testcases below. Only the first two have detailed +; explanations. +; +; Testcase #1 +; =========== +; +; In this case c1 and c2 should have frequencies of 15/7 and 13/7, +; respectively. To calculate this, consider assigning 1.0 to entry, and +; distributing frequency iteratively (to infinity). At the first iteration, +; entry gives 3/4 to c1 and 1/4 to c2. At every step after, c1 and c2 give 3/4 +; of what they have to each other. Somehow, all of it comes out to exit. +; +; c1 = 3/4 + 1/4*3/4 + 3/4*3^2/4^2 + 1/4*3^3/4^3 + 3/4*3^3/4^3 + ... +; c2 = 1/4 + 3/4*3/4 + 1/4*3^2/4^2 + 3/4*3^3/4^3 + 1/4*3^3/4^3 + ... +; +; Simplify by splitting up the odd and even terms of the series and taking out +; factors so that the infite series matches: +; +; c1 = 3/4 *(9^0/16^0 + 9^1/16^1 + 9^2/16^2 + ...) +; + 3/16*(9^0/16^0 + 9^1/16^1 + 9^2/16^2 + ...) +; c2 = 1/4 *(9^0/16^0 + 9^1/16^1 + 9^2/16^2 + ...) +; + 9/16*(9^0/16^0 + 9^1/16^1 + 9^2/16^2 + ...) +; +; c1 = 15/16*(9^0/16^0 + 9^1/16^1 + 9^2/16^2 + ...) +; c2 = 13/16*(9^0/16^0 + 9^1/16^1 + 9^2/16^2 + ...) +; +; Since this geometric series sums to 16/7: +; +; c1 = 15/7 +; c2 = 13/7 +; +; If we treat c1 and c2 as members of the same loop, the exit frequency of the +; loop as a whole is 1/4, so the loop scale should be 4. Summing c1 and c2 +; gives 28/7, or 4.0, which is nice confirmation of the math above. +; +; -block-freq currently treats the two nodes as equals. +define void @multientry(i1 %x) { +; CHECK-LABEL: Printing analysis {{.*}} for function 'multientry': +; CHECK-NEXT: block-frequency-info: multientry +entry: +; CHECK-NEXT: entry: float = 1.0, int = [[ENTRY:[0-9]+]] + br i1 %x, label %c1, label %c2, !prof !2 + +c1: +; CHECK-NEXT: c1: float = 2.0, +; The "correct" answer is: float = 2.142857{{[0-9]*}}, + br i1 %x, label %c2, label %exit, !prof !2 + +c2: +; CHECK-NEXT: c2: float = 2.0, +; The "correct" answer is: float = 1.857142{{[0-9]*}}, + br i1 %x, label %c1, label %exit, !prof !2 + +exit: +; CHECK-NEXT: exit: float = 1.0, int = [[ENTRY]] + ret void +} + +!2 = metadata !{metadata !"branch_weights", i32 3, i32 1} + +; Testcase #2 +; =========== +; +; In this case c1 and c2 should be treated as equals in a single loop. The +; exit frequency is 1/3, so the scaling factor for the loop should be 3.0. The +; loop is entered 2/3 of the time, and c1 and c2 split the total loop frequency +; evenly (1/2), so they should each have frequencies of 1.0 (3.0*2/3*1/2). +; Another way of computing this result is by assigning 1.0 to entry and showing +; that c1 and c2 should accumulate frequencies of: +; +; 1/3 + 2/9 + 4/27 + 8/81 + ... +; 2^0/3^1 + 2^1/3^2 + 2^2/3^3 + 2^3/3^4 + ... +; +; At the first step, c1 and c2 each get 1/3 of the entry. At each subsequent +; step, c1 and c2 each get 1/3 of what's left in c1 and c2 combined. This +; infinite series sums to 1. +; +; Since the currently algorithm *always* assumes entry blocks are equal, +; -block-freq gets the right answers here. +define void @crossloops(i2 %x) { +; CHECK-LABEL: Printing analysis {{.*}} for function 'crossloops': +; CHECK-NEXT: block-frequency-info: crossloops +entry: +; CHECK-NEXT: entry: float = 1.0, int = [[ENTRY:[0-9]+]] + switch i2 %x, label %exit [ i2 1, label %c1 + i2 2, label %c2 ], !prof !3 + +c1: +; CHECK-NEXT: c1: float = 1.0, + switch i2 %x, label %exit [ i2 1, label %c1 + i2 2, label %c2 ], !prof !3 + +c2: +; CHECK-NEXT: c2: float = 1.0, + switch i2 %x, label %exit [ i2 1, label %c1 + i2 2, label %c2 ], !prof !3 + +exit: +; CHECK-NEXT: exit: float = 1.0, int = [[ENTRY]] + ret void +} + +!3 = metadata !{metadata !"branch_weights", i32 2, i32 2, i32 2} + +; A true loop with irreducible control flow inside. +define void @loop_around_irreducible(i1 %x) { +; CHECK-LABEL: Printing analysis {{.*}} for function 'loop_around_irreducible': +; CHECK-NEXT: block-frequency-info: loop_around_irreducible +entry: +; CHECK-NEXT: entry: float = 1.0, int = [[ENTRY:[0-9]+]] + br label %loop + +loop: +; CHECK-NEXT: loop: float = 4.0, int = [[HEAD:[0-9]+]] + br i1 %x, label %left, label %right, !prof !4 + +left: +; CHECK-NEXT: left: float = 8.0, + br i1 %x, label %right, label %loop.end, !prof !5 + +right: +; CHECK-NEXT: right: float = 8.0, + br i1 %x, label %left, label %loop.end, !prof !5 + +loop.end: +; CHECK-NEXT: loop.end: float = 4.0, int = [[HEAD]] + br i1 %x, label %loop, label %exit, !prof !5 + +exit: +; CHECK-NEXT: exit: float = 1.0, int = [[ENTRY]] + ret void +} +!4 = metadata !{metadata !"branch_weights", i32 1, i32 1} +!5 = metadata !{metadata !"branch_weights", i32 3, i32 1} + +; Two unrelated irreducible SCCs. +define void @two_sccs(i1 %x) { +; CHECK-LABEL: Printing analysis {{.*}} for function 'two_sccs': +; CHECK-NEXT: block-frequency-info: two_sccs +entry: +; CHECK-NEXT: entry: float = 1.0, int = [[ENTRY:[0-9]+]] + br i1 %x, label %a, label %b, !prof !6 + +a: +; CHECK-NEXT: a: float = 0.75, + br i1 %x, label %a.left, label %a.right, !prof !7 + +a.left: +; CHECK-NEXT: a.left: float = 1.5, + br i1 %x, label %a.right, label %exit, !prof !6 + +a.right: +; CHECK-NEXT: a.right: float = 1.5, + br i1 %x, label %a.left, label %exit, !prof !6 + +b: +; CHECK-NEXT: b: float = 0.25, + br i1 %x, label %b.left, label %b.right, !prof !7 + +b.left: +; CHECK-NEXT: b.left: float = 0.625, + br i1 %x, label %b.right, label %exit, !prof !8 + +b.right: +; CHECK-NEXT: b.right: float = 0.625, + br i1 %x, label %b.left, label %exit, !prof !8 + +exit: +; CHECK-NEXT: exit: float = 1.0, int = [[ENTRY]] + ret void +} +!6 = metadata !{metadata !"branch_weights", i32 3, i32 1} +!7 = metadata !{metadata !"branch_weights", i32 1, i32 1} +!8 = metadata !{metadata !"branch_weights", i32 4, i32 1} + +; A true loop inside irreducible control flow. +define void @loop_inside_irreducible(i1 %x) { +; CHECK-LABEL: Printing analysis {{.*}} for function 'loop_inside_irreducible': +; CHECK-NEXT: block-frequency-info: loop_inside_irreducible +entry: +; CHECK-NEXT: entry: float = 1.0, int = [[ENTRY:[0-9]+]] + br i1 %x, label %left, label %right, !prof !9 + +left: +; CHECK-NEXT: left: float = 2.0, + br i1 %x, label %right, label %exit, !prof !10 + +right: +; CHECK-NEXT: right: float = 2.0, int = [[RIGHT:[0-9]+]] + br label %loop + +loop: +; CHECK-NEXT: loop: float = 6.0, + br i1 %x, label %loop, label %right.end, !prof !11 + +right.end: +; CHECK-NEXT: right.end: float = 2.0, int = [[RIGHT]] + br i1 %x, label %left, label %exit, !prof !10 + +exit: +; CHECK-NEXT: exit: float = 1.0, int = [[ENTRY]] + ret void +} +!9 = metadata !{metadata !"branch_weights", i32 1, i32 1} +!10 = metadata !{metadata !"branch_weights", i32 3, i32 1} +!11 = metadata !{metadata !"branch_weights", i32 2, i32 1} + +; Irreducible control flow in a branch that's in a true loop. +define void @loop_around_branch_with_irreducible(i1 %x) { +; CHECK-LABEL: Printing analysis {{.*}} for function 'loop_around_branch_with_irreducible': +; CHECK-NEXT: block-frequency-info: loop_around_branch_with_irreducible +entry: +; CHECK-NEXT: entry: float = 1.0, int = [[ENTRY:[0-9]+]] + br label %loop + +loop: +; CHECK-NEXT: loop: float = 2.0, int = [[LOOP:[0-9]+]] + br i1 %x, label %normal, label %irreducible.entry, !prof !12 + +normal: +; CHECK-NEXT: normal: float = 1.5, + br label %loop.end + +irreducible.entry: +; CHECK-NEXT: irreducible.entry: float = 0.5, int = [[IRREDUCIBLE:[0-9]+]] + br i1 %x, label %left, label %right, !prof !13 + +left: +; CHECK-NEXT: left: float = 1.0, + br i1 %x, label %right, label %irreducible.exit, !prof !12 + +right: +; CHECK-NEXT: right: float = 1.0, + br i1 %x, label %left, label %irreducible.exit, !prof !12 + +irreducible.exit: +; CHECK-NEXT: irreducible.exit: float = 0.5, int = [[IRREDUCIBLE]] + br label %loop.end + +loop.end: +; CHECK-NEXT: loop.end: float = 2.0, int = [[LOOP]] + br i1 %x, label %loop, label %exit, !prof !13 + +exit: +; CHECK-NEXT: exit: float = 1.0, int = [[ENTRY]] + ret void +} +!12 = metadata !{metadata !"branch_weights", i32 3, i32 1} +!13 = metadata !{metadata !"branch_weights", i32 1, i32 1} + +; Irreducible control flow between two true loops. +define void @loop_around_branch_with_irreducible_around_loop(i1 %x) { +; CHECK-LABEL: Printing analysis {{.*}} for function 'loop_around_branch_with_irreducible_around_loop': +; CHECK-NEXT: block-frequency-info: loop_around_branch_with_irreducible_around_loop +entry: +; CHECK-NEXT: entry: float = 1.0, int = [[ENTRY:[0-9]+]] + br label %loop + +loop: +; CHECK-NEXT: loop: float = 3.0, int = [[LOOP:[0-9]+]] + br i1 %x, label %normal, label %irreducible, !prof !14 + +normal: +; CHECK-NEXT: normal: float = 2.0, + br label %loop.end + +irreducible: +; CHECK-NEXT: irreducible: float = 1.0, + br i1 %x, label %left, label %right, !prof !15 + +left: +; CHECK-NEXT: left: float = 2.0, + br i1 %x, label %right, label %loop.end, !prof !16 + +right: +; CHECK-NEXT: right: float = 2.0, int = [[RIGHT:[0-9]+]] + br label %right.loop + +right.loop: +; CHECK-NEXT: right.loop: float = 10.0, + br i1 %x, label %right.loop, label %right.end, !prof !17 + +right.end: +; CHECK-NEXT: right.end: float = 2.0, int = [[RIGHT]] + br i1 %x, label %left, label %loop.end, !prof !16 + +loop.end: +; CHECK-NEXT: loop.end: float = 3.0, int = [[LOOP]] + br i1 %x, label %loop, label %exit, !prof !14 + +exit: +; CHECK-NEXT: exit: float = 1.0, int = [[ENTRY]] + ret void +} +!14 = metadata !{metadata !"branch_weights", i32 2, i32 1} +!15 = metadata !{metadata !"branch_weights", i32 1, i32 1} +!16 = metadata !{metadata !"branch_weights", i32 3, i32 1} +!17 = metadata !{metadata !"branch_weights", i32 4, i32 1} + +; An irreducible SCC with a non-header. +define void @nonheader(i1 %x) { +; CHECK-LABEL: Printing analysis {{.*}} for function 'nonheader': +; CHECK-NEXT: block-frequency-info: nonheader +entry: +; CHECK-NEXT: entry: float = 1.0, int = [[ENTRY:[0-9]+]] + br i1 %x, label %left, label %right, !prof !18 + +left: +; CHECK-NEXT: left: float = 1.0, + br i1 %x, label %bottom, label %exit, !prof !19 + +right: +; CHECK-NEXT: right: float = 1.0, + br i1 %x, label %bottom, label %exit, !prof !20 + +bottom: +; CHECK-NEXT: bottom: float = 1.0, + br i1 %x, label %left, label %right, !prof !18 + +exit: +; CHECK-NEXT: exit: float = 1.0, int = [[ENTRY]] + ret void +} +!18 = metadata !{metadata !"branch_weights", i32 1, i32 1} +!19 = metadata !{metadata !"branch_weights", i32 1, i32 3} +!20 = metadata !{metadata !"branch_weights", i32 3, i32 1} + +; An irreducible SCC with an irreducible sub-SCC. In the current version of +; -block-freq, this means an extra header. +; +; This testcases uses non-trivial branch weights. The CHECK statements here +; will start to fail if we change -block-freq to be more accurate. Currently, +; we expect left, right and top to be treated as equal headers. +define void @nonentry_header(i1 %x, i2 %y) { +; CHECK-LABEL: Printing analysis {{.*}} for function 'nonentry_header': +; CHECK-NEXT: block-frequency-info: nonentry_header +entry: +; CHECK-NEXT: entry: float = 1.0, int = [[ENTRY:[0-9]+]] + br i1 %x, label %left, label %right, !prof !21 + +left: +; CHECK-NEXT: left: float = 3.0, + br i1 %x, label %top, label %bottom, !prof !22 + +right: +; CHECK-NEXT: right: float = 3.0, + br i1 %x, label %top, label %bottom, !prof !22 + +top: +; CHECK-NEXT: top: float = 3.0, + switch i2 %y, label %exit [ i2 0, label %left + i2 1, label %right + i2 2, label %bottom ], !prof !23 + +bottom: +; CHECK-NEXT: bottom: float = 4.5, + br label %top + +exit: +; CHECK-NEXT: exit: float = 1.0, int = [[ENTRY]] + ret void +} +!21 = metadata !{metadata !"branch_weights", i32 2, i32 1} +!22 = metadata !{metadata !"branch_weights", i32 1, i32 1} +!23 = metadata !{metadata !"branch_weights", i32 8, i32 1, i32 3, i32 12} diff --git a/test/Analysis/BlockFrequencyInfo/loop_with_branch.ll b/test/Analysis/BlockFrequencyInfo/loop_with_branch.ll new file mode 100644 index 0000000..9d27b6b --- /dev/null +++ b/test/Analysis/BlockFrequencyInfo/loop_with_branch.ll @@ -0,0 +1,44 @@ +; RUN: opt < %s -analyze -block-freq | FileCheck %s + +; CHECK-LABEL: Printing analysis {{.*}} for function 'loop_with_branch': +; CHECK-NEXT: block-frequency-info: loop_with_branch +define void @loop_with_branch(i32 %a) { +; CHECK-NEXT: entry: float = 1.0, int = [[ENTRY:[0-9]+]] +entry: + %skip_loop = call i1 @foo0(i32 %a) + br i1 %skip_loop, label %skip, label %header, !prof !0 + +; CHECK-NEXT: skip: float = 0.25, +skip: + br label %exit + +; CHECK-NEXT: header: float = 4.5, +header: + %i = phi i32 [ 0, %entry ], [ %i.next, %back ] + %i.next = add i32 %i, 1 + %choose = call i2 @foo1(i32 %i) + switch i2 %choose, label %exit [ i2 0, label %left + i2 1, label %right ], !prof !1 + +; CHECK-NEXT: left: float = 1.5, +left: + br label %back + +; CHECK-NEXT: right: float = 2.25, +right: + br label %back + +; CHECK-NEXT: back: float = 3.75, +back: + br label %header + +; CHECK-NEXT: exit: float = 1.0, int = [[ENTRY]] +exit: + ret void +} + +declare i1 @foo0(i32) +declare i2 @foo1(i32) + +!0 = metadata !{metadata !"branch_weights", i32 1, i32 3} +!1 = metadata !{metadata !"branch_weights", i32 1, i32 2, i32 3} diff --git a/test/Analysis/BlockFrequencyInfo/nested_loop_with_branches.ll b/test/Analysis/BlockFrequencyInfo/nested_loop_with_branches.ll new file mode 100644 index 0000000..d93ffce --- /dev/null +++ b/test/Analysis/BlockFrequencyInfo/nested_loop_with_branches.ll @@ -0,0 +1,59 @@ +; RUN: opt < %s -analyze -block-freq | FileCheck %s + +; CHECK-LABEL: Printing analysis {{.*}} for function 'nested_loop_with_branches' +; CHECK-NEXT: block-frequency-info: nested_loop_with_branches +define void @nested_loop_with_branches(i32 %a) { +; CHECK-NEXT: entry: float = 1.0, int = [[ENTRY:[0-9]+]] +entry: + %v0 = call i1 @foo0(i32 %a) + br i1 %v0, label %exit, label %outer, !prof !0 + +; CHECK-NEXT: outer: float = 12.0, +outer: + %i = phi i32 [ 0, %entry ], [ %i.next, %inner.end ], [ %i.next, %no_inner ] + %i.next = add i32 %i, 1 + %do_inner = call i1 @foo1(i32 %i) + br i1 %do_inner, label %no_inner, label %inner, !prof !0 + +; CHECK-NEXT: inner: float = 36.0, +inner: + %j = phi i32 [ 0, %outer ], [ %j.next, %inner.end ] + %side = call i1 @foo3(i32 %j) + br i1 %side, label %left, label %right, !prof !0 + +; CHECK-NEXT: left: float = 9.0, +left: + %v4 = call i1 @foo4(i32 %j) + br label %inner.end + +; CHECK-NEXT: right: float = 27.0, +right: + %v5 = call i1 @foo5(i32 %j) + br label %inner.end + +; CHECK-NEXT: inner.end: float = 36.0, +inner.end: + %stay_inner = phi i1 [ %v4, %left ], [ %v5, %right ] + %j.next = add i32 %j, 1 + br i1 %stay_inner, label %inner, label %outer, !prof !1 + +; CHECK-NEXT: no_inner: float = 3.0, +no_inner: + %continue = call i1 @foo6(i32 %i) + br i1 %continue, label %outer, label %exit, !prof !1 + +; CHECK-NEXT: exit: float = 1.0, int = [[ENTRY]] +exit: + ret void +} + +declare i1 @foo0(i32) +declare i1 @foo1(i32) +declare i1 @foo2(i32) +declare i1 @foo3(i32) +declare i1 @foo4(i32) +declare i1 @foo5(i32) +declare i1 @foo6(i32) + +!0 = metadata !{metadata !"branch_weights", i32 1, i32 3} +!1 = metadata !{metadata !"branch_weights", i32 3, i32 1} |