maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/Analysis/BranchProbabilityInfo/basic.ll

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

600 lines
19 KiB
LLVM
Raw Normal View History

Teach the BranchProbabilityInfo pass to print its results, and use that to bring it under direct test instead of merely indirectly testing it in the BlockFrequencyInfo pass. The next step is to start adding tests for the various heuristics employed, and to start fixing those heuristics once they're under test. llvm-svn: 142778
2011-10-24 05:21:50 +08:00
; RUN: opt < %s -analyze -branch-prob | FileCheck %s
[BPI] Add new LazyBPI analysis Summary: The motivation is the same as in D22141: In order to add the hotness attribute to optimization remarks we need BFI to be available in all passes that emit optimization remarks. BFI depends on BPI so unless we make this lazy as well we would still compute BPI unconditionally. The solution is to use the new LazyBPI pass in LazyBFI and only compute BPI when computation of BFI is requested by the client. I extended the laziness test using a LoopDistribute test to also cover BPI. Reviewers: hfinkel, davidxl Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D22835 llvm-svn: 277083
2016-07-29 07:31:12 +08:00
; RUN: opt < %s -analyze -lazy-branch-prob | FileCheck %s
[PM] Port Branch Probability Analysis pass to the new pass manager. Differential Revision: http://reviews.llvm.org/D19839 llvm-svn: 268601
2016-05-05 10:59:57 +08:00
; RUN: opt < %s -passes='print<branch-prob>' -disable-output 2>&1 | FileCheck %s
Teach the BranchProbabilityInfo pass to print its results, and use that to bring it under direct test instead of merely indirectly testing it in the BlockFrequencyInfo pass. The next step is to start adding tests for the various heuristics employed, and to start fixing those heuristics once they're under test. llvm-svn: 142778
2011-10-24 05:21:50 +08:00
define i32 @test1(i32 %i, i32* %a) {
; CHECK: Printing analysis {{.*}} for function 'test1'
entry:
br label %body
Use fixed-point representation for BranchProbability. BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change: Pros: 1. It uses only a half space of the current one. 2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer. Cons: 1. Constructing a probability using arbitrary numerator and denominator needs additional calculations. 2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio. One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern. Differential revision: http://reviews.llvm.org/D12603 llvm-svn: 248633
2015-09-26 07:09:59 +08:00
; CHECK: edge entry -> body probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
Teach the BranchProbabilityInfo pass to print its results, and use that to bring it under direct test instead of merely indirectly testing it in the BlockFrequencyInfo pass. The next step is to start adding tests for the various heuristics employed, and to start fixing those heuristics once they're under test. llvm-svn: 142778
2011-10-24 05:21:50 +08:00
body:
%iv = phi i32 [ 0, %entry ], [ %next, %body ]
%base = phi i32 [ 0, %entry ], [ %sum, %body ]
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction One of several parallel first steps to remove the target type of pointers, replacing them with a single opaque pointer type. This adds an explicit type parameter to the gep instruction so that when the first parameter becomes an opaque pointer type, the type to gep through is still available to the instructions. * This doesn't modify gep operators, only instructions (operators will be handled separately) * Textual IR changes only. Bitcode (including upgrade) and changing the in-memory representation will be in separate changes. * geps of vectors are transformed as: getelementptr <4 x float*> %x, ... ->getelementptr float, <4 x float*> %x, ... Then, once the opaque pointer type is introduced, this will ultimately look like: getelementptr float, <4 x ptr> %x with the unambiguous interpretation that it is a vector of pointers to float. * address spaces remain on the pointer, not the type: getelementptr float addrspace(1)* %x ->getelementptr float, float addrspace(1)* %x Then, eventually: getelementptr float, ptr addrspace(1) %x Importantly, the massive amount of test case churn has been automated by same crappy python code. I had to manually update a few test cases that wouldn't fit the script's model (r228970,r229196,r229197,r229198). The python script just massages stdin and writes the result to stdout, I then wrapped that in a shell script to handle replacing files, then using the usual find+xargs to migrate all the files. update.py: import fileinput import sys import re ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") def conv(match, line): if not match: return line line = match.groups()[0] if len(match.groups()[5]) == 0: line += match.groups()[2] line += match.groups()[3] line += ", " line += match.groups()[1] line += "\n" return line for line in sys.stdin: if line.find("getelementptr ") == line.find("getelementptr inbounds"): if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("): line = conv(re.match(ibrep, line), line) elif line.find("getelementptr ") != line.find("getelementptr ("): line = conv(re.match(normrep, line), line) sys.stdout.write(line) apply.sh: for name in "$@" do python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name" rm -f "$name.tmp" done The actual commands: From llvm/src: find test/ -name *.ll | xargs ./apply.sh From llvm/src/tools/clang: find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}" From llvm/src/tools/polly: find test/ -name *.ll | xargs ./apply.sh After that, check-all (with llvm, clang, clang-tools-extra, lld, compiler-rt, and polly all checked out). The extra 'rm' in the apply.sh script is due to a few files in clang's test suite using interesting unicode stuff that my python script was throwing exceptions on. None of those files needed to be migrated, so it seemed sufficient to ignore those cases. Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7636 llvm-svn: 230786
2015-02-28 03:29:02 +08:00
%arrayidx = getelementptr inbounds i32, i32* %a, i32 %iv
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%0 = load i32, i32* %arrayidx
Teach the BranchProbabilityInfo pass to print its results, and use that to bring it under direct test instead of merely indirectly testing it in the BlockFrequencyInfo pass. The next step is to start adding tests for the various heuristics employed, and to start fixing those heuristics once they're under test. llvm-svn: 142778
2011-10-24 05:21:50 +08:00
%sum = add nsw i32 %0, %base
%next = add i32 %iv, 1
%exitcond = icmp eq i32 %next, %i
br i1 %exitcond, label %exit, label %body
Use fixed-point representation for BranchProbability. BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change: Pros: 1. It uses only a half space of the current one. 2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer. Cons: 1. Constructing a probability using arbitrary numerator and denominator needs additional calculations. 2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio. One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern. Differential revision: http://reviews.llvm.org/D12603 llvm-svn: 248633
2015-09-26 07:09:59 +08:00
; CHECK: edge body -> exit probability is 0x04000000 / 0x80000000 = 3.12%
; CHECK: edge body -> body probability is 0x7c000000 / 0x80000000 = 96.88% [HOT edge]
Teach the BranchProbabilityInfo pass to print its results, and use that to bring it under direct test instead of merely indirectly testing it in the BlockFrequencyInfo pass. The next step is to start adding tests for the various heuristics employed, and to start fixing those heuristics once they're under test. llvm-svn: 142778
2011-10-24 05:21:50 +08:00
exit:
ret i32 %sum
}
define i32 @test2(i32 %i, i32 %a, i32 %b) {
; CHECK: Printing analysis {{.*}} for function 'test2'
entry:
%cond = icmp ult i32 %i, 42
br i1 %cond, label %then, label %else, !prof !0
Use fixed-point representation for BranchProbability. BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change: Pros: 1. It uses only a half space of the current one. 2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer. Cons: 1. Constructing a probability using arbitrary numerator and denominator needs additional calculations. 2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio. One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern. Differential revision: http://reviews.llvm.org/D12603 llvm-svn: 248633
2015-09-26 07:09:59 +08:00
; CHECK: edge entry -> then probability is 0x78787878 / 0x80000000 = 94.12% [HOT edge]
; CHECK: edge entry -> else probability is 0x07878788 / 0x80000000 = 5.88%
Teach the BranchProbabilityInfo pass to print its results, and use that to bring it under direct test instead of merely indirectly testing it in the BlockFrequencyInfo pass. The next step is to start adding tests for the various heuristics employed, and to start fixing those heuristics once they're under test. llvm-svn: 142778
2011-10-24 05:21:50 +08:00
then:
br label %exit
Use fixed-point representation for BranchProbability. BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change: Pros: 1. It uses only a half space of the current one. 2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer. Cons: 1. Constructing a probability using arbitrary numerator and denominator needs additional calculations. 2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio. One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern. Differential revision: http://reviews.llvm.org/D12603 llvm-svn: 248633
2015-09-26 07:09:59 +08:00
; CHECK: edge then -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
Teach the BranchProbabilityInfo pass to print its results, and use that to bring it under direct test instead of merely indirectly testing it in the BlockFrequencyInfo pass. The next step is to start adding tests for the various heuristics employed, and to start fixing those heuristics once they're under test. llvm-svn: 142778
2011-10-24 05:21:50 +08:00
else:
br label %exit
Use fixed-point representation for BranchProbability. BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change: Pros: 1. It uses only a half space of the current one. 2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer. Cons: 1. Constructing a probability using arbitrary numerator and denominator needs additional calculations. 2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio. One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern. Differential revision: http://reviews.llvm.org/D12603 llvm-svn: 248633
2015-09-26 07:09:59 +08:00
; CHECK: edge else -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
Teach the BranchProbabilityInfo pass to print its results, and use that to bring it under direct test instead of merely indirectly testing it in the BlockFrequencyInfo pass. The next step is to start adding tests for the various heuristics employed, and to start fixing those heuristics once they're under test. llvm-svn: 142778
2011-10-24 05:21:50 +08:00
exit:
%result = phi i32 [ %a, %then ], [ %b, %else ]
ret i32 %result
}
IR: Make metadata typeless in assembly Now that `Metadata` is typeless, reflect that in the assembly. These are the matching assembly changes for the metadata/value split in r223802. - Only use the `metadata` type when referencing metadata from a call intrinsic -- i.e., only when it's used as a `Value`. - Stop pretending that `ValueAsMetadata` is wrapped in an `MDNode` when referencing it from call intrinsics. So, assembly like this: define @foo(i32 %v) { call void @llvm.foo(metadata !{i32 %v}, metadata !0) call void @llvm.foo(metadata !{i32 7}, metadata !0) call void @llvm.foo(metadata !1, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{metadata !3}, metadata !0) ret void, !bar !2 } !0 = metadata !{metadata !2} !1 = metadata !{i32* @global} !2 = metadata !{metadata !3} !3 = metadata !{} turns into this: define @foo(i32 %v) { call void @llvm.foo(metadata i32 %v, metadata !0) call void @llvm.foo(metadata i32 7, metadata !0) call void @llvm.foo(metadata i32* @global, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{!3}, metadata !0) ret void, !bar !2 } !0 = !{!2} !1 = !{i32* @global} !2 = !{!3} !3 = !{} I wrote an upgrade script that handled almost all of the tests in llvm and many of the tests in cfe (even handling many `CHECK` lines). I've attached it (or will attach it in a moment if you're speedy) to PR21532 to help everyone update their out-of-tree testcases. This is part of PR21532. llvm-svn: 224257
2014-12-16 03:07:53 +08:00
!0 = !{!"branch_weights", i32 64, i32 4}
Teach the BranchProbabilityInfo pass to print its results, and use that to bring it under direct test instead of merely indirectly testing it in the BlockFrequencyInfo pass. The next step is to start adding tests for the various heuristics employed, and to start fixing those heuristics once they're under test. llvm-svn: 142778
2011-10-24 05:21:50 +08:00
define i32 @test3(i32 %i, i32 %a, i32 %b, i32 %c, i32 %d, i32 %e) {
; CHECK: Printing analysis {{.*}} for function 'test3'
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
Use fixed-point representation for BranchProbability. BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change: Pros: 1. It uses only a half space of the current one. 2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer. Cons: 1. Constructing a probability using arbitrary numerator and denominator needs additional calculations. 2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio. One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern. Differential revision: http://reviews.llvm.org/D12603 llvm-svn: 248633
2015-09-26 07:09:59 +08:00
; CHECK: edge entry -> case_a probability is 0x06666666 / 0x80000000 = 5.00%
; CHECK: edge entry -> case_b probability is 0x06666666 / 0x80000000 = 5.00%
; CHECK: edge entry -> case_c probability is 0x66666666 / 0x80000000 = 80.00%
; CHECK: edge entry -> case_d probability is 0x06666666 / 0x80000000 = 5.00%
; CHECK: edge entry -> case_e probability is 0x06666666 / 0x80000000 = 5.00%
Teach the BranchProbabilityInfo pass to print its results, and use that to bring it under direct test instead of merely indirectly testing it in the BlockFrequencyInfo pass. The next step is to start adding tests for the various heuristics employed, and to start fixing those heuristics once they're under test. llvm-svn: 142778
2011-10-24 05:21:50 +08:00
case_a:
br label %exit
Use fixed-point representation for BranchProbability. BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change: Pros: 1. It uses only a half space of the current one. 2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer. Cons: 1. Constructing a probability using arbitrary numerator and denominator needs additional calculations. 2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio. One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern. Differential revision: http://reviews.llvm.org/D12603 llvm-svn: 248633
2015-09-26 07:09:59 +08:00
; CHECK: edge case_a -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
Teach the BranchProbabilityInfo pass to print its results, and use that to bring it under direct test instead of merely indirectly testing it in the BlockFrequencyInfo pass. The next step is to start adding tests for the various heuristics employed, and to start fixing those heuristics once they're under test. llvm-svn: 142778
2011-10-24 05:21:50 +08:00
case_b:
br label %exit
Use fixed-point representation for BranchProbability. BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change: Pros: 1. It uses only a half space of the current one. 2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer. Cons: 1. Constructing a probability using arbitrary numerator and denominator needs additional calculations. 2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio. One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern. Differential revision: http://reviews.llvm.org/D12603 llvm-svn: 248633
2015-09-26 07:09:59 +08:00
; CHECK: edge case_b -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
Teach the BranchProbabilityInfo pass to print its results, and use that to bring it under direct test instead of merely indirectly testing it in the BlockFrequencyInfo pass. The next step is to start adding tests for the various heuristics employed, and to start fixing those heuristics once they're under test. llvm-svn: 142778
2011-10-24 05:21:50 +08:00
case_c:
br label %exit
Use fixed-point representation for BranchProbability. BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change: Pros: 1. It uses only a half space of the current one. 2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer. Cons: 1. Constructing a probability using arbitrary numerator and denominator needs additional calculations. 2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio. One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern. Differential revision: http://reviews.llvm.org/D12603 llvm-svn: 248633
2015-09-26 07:09:59 +08:00
; CHECK: edge case_c -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
Teach the BranchProbabilityInfo pass to print its results, and use that to bring it under direct test instead of merely indirectly testing it in the BlockFrequencyInfo pass. The next step is to start adding tests for the various heuristics employed, and to start fixing those heuristics once they're under test. llvm-svn: 142778
2011-10-24 05:21:50 +08:00
case_d:
br label %exit
Use fixed-point representation for BranchProbability. BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change: Pros: 1. It uses only a half space of the current one. 2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer. Cons: 1. Constructing a probability using arbitrary numerator and denominator needs additional calculations. 2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio. One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern. Differential revision: http://reviews.llvm.org/D12603 llvm-svn: 248633
2015-09-26 07:09:59 +08:00
; CHECK: edge case_d -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
Teach the BranchProbabilityInfo pass to print its results, and use that to bring it under direct test instead of merely indirectly testing it in the BlockFrequencyInfo pass. The next step is to start adding tests for the various heuristics employed, and to start fixing those heuristics once they're under test. llvm-svn: 142778
2011-10-24 05:21:50 +08:00
case_e:
br label %exit
Use fixed-point representation for BranchProbability. BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change: Pros: 1. It uses only a half space of the current one. 2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer. Cons: 1. Constructing a probability using arbitrary numerator and denominator needs additional calculations. 2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio. One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern. Differential revision: http://reviews.llvm.org/D12603 llvm-svn: 248633
2015-09-26 07:09:59 +08:00
; CHECK: edge case_e -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
Teach the BranchProbabilityInfo pass to print its results, and use that to bring it under direct test instead of merely indirectly testing it in the BlockFrequencyInfo pass. The next step is to start adding tests for the various heuristics employed, and to start fixing those heuristics once they're under test. llvm-svn: 142778
2011-10-24 05:21:50 +08:00
exit:
%result = phi i32 [ %a, %case_a ],
[ %b, %case_b ],
[ %c, %case_c ],
[ %d, %case_d ],
[ %e, %case_e ]
ret i32 %result
}
IR: Make metadata typeless in assembly Now that `Metadata` is typeless, reflect that in the assembly. These are the matching assembly changes for the metadata/value split in r223802. - Only use the `metadata` type when referencing metadata from a call intrinsic -- i.e., only when it's used as a `Value`. - Stop pretending that `ValueAsMetadata` is wrapped in an `MDNode` when referencing it from call intrinsics. So, assembly like this: define @foo(i32 %v) { call void @llvm.foo(metadata !{i32 %v}, metadata !0) call void @llvm.foo(metadata !{i32 7}, metadata !0) call void @llvm.foo(metadata !1, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{metadata !3}, metadata !0) ret void, !bar !2 } !0 = metadata !{metadata !2} !1 = metadata !{i32* @global} !2 = metadata !{metadata !3} !3 = metadata !{} turns into this: define @foo(i32 %v) { call void @llvm.foo(metadata i32 %v, metadata !0) call void @llvm.foo(metadata i32 7, metadata !0) call void @llvm.foo(metadata i32* @global, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{!3}, metadata !0) ret void, !bar !2 } !0 = !{!2} !1 = !{i32* @global} !2 = !{!3} !3 = !{} I wrote an upgrade script that handled almost all of the tests in llvm and many of the tests in cfe (even handling many `CHECK` lines). I've attached it (or will attach it in a moment if you're speedy) to PR21532 to help everyone update their out-of-tree testcases. This is part of PR21532. llvm-svn: 224257
2014-12-16 03:07:53 +08:00
!1 = !{!"branch_weights", i32 4, i32 4, i32 64, i32 4, i32 4}
BranchProb: modify the definition of an edge in BranchProbabilityInfo to handle the case of multiple edges from one block to another. A simple example is a switch statement with multiple values to the same destination. The definition of an edge is modified from a pair of blocks to a pair of PredBlock and an index into the successors. Also set the weight correctly when building SelectionDAG from LLVM IR, especially when converting a Switch. IntegersSubsetMapping is updated to calculate the weight for each cluster. llvm-svn: 162572
2012-08-25 02:14:27 +08:00
define i32 @test4(i32 %x) nounwind uwtable readnone ssp {
; CHECK: Printing analysis {{.*}} for function 'test4'
entry:
%conv = sext i32 %x to i64
switch i64 %conv, label %return [
i64 0, label %sw.bb
i64 1, label %sw.bb
i64 2, label %sw.bb
i64 5, label %sw.bb1
], !prof !2
Use fixed-point representation for BranchProbability. BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change: Pros: 1. It uses only a half space of the current one. 2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer. Cons: 1. Constructing a probability using arbitrary numerator and denominator needs additional calculations. 2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio. One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern. Differential revision: http://reviews.llvm.org/D12603 llvm-svn: 248633
2015-09-26 07:09:59 +08:00
; CHECK: edge entry -> return probability is 0x0a8a8a8b / 0x80000000 = 8.24%
; CHECK: edge entry -> sw.bb probability is 0x15151515 / 0x80000000 = 16.47%
; CHECK: edge entry -> sw.bb1 probability is 0x60606060 / 0x80000000 = 75.29%
BranchProb: modify the definition of an edge in BranchProbabilityInfo to handle the case of multiple edges from one block to another. A simple example is a switch statement with multiple values to the same destination. The definition of an edge is modified from a pair of blocks to a pair of PredBlock and an index into the successors. Also set the weight correctly when building SelectionDAG from LLVM IR, especially when converting a Switch. IntegersSubsetMapping is updated to calculate the weight for each cluster. llvm-svn: 162572
2012-08-25 02:14:27 +08:00
sw.bb:
br label %return
sw.bb1:
br label %return
return:
%retval.0 = phi i32 [ 5, %sw.bb1 ], [ 1, %sw.bb ], [ 0, %entry ]
ret i32 %retval.0
}
IR: Make metadata typeless in assembly Now that `Metadata` is typeless, reflect that in the assembly. These are the matching assembly changes for the metadata/value split in r223802. - Only use the `metadata` type when referencing metadata from a call intrinsic -- i.e., only when it's used as a `Value`. - Stop pretending that `ValueAsMetadata` is wrapped in an `MDNode` when referencing it from call intrinsics. So, assembly like this: define @foo(i32 %v) { call void @llvm.foo(metadata !{i32 %v}, metadata !0) call void @llvm.foo(metadata !{i32 7}, metadata !0) call void @llvm.foo(metadata !1, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{metadata !3}, metadata !0) ret void, !bar !2 } !0 = metadata !{metadata !2} !1 = metadata !{i32* @global} !2 = metadata !{metadata !3} !3 = metadata !{} turns into this: define @foo(i32 %v) { call void @llvm.foo(metadata i32 %v, metadata !0) call void @llvm.foo(metadata i32 7, metadata !0) call void @llvm.foo(metadata i32* @global, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{!3}, metadata !0) ret void, !bar !2 } !0 = !{!2} !1 = !{i32* @global} !2 = !{!3} !3 = !{} I wrote an upgrade script that handled almost all of the tests in llvm and many of the tests in cfe (even handling many `CHECK` lines). I've attached it (or will attach it in a moment if you're speedy) to PR21532 to help everyone update their out-of-tree testcases. This is part of PR21532. llvm-svn: 224257
2014-12-16 03:07:53 +08:00
!2 = !{!"branch_weights", i32 7, i32 6, i32 4, i32 4, i32 64}
Add a new function attribute 'cold' to functions. Other than recognizing the attribute, the patch does little else. It changes the branch probability analyzer so that edges into blocks postdominated by a cold function are given low weight. Added analysis and code generation tests. Added documentation for the new attribute. llvm-svn: 182638
2013-05-24 20:26:52 +08:00
declare void @coldfunc() cold
define i32 @test5(i32 %a, i32 %b, i1 %flag) {
; CHECK: Printing analysis {{.*}} for function 'test5'
entry:
br i1 %flag, label %then, label %else
Use fixed-point representation for BranchProbability. BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change: Pros: 1. It uses only a half space of the current one. 2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer. Cons: 1. Constructing a probability using arbitrary numerator and denominator needs additional calculations. 2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio. One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern. Differential revision: http://reviews.llvm.org/D12603 llvm-svn: 248633
2015-09-26 07:09:59 +08:00
; CHECK: edge entry -> then probability is 0x07878788 / 0x80000000 = 5.88%
; CHECK: edge entry -> else probability is 0x78787878 / 0x80000000 = 94.12% [HOT edge]
Add a new function attribute 'cold' to functions. Other than recognizing the attribute, the patch does little else. It changes the branch probability analyzer so that edges into blocks postdominated by a cold function are given low weight. Added analysis and code generation tests. Added documentation for the new attribute. llvm-svn: 182638
2013-05-24 20:26:52 +08:00
then:
call void @coldfunc()
br label %exit
Use fixed-point representation for BranchProbability. BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change: Pros: 1. It uses only a half space of the current one. 2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer. Cons: 1. Constructing a probability using arbitrary numerator and denominator needs additional calculations. 2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio. One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern. Differential revision: http://reviews.llvm.org/D12603 llvm-svn: 248633
2015-09-26 07:09:59 +08:00
; CHECK: edge then -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
Add a new function attribute 'cold' to functions. Other than recognizing the attribute, the patch does little else. It changes the branch probability analyzer so that edges into blocks postdominated by a cold function are given low weight. Added analysis and code generation tests. Added documentation for the new attribute. llvm-svn: 182638
2013-05-24 20:26:52 +08:00
else:
br label %exit
Use fixed-point representation for BranchProbability. BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change: Pros: 1. It uses only a half space of the current one. 2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer. Cons: 1. Constructing a probability using arbitrary numerator and denominator needs additional calculations. 2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio. One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern. Differential revision: http://reviews.llvm.org/D12603 llvm-svn: 248633
2015-09-26 07:09:59 +08:00
; CHECK: edge else -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
Add a new function attribute 'cold' to functions. Other than recognizing the attribute, the patch does little else. It changes the branch probability analyzer so that edges into blocks postdominated by a cold function are given low weight. Added analysis and code generation tests. Added documentation for the new attribute. llvm-svn: 182638
2013-05-24 20:26:52 +08:00
exit:
%result = phi i32 [ %a, %then ], [ %b, %else ]
ret i32 %result
}
declare i32 @regular_function(i32 %i)
[BPI] Refactor post domination calculation and simple fix for ColdCall Collection of PostDominatedByUnreachable and PostDominatedByColdCall have been split out of heuristics itself. Update of the data happens now for each basic block (before update for PostDominatedByColdCall might be skipped if unreachable or matadata heuristic handled this basic block). This separation allows re-ordering of heuristics without loosing the post-domination information. Reviewers: sanjoy, junbuml, vsk, chandlerc, reames Reviewed By: chandlerc Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D31701 llvm-svn: 300029
2017-04-12 13:42:14 +08:00
define i32 @test_cold_call_sites_with_prof(i32 %a, i32 %b, i1 %flag, i1 %flag2) {
; CHECK: Printing analysis {{.*}} for function 'test_cold_call_sites_with_prof'
entry:
br i1 %flag, label %then, label %else
; CHECK: edge entry -> then probability is 0x07878788 / 0x80000000 = 5.88%
; CHECK: edge entry -> else probability is 0x78787878 / 0x80000000 = 94.12% [HOT edge]
then:
br i1 %flag2, label %then2, label %else2, !prof !3
; CHECK: edge then -> then2 probability is 0x7ebb907a / 0x80000000 = 99.01% [HOT edge]
; CHECK: edge then -> else2 probability is 0x01446f86 / 0x80000000 = 0.99%
then2:
br label %join
; CHECK: edge then2 -> join probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
else2:
br label %join
; CHECK: edge else2 -> join probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
join:
%joinresult = phi i32 [ %a, %then2 ], [ %b, %else2 ]
call void @coldfunc()
br label %exit
; CHECK: edge join -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
else:
br label %exit
; CHECK: edge else -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
exit:
%result = phi i32 [ %joinresult, %join ], [ %b, %else ]
ret i32 %result
}
!3 = !{!"branch_weights", i32 100, i32 1}
Add a new function attribute 'cold' to functions. Other than recognizing the attribute, the patch does little else. It changes the branch probability analyzer so that edges into blocks postdominated by a cold function are given low weight. Added analysis and code generation tests. Added documentation for the new attribute. llvm-svn: 182638
2013-05-24 20:26:52 +08:00
define i32 @test_cold_call_sites(i32* %a) {
; Test that edges to blocks post-dominated by cold call sites
; are marked as not expected to be taken.
; TODO(dnovillo) The calls to regular_function should not be merged, but
; they are currently being merged. Convert this into a code generation test
; after that is fixed.
; CHECK: Printing analysis {{.*}} for function 'test_cold_call_sites'
Use fixed-point representation for BranchProbability. BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change: Pros: 1. It uses only a half space of the current one. 2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer. Cons: 1. Constructing a probability using arbitrary numerator and denominator needs additional calculations. 2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio. One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern. Differential revision: http://reviews.llvm.org/D12603 llvm-svn: 248633
2015-09-26 07:09:59 +08:00
; CHECK: edge entry -> then probability is 0x07878788 / 0x80000000 = 5.88%
; CHECK: edge entry -> else probability is 0x78787878 / 0x80000000 = 94.12% [HOT edge]
Add a new function attribute 'cold' to functions. Other than recognizing the attribute, the patch does little else. It changes the branch probability analyzer so that edges into blocks postdominated by a cold function are given low weight. Added analysis and code generation tests. Added documentation for the new attribute. llvm-svn: 182638
2013-05-24 20:26:52 +08:00
entry:
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction One of several parallel first steps to remove the target type of pointers, replacing them with a single opaque pointer type. This adds an explicit type parameter to the gep instruction so that when the first parameter becomes an opaque pointer type, the type to gep through is still available to the instructions. * This doesn't modify gep operators, only instructions (operators will be handled separately) * Textual IR changes only. Bitcode (including upgrade) and changing the in-memory representation will be in separate changes. * geps of vectors are transformed as: getelementptr <4 x float*> %x, ... ->getelementptr float, <4 x float*> %x, ... Then, once the opaque pointer type is introduced, this will ultimately look like: getelementptr float, <4 x ptr> %x with the unambiguous interpretation that it is a vector of pointers to float. * address spaces remain on the pointer, not the type: getelementptr float addrspace(1)* %x ->getelementptr float, float addrspace(1)* %x Then, eventually: getelementptr float, ptr addrspace(1) %x Importantly, the massive amount of test case churn has been automated by same crappy python code. I had to manually update a few test cases that wouldn't fit the script's model (r228970,r229196,r229197,r229198). The python script just massages stdin and writes the result to stdout, I then wrapped that in a shell script to handle replacing files, then using the usual find+xargs to migrate all the files. update.py: import fileinput import sys import re ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") def conv(match, line): if not match: return line line = match.groups()[0] if len(match.groups()[5]) == 0: line += match.groups()[2] line += match.groups()[3] line += ", " line += match.groups()[1] line += "\n" return line for line in sys.stdin: if line.find("getelementptr ") == line.find("getelementptr inbounds"): if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("): line = conv(re.match(ibrep, line), line) elif line.find("getelementptr ") != line.find("getelementptr ("): line = conv(re.match(normrep, line), line) sys.stdout.write(line) apply.sh: for name in "$@" do python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name" rm -f "$name.tmp" done The actual commands: From llvm/src: find test/ -name *.ll | xargs ./apply.sh From llvm/src/tools/clang: find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}" From llvm/src/tools/polly: find test/ -name *.ll | xargs ./apply.sh After that, check-all (with llvm, clang, clang-tools-extra, lld, compiler-rt, and polly all checked out). The extra 'rm' in the apply.sh script is due to a few files in clang's test suite using interesting unicode stuff that my python script was throwing exceptions on. None of those files needed to be migrated, so it seemed sufficient to ignore those cases. Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7636 llvm-svn: 230786
2015-02-28 03:29:02 +08:00
%gep1 = getelementptr i32, i32* %a, i32 1
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%val1 = load i32, i32* %gep1
Add a new function attribute 'cold' to functions. Other than recognizing the attribute, the patch does little else. It changes the branch probability analyzer so that edges into blocks postdominated by a cold function are given low weight. Added analysis and code generation tests. Added documentation for the new attribute. llvm-svn: 182638
2013-05-24 20:26:52 +08:00
%cond1 = icmp ugt i32 %val1, 1
br i1 %cond1, label %then, label %else
then:
; This function is not declared cold, but this call site is.
%val4 = call i32 @regular_function(i32 %val1) cold
br label %exit
else:
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction One of several parallel first steps to remove the target type of pointers, replacing them with a single opaque pointer type. This adds an explicit type parameter to the gep instruction so that when the first parameter becomes an opaque pointer type, the type to gep through is still available to the instructions. * This doesn't modify gep operators, only instructions (operators will be handled separately) * Textual IR changes only. Bitcode (including upgrade) and changing the in-memory representation will be in separate changes. * geps of vectors are transformed as: getelementptr <4 x float*> %x, ... ->getelementptr float, <4 x float*> %x, ... Then, once the opaque pointer type is introduced, this will ultimately look like: getelementptr float, <4 x ptr> %x with the unambiguous interpretation that it is a vector of pointers to float. * address spaces remain on the pointer, not the type: getelementptr float addrspace(1)* %x ->getelementptr float, float addrspace(1)* %x Then, eventually: getelementptr float, ptr addrspace(1) %x Importantly, the massive amount of test case churn has been automated by same crappy python code. I had to manually update a few test cases that wouldn't fit the script's model (r228970,r229196,r229197,r229198). The python script just massages stdin and writes the result to stdout, I then wrapped that in a shell script to handle replacing files, then using the usual find+xargs to migrate all the files. update.py: import fileinput import sys import re ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") def conv(match, line): if not match: return line line = match.groups()[0] if len(match.groups()[5]) == 0: line += match.groups()[2] line += match.groups()[3] line += ", " line += match.groups()[1] line += "\n" return line for line in sys.stdin: if line.find("getelementptr ") == line.find("getelementptr inbounds"): if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("): line = conv(re.match(ibrep, line), line) elif line.find("getelementptr ") != line.find("getelementptr ("): line = conv(re.match(normrep, line), line) sys.stdout.write(line) apply.sh: for name in "$@" do python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name" rm -f "$name.tmp" done The actual commands: From llvm/src: find test/ -name *.ll | xargs ./apply.sh From llvm/src/tools/clang: find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}" From llvm/src/tools/polly: find test/ -name *.ll | xargs ./apply.sh After that, check-all (with llvm, clang, clang-tools-extra, lld, compiler-rt, and polly all checked out). The extra 'rm' in the apply.sh script is due to a few files in clang's test suite using interesting unicode stuff that my python script was throwing exceptions on. None of those files needed to be migrated, so it seemed sufficient to ignore those cases. Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7636 llvm-svn: 230786
2015-02-28 03:29:02 +08:00
%gep2 = getelementptr i32, i32* %a, i32 2
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-28 05:17:42 +08:00
%val2 = load i32, i32* %gep2
Add a new function attribute 'cold' to functions. Other than recognizing the attribute, the patch does little else. It changes the branch probability analyzer so that edges into blocks postdominated by a cold function are given low weight. Added analysis and code generation tests. Added documentation for the new attribute. llvm-svn: 182638
2013-05-24 20:26:52 +08:00
%val3 = call i32 @regular_function(i32 %val2)
br label %exit
exit:
%ret = phi i32 [ %val4, %then ], [ %val3, %else ]
ret i32 %ret
}
Consider (x == -1) unlikely in BranchProbabilityInfo This adds another heuristic to BPI, similar to the existing heuristic that considers (x == 0) unlikely to be true. As suggested in the PACT'98 paper by Deitrich, Cheng, and Hwu, -1 is often used to indicate an invalid index, and equality comparisons with -1 are also unlikely to succeed. Local experimentation supports this hypothesis: This yields a 1-2% speedup in the test-suite sqlite benchmark on the PPC A2 core, with no significant regressions. llvm-svn: 193855
2013-11-01 18:58:22 +08:00
Enhance calcColdCallHeuristics for InvokeInst Summary: When identifying cold blocks, consider only the edge to the normal destination if the terminator is InvokeInst and let calcInvokeHeuristics() decide edge weights for the InvokeInst. Reviewers: mcrosier, hfinkel, davidxl Subscribers: mcrosier, llvm-commits Differential Revision: https://reviews.llvm.org/D24868 llvm-svn: 282262
2016-09-24 01:26:14 +08:00
; CHECK-LABEL: test_invoke_code_callsite1
define i32 @test_invoke_code_callsite1(i1 %c) personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
entry:
br i1 %c, label %if.then, label %if.end
; Edge "entry->if.end" should have higher probability based on the cold call
; heuristic which treat %if.then as a cold block because the normal destination
; of the invoke instruction in %if.then is post-dominated by ColdFunc().
; CHECK: edge entry -> if.then probability is 0x07878788 / 0x80000000 = 5.88%
; CHECK: edge entry -> if.end probability is 0x78787878 / 0x80000000 = 94.12% [HOT edge]
if.then:
invoke i32 @InvokeCall()
to label %invoke.cont unwind label %lpad
; CHECK: edge if.then -> invoke.cont probability is 0x7ffff800 / 0x80000000 = 100.00% [HOT edge]
; CHECK: edge if.then -> lpad probability is 0x00000800 / 0x80000000 = 0.00%
invoke.cont:
call void @ColdFunc() #0
br label %if.end
lpad:
%ll = landingpad { i8*, i32 }
cleanup
br label %if.end
if.end:
ret i32 0
}
; CHECK-LABEL: test_invoke_code_callsite2
define i32 @test_invoke_code_callsite2(i1 %c) personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
entry:
br i1 %c, label %if.then, label %if.end
; CHECK: edge entry -> if.then probability is 0x40000000 / 0x80000000 = 50.00%
; CHECK: edge entry -> if.end probability is 0x40000000 / 0x80000000 = 50.00%
if.then:
invoke i32 @InvokeCall()
to label %invoke.cont unwind label %lpad
; The cold call heuristic should not kick in when the cold callsite is in EH path.
; CHECK: edge if.then -> invoke.cont probability is 0x7ffff800 / 0x80000000 = 100.00% [HOT edge]
; CHECK: edge if.then -> lpad probability is 0x00000800 / 0x80000000 = 0.00%
invoke.cont:
br label %if.end
lpad:
%ll = landingpad { i8*, i32 }
cleanup
call void @ColdFunc() #0
br label %if.end
if.end:
ret i32 0
}
; CHECK-LABEL: test_invoke_code_callsite3
define i32 @test_invoke_code_callsite3(i1 %c) personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
entry:
br i1 %c, label %if.then, label %if.end
; CHECK: edge entry -> if.then probability is 0x07878788 / 0x80000000 = 5.88%
; CHECK: edge entry -> if.end probability is 0x78787878 / 0x80000000 = 94.12% [HOT edge]
if.then:
invoke i32 @InvokeCall()
to label %invoke.cont unwind label %lpad
; Regardless of cold calls, edge weights from a invoke instruction should be
; determined by the invoke heuristic.
; CHECK: edge if.then -> invoke.cont probability is 0x7ffff800 / 0x80000000 = 100.00% [HOT edge]
; CHECK: edge if.then -> lpad probability is 0x00000800 / 0x80000000 = 0.00%
invoke.cont:
call void @ColdFunc() #0
br label %if.end
lpad:
%ll = landingpad { i8*, i32 }
cleanup
call void @ColdFunc() #0
br label %if.end
if.end:
ret i32 0
}
declare i32 @__gxx_personality_v0(...)
declare void @ColdFunc()
declare i32 @InvokeCall()
attributes #0 = { cold }
Consider (x == -1) unlikely in BranchProbabilityInfo This adds another heuristic to BPI, similar to the existing heuristic that considers (x == 0) unlikely to be true. As suggested in the PACT'98 paper by Deitrich, Cheng, and Hwu, -1 is often used to indicate an invalid index, and equality comparisons with -1 are also unlikely to succeed. Local experimentation supports this hypothesis: This yields a 1-2% speedup in the test-suite sqlite benchmark on the PPC A2 core, with no significant regressions. llvm-svn: 193855
2013-11-01 18:58:22 +08:00
define i32 @zero1(i32 %i, i32 %a, i32 %b) {
; CHECK: Printing analysis {{.*}} for function 'zero1'
entry:
%cond = icmp eq i32 %i, 0
br i1 %cond, label %then, label %else
Use fixed-point representation for BranchProbability. BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change: Pros: 1. It uses only a half space of the current one. 2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer. Cons: 1. Constructing a probability using arbitrary numerator and denominator needs additional calculations. 2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio. One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern. Differential revision: http://reviews.llvm.org/D12603 llvm-svn: 248633
2015-09-26 07:09:59 +08:00
; CHECK: edge entry -> then probability is 0x30000000 / 0x80000000 = 37.50%
; CHECK: edge entry -> else probability is 0x50000000 / 0x80000000 = 62.50%
Consider (x == -1) unlikely in BranchProbabilityInfo This adds another heuristic to BPI, similar to the existing heuristic that considers (x == 0) unlikely to be true. As suggested in the PACT'98 paper by Deitrich, Cheng, and Hwu, -1 is often used to indicate an invalid index, and equality comparisons with -1 are also unlikely to succeed. Local experimentation supports this hypothesis: This yields a 1-2% speedup in the test-suite sqlite benchmark on the PPC A2 core, with no significant regressions. llvm-svn: 193855
2013-11-01 18:58:22 +08:00
then:
br label %exit
else:
br label %exit
exit:
%result = phi i32 [ %a, %then ], [ %b, %else ]
ret i32 %result
}
define i32 @zero2(i32 %i, i32 %a, i32 %b) {
; CHECK: Printing analysis {{.*}} for function 'zero2'
entry:
%cond = icmp ne i32 %i, -1
br i1 %cond, label %then, label %else
Use fixed-point representation for BranchProbability. BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change: Pros: 1. It uses only a half space of the current one. 2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer. Cons: 1. Constructing a probability using arbitrary numerator and denominator needs additional calculations. 2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio. One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern. Differential revision: http://reviews.llvm.org/D12603 llvm-svn: 248633
2015-09-26 07:09:59 +08:00
; CHECK: edge entry -> then probability is 0x50000000 / 0x80000000 = 62.50%
; CHECK: edge entry -> else probability is 0x30000000 / 0x80000000 = 37.50%
Consider (x == -1) unlikely in BranchProbabilityInfo This adds another heuristic to BPI, similar to the existing heuristic that considers (x == 0) unlikely to be true. As suggested in the PACT'98 paper by Deitrich, Cheng, and Hwu, -1 is often used to indicate an invalid index, and equality comparisons with -1 are also unlikely to succeed. Local experimentation supports this hypothesis: This yields a 1-2% speedup in the test-suite sqlite benchmark on the PPC A2 core, with no significant regressions. llvm-svn: 193855
2013-11-01 18:58:22 +08:00
then:
br label %exit
else:
br label %exit
exit:
%result = phi i32 [ %a, %then ], [ %b, %else ]
ret i32 %result
}
Re-apply r234898 and fix tests. This commit makes LLVM not estimate branch probabilities when doing a single bit bitmask tests. The code that originally made me discover this is: if ((a & 0x1) == 0x1) { .. } In this case we don't actually have any branch probability information and should not assume to have any. LLVM transforms this into: %and = and i32 %a, 1 %tobool = icmp eq i32 %and, 0 So, in this case, the result of a bitwise and is compared against 0, but nevertheless, we should not assume to have probability information. CodeGen/ARM/2013-10-11-select-stalls.ll started failing because the changed probabilities changed the results of ARMBaseInstrInfo::isProfitableToIfCvt() and led to an Ifcvt of the diamond in the test. AFAICT, the test was never meant to test this and thus changing the test input slightly to not change the probabilities seems like the best way to preserve the meaning of the test. llvm-svn: 234979
2015-04-15 14:24:07 +08:00
define i32 @zero3(i32 %i, i32 %a, i32 %b) {
; CHECK: Printing analysis {{.*}} for function 'zero3'
entry:
; AND'ing with a single bit bitmask essentially leads to a bool comparison,
; meaning we don't have probability information.
%and = and i32 %i, 2
%tobool = icmp eq i32 %and, 0
br i1 %tobool, label %then, label %else
Use fixed-point representation for BranchProbability. BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change: Pros: 1. It uses only a half space of the current one. 2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer. Cons: 1. Constructing a probability using arbitrary numerator and denominator needs additional calculations. 2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio. One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern. Differential revision: http://reviews.llvm.org/D12603 llvm-svn: 248633
2015-09-26 07:09:59 +08:00
; CHECK: edge entry -> then probability is 0x40000000 / 0x80000000 = 50.00%
; CHECK: edge entry -> else probability is 0x40000000 / 0x80000000 = 50.00%
Re-apply r234898 and fix tests. This commit makes LLVM not estimate branch probabilities when doing a single bit bitmask tests. The code that originally made me discover this is: if ((a & 0x1) == 0x1) { .. } In this case we don't actually have any branch probability information and should not assume to have any. LLVM transforms this into: %and = and i32 %a, 1 %tobool = icmp eq i32 %and, 0 So, in this case, the result of a bitwise and is compared against 0, but nevertheless, we should not assume to have probability information. CodeGen/ARM/2013-10-11-select-stalls.ll started failing because the changed probabilities changed the results of ARMBaseInstrInfo::isProfitableToIfCvt() and led to an Ifcvt of the diamond in the test. AFAICT, the test was never meant to test this and thus changing the test input slightly to not change the probabilities seems like the best way to preserve the meaning of the test. llvm-svn: 234979
2015-04-15 14:24:07 +08:00
then:
; AND'ing with other bitmask might be something else, so we still assume the
; usual probabilities.
%and2 = and i32 %i, 5
%tobool2 = icmp eq i32 %and2, 0
br i1 %tobool2, label %else, label %exit
Use fixed-point representation for BranchProbability. BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change: Pros: 1. It uses only a half space of the current one. 2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer. Cons: 1. Constructing a probability using arbitrary numerator and denominator needs additional calculations. 2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio. One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern. Differential revision: http://reviews.llvm.org/D12603 llvm-svn: 248633
2015-09-26 07:09:59 +08:00
; CHECK: edge then -> else probability is 0x30000000 / 0x80000000 = 37.50%
; CHECK: edge then -> exit probability is 0x50000000 / 0x80000000 = 62.50%
Re-apply r234898 and fix tests. This commit makes LLVM not estimate branch probabilities when doing a single bit bitmask tests. The code that originally made me discover this is: if ((a & 0x1) == 0x1) { .. } In this case we don't actually have any branch probability information and should not assume to have any. LLVM transforms this into: %and = and i32 %a, 1 %tobool = icmp eq i32 %and, 0 So, in this case, the result of a bitwise and is compared against 0, but nevertheless, we should not assume to have probability information. CodeGen/ARM/2013-10-11-select-stalls.ll started failing because the changed probabilities changed the results of ARMBaseInstrInfo::isProfitableToIfCvt() and led to an Ifcvt of the diamond in the test. AFAICT, the test was never meant to test this and thus changing the test input slightly to not change the probabilities seems like the best way to preserve the meaning of the test. llvm-svn: 234979
2015-04-15 14:24:07 +08:00
else:
br label %exit
exit:
%result = phi i32 [ %a, %then ], [ %b, %else ]
ret i32 %result
}
[BPI] Use metadata info before any other heuristics Metadata potentially is more precise than any heuristics we use, so it makes sense to use first metadata info if it is available. However it makes sense to examine it against other strong heuristics like unreachable one. If edge coming to unreachable block has higher probability then it is expected by unreachable heuristic then we use heuristic and remaining probability is distributed among other reachable blocks equally. An example where metadata might be more strong then unreachable heuristic is as follows: it is possible that there are two branches and for the branch A metadata says that its probability is (0, 2^25). For the branch B the probability is (1, 2^25). So the expectation is that first edge of B is hotter than first edge of A because first edge of A did not executed at least once. If first edge of A points to the unreachable block then using the unreachable heuristics we'll set the probability for A to (1, 2^20) and now edge of A becomes hotter than edge of B. This is unexpected behavior. This fixed the biggest part of https://bugs.llvm.org/show_bug.cgi?id=32214 Reviewers: sanjoy, junbuml, vsk, chandlerc Reviewed By: chandlerc Subscribers: llvm-commits, reames, davidxl Differential Revision: https://reviews.llvm.org/D30631 llvm-svn: 300440
2017-04-17 12:33:04 +08:00
define i32 @test_unreachable_with_prof_greater(i32 %a, i32 %b) {
; CHECK: Printing analysis {{.*}} for function 'test_unreachable_with_prof_greater'
entry:
%cond = icmp eq i32 %a, 42
br i1 %cond, label %exit, label %unr, !prof !4
[BPI] Reduce the probability of unreachable edge to minimal value greater than 0 The probability of edge coming to unreachable block should be as low as possible. The change reduces the probability to minimal value greater than zero. The bug https://bugs.llvm.org/show_bug.cgi?id=32214 show the example when the probability of edge coming to unreachable block is greater than for edge coming to out of the loop and it causes incorrect loop rotation. Please note that with this change the behavior of unreachable heuristic is a bit different than others. Specifically, before this change the sum of probabilities coming to unreachable blocks have the same weight for all branches (it was just split over all edges of this block coming to unreachable blocks). With this change it might be slightly different but not to much due to probability of taken branch to unreachable block is really small. Reviewers: chandlerc, sanjoy, vsk, congh, junbuml, davidxl, dexonsmith Reviewed By: chandlerc, dexonsmith Subscribers: reames, llvm-commits Differential Revision: https://reviews.llvm.org/D30633 llvm-svn: 303327
2017-05-18 14:11:56 +08:00
; CHECK: edge entry -> exit probability is 0x7fffffff / 0x80000000 = 100.00% [HOT edge]
; CHECK: edge entry -> unr probability is 0x00000001 / 0x80000000 = 0.00%
[BPI] Use metadata info before any other heuristics Metadata potentially is more precise than any heuristics we use, so it makes sense to use first metadata info if it is available. However it makes sense to examine it against other strong heuristics like unreachable one. If edge coming to unreachable block has higher probability then it is expected by unreachable heuristic then we use heuristic and remaining probability is distributed among other reachable blocks equally. An example where metadata might be more strong then unreachable heuristic is as follows: it is possible that there are two branches and for the branch A metadata says that its probability is (0, 2^25). For the branch B the probability is (1, 2^25). So the expectation is that first edge of B is hotter than first edge of A because first edge of A did not executed at least once. If first edge of A points to the unreachable block then using the unreachable heuristics we'll set the probability for A to (1, 2^20) and now edge of A becomes hotter than edge of B. This is unexpected behavior. This fixed the biggest part of https://bugs.llvm.org/show_bug.cgi?id=32214 Reviewers: sanjoy, junbuml, vsk, chandlerc Reviewed By: chandlerc Subscribers: llvm-commits, reames, davidxl Differential Revision: https://reviews.llvm.org/D30631 llvm-svn: 300440
2017-04-17 12:33:04 +08:00
unr:
unreachable
exit:
ret i32 %b
}
!4 = !{!"branch_weights", i32 0, i32 1}
define i32 @test_unreachable_with_prof_equal(i32 %a, i32 %b) {
; CHECK: Printing analysis {{.*}} for function 'test_unreachable_with_prof_equal'
entry:
%cond = icmp eq i32 %a, 42
br i1 %cond, label %exit, label %unr, !prof !5
[BPI] Reduce the probability of unreachable edge to minimal value greater than 0 The probability of edge coming to unreachable block should be as low as possible. The change reduces the probability to minimal value greater than zero. The bug https://bugs.llvm.org/show_bug.cgi?id=32214 show the example when the probability of edge coming to unreachable block is greater than for edge coming to out of the loop and it causes incorrect loop rotation. Please note that with this change the behavior of unreachable heuristic is a bit different than others. Specifically, before this change the sum of probabilities coming to unreachable blocks have the same weight for all branches (it was just split over all edges of this block coming to unreachable blocks). With this change it might be slightly different but not to much due to probability of taken branch to unreachable block is really small. Reviewers: chandlerc, sanjoy, vsk, congh, junbuml, davidxl, dexonsmith Reviewed By: chandlerc, dexonsmith Subscribers: reames, llvm-commits Differential Revision: https://reviews.llvm.org/D30633 llvm-svn: 303327
2017-05-18 14:11:56 +08:00
; CHECK: edge entry -> exit probability is 0x7fffffff / 0x80000000 = 100.00% [HOT edge]
; CHECK: edge entry -> unr probability is 0x00000001 / 0x80000000 = 0.00%
[BPI] Use metadata info before any other heuristics Metadata potentially is more precise than any heuristics we use, so it makes sense to use first metadata info if it is available. However it makes sense to examine it against other strong heuristics like unreachable one. If edge coming to unreachable block has higher probability then it is expected by unreachable heuristic then we use heuristic and remaining probability is distributed among other reachable blocks equally. An example where metadata might be more strong then unreachable heuristic is as follows: it is possible that there are two branches and for the branch A metadata says that its probability is (0, 2^25). For the branch B the probability is (1, 2^25). So the expectation is that first edge of B is hotter than first edge of A because first edge of A did not executed at least once. If first edge of A points to the unreachable block then using the unreachable heuristics we'll set the probability for A to (1, 2^20) and now edge of A becomes hotter than edge of B. This is unexpected behavior. This fixed the biggest part of https://bugs.llvm.org/show_bug.cgi?id=32214 Reviewers: sanjoy, junbuml, vsk, chandlerc Reviewed By: chandlerc Subscribers: llvm-commits, reames, davidxl Differential Revision: https://reviews.llvm.org/D30631 llvm-svn: 300440
2017-04-17 12:33:04 +08:00
unr:
unreachable
exit:
ret i32 %b
}
[BPI] Reduce the probability of unreachable edge to minimal value greater than 0 The probability of edge coming to unreachable block should be as low as possible. The change reduces the probability to minimal value greater than zero. The bug https://bugs.llvm.org/show_bug.cgi?id=32214 show the example when the probability of edge coming to unreachable block is greater than for edge coming to out of the loop and it causes incorrect loop rotation. Please note that with this change the behavior of unreachable heuristic is a bit different than others. Specifically, before this change the sum of probabilities coming to unreachable blocks have the same weight for all branches (it was just split over all edges of this block coming to unreachable blocks). With this change it might be slightly different but not to much due to probability of taken branch to unreachable block is really small. Reviewers: chandlerc, sanjoy, vsk, congh, junbuml, davidxl, dexonsmith Reviewed By: chandlerc, dexonsmith Subscribers: reames, llvm-commits Differential Revision: https://reviews.llvm.org/D30633 llvm-svn: 303327
2017-05-18 14:11:56 +08:00
!5 = !{!"branch_weights", i32 2147483647, i32 1}
[BPI] Use metadata info before any other heuristics Metadata potentially is more precise than any heuristics we use, so it makes sense to use first metadata info if it is available. However it makes sense to examine it against other strong heuristics like unreachable one. If edge coming to unreachable block has higher probability then it is expected by unreachable heuristic then we use heuristic and remaining probability is distributed among other reachable blocks equally. An example where metadata might be more strong then unreachable heuristic is as follows: it is possible that there are two branches and for the branch A metadata says that its probability is (0, 2^25). For the branch B the probability is (1, 2^25). So the expectation is that first edge of B is hotter than first edge of A because first edge of A did not executed at least once. If first edge of A points to the unreachable block then using the unreachable heuristics we'll set the probability for A to (1, 2^20) and now edge of A becomes hotter than edge of B. This is unexpected behavior. This fixed the biggest part of https://bugs.llvm.org/show_bug.cgi?id=32214 Reviewers: sanjoy, junbuml, vsk, chandlerc Reviewed By: chandlerc Subscribers: llvm-commits, reames, davidxl Differential Revision: https://reviews.llvm.org/D30631 llvm-svn: 300440
2017-04-17 12:33:04 +08:00
define i32 @test_unreachable_with_prof_zero(i32 %a, i32 %b) {
; CHECK: Printing analysis {{.*}} for function 'test_unreachable_with_prof_zero'
entry:
%cond = icmp eq i32 %a, 42
br i1 %cond, label %exit, label %unr, !prof !6
[BPI] Reduce the probability of unreachable edge to minimal value greater than 0 The probability of edge coming to unreachable block should be as low as possible. The change reduces the probability to minimal value greater than zero. The bug https://bugs.llvm.org/show_bug.cgi?id=32214 show the example when the probability of edge coming to unreachable block is greater than for edge coming to out of the loop and it causes incorrect loop rotation. Please note that with this change the behavior of unreachable heuristic is a bit different than others. Specifically, before this change the sum of probabilities coming to unreachable blocks have the same weight for all branches (it was just split over all edges of this block coming to unreachable blocks). With this change it might be slightly different but not to much due to probability of taken branch to unreachable block is really small. Reviewers: chandlerc, sanjoy, vsk, congh, junbuml, davidxl, dexonsmith Reviewed By: chandlerc, dexonsmith Subscribers: reames, llvm-commits Differential Revision: https://reviews.llvm.org/D30633 llvm-svn: 303327
2017-05-18 14:11:56 +08:00
; CHECK: edge entry -> exit probability is 0x7fffffff / 0x80000000 = 100.00% [HOT edge]
; CHECK: edge entry -> unr probability is 0x00000001 / 0x80000000 = 0.00%
[BPI] Use metadata info before any other heuristics Metadata potentially is more precise than any heuristics we use, so it makes sense to use first metadata info if it is available. However it makes sense to examine it against other strong heuristics like unreachable one. If edge coming to unreachable block has higher probability then it is expected by unreachable heuristic then we use heuristic and remaining probability is distributed among other reachable blocks equally. An example where metadata might be more strong then unreachable heuristic is as follows: it is possible that there are two branches and for the branch A metadata says that its probability is (0, 2^25). For the branch B the probability is (1, 2^25). So the expectation is that first edge of B is hotter than first edge of A because first edge of A did not executed at least once. If first edge of A points to the unreachable block then using the unreachable heuristics we'll set the probability for A to (1, 2^20) and now edge of A becomes hotter than edge of B. This is unexpected behavior. This fixed the biggest part of https://bugs.llvm.org/show_bug.cgi?id=32214 Reviewers: sanjoy, junbuml, vsk, chandlerc Reviewed By: chandlerc Subscribers: llvm-commits, reames, davidxl Differential Revision: https://reviews.llvm.org/D30631 llvm-svn: 300440
2017-04-17 12:33:04 +08:00
unr:
unreachable
exit:
ret i32 %b
}
!6 = !{!"branch_weights", i32 0, i32 0}
define i32 @test_unreachable_with_prof_less(i32 %a, i32 %b) {
; CHECK: Printing analysis {{.*}} for function 'test_unreachable_with_prof_less'
entry:
%cond = icmp eq i32 %a, 42
br i1 %cond, label %exit, label %unr, !prof !7
; CHECK: edge entry -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
; CHECK: edge entry -> unr probability is 0x00000000 / 0x80000000 = 0.00%
unr:
unreachable
exit:
ret i32 %b
}
!7 = !{!"branch_weights", i32 1, i32 0}
define i32 @test_unreachable_with_switch_prof1(i32 %i, i32 %a, i32 %b, i32 %c, i32 %d, i32 %e) {
; CHECK: Printing analysis {{.*}} for function 'test_unreachable_with_switch_prof1'
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 !8
[BPI] Reduce the probability of unreachable edge to minimal value greater than 0 The probability of edge coming to unreachable block should be as low as possible. The change reduces the probability to minimal value greater than zero. The bug https://bugs.llvm.org/show_bug.cgi?id=32214 show the example when the probability of edge coming to unreachable block is greater than for edge coming to out of the loop and it causes incorrect loop rotation. Please note that with this change the behavior of unreachable heuristic is a bit different than others. Specifically, before this change the sum of probabilities coming to unreachable blocks have the same weight for all branches (it was just split over all edges of this block coming to unreachable blocks). With this change it might be slightly different but not to much due to probability of taken branch to unreachable block is really small. Reviewers: chandlerc, sanjoy, vsk, congh, junbuml, davidxl, dexonsmith Reviewed By: chandlerc, dexonsmith Subscribers: reames, llvm-commits Differential Revision: https://reviews.llvm.org/D30633 llvm-svn: 303327
2017-05-18 14:11:56 +08:00
; CHECK: edge entry -> case_a probability is 0x00000001 / 0x80000000 = 0.00%
; CHECK: edge entry -> case_b probability is 0x07ffffff / 0x80000000 = 6.25%
; CHECK: edge entry -> case_c probability is 0x67ffffff / 0x80000000 = 81.25% [HOT edge]
; CHECK: edge entry -> case_d probability is 0x07ffffff / 0x80000000 = 6.25%
; CHECK: edge entry -> case_e probability is 0x07ffffff / 0x80000000 = 6.25%
[BPI] Use metadata info before any other heuristics Metadata potentially is more precise than any heuristics we use, so it makes sense to use first metadata info if it is available. However it makes sense to examine it against other strong heuristics like unreachable one. If edge coming to unreachable block has higher probability then it is expected by unreachable heuristic then we use heuristic and remaining probability is distributed among other reachable blocks equally. An example where metadata might be more strong then unreachable heuristic is as follows: it is possible that there are two branches and for the branch A metadata says that its probability is (0, 2^25). For the branch B the probability is (1, 2^25). So the expectation is that first edge of B is hotter than first edge of A because first edge of A did not executed at least once. If first edge of A points to the unreachable block then using the unreachable heuristics we'll set the probability for A to (1, 2^20) and now edge of A becomes hotter than edge of B. This is unexpected behavior. This fixed the biggest part of https://bugs.llvm.org/show_bug.cgi?id=32214 Reviewers: sanjoy, junbuml, vsk, chandlerc Reviewed By: chandlerc Subscribers: llvm-commits, reames, davidxl Differential Revision: https://reviews.llvm.org/D30631 llvm-svn: 300440
2017-04-17 12:33:04 +08:00
case_a:
unreachable
case_b:
br label %exit
; CHECK: edge case_b -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
case_c:
br label %exit
; CHECK: edge case_c -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
case_d:
br label %exit
; CHECK: edge case_d -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
case_e:
br label %exit
; CHECK: edge case_e -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
exit:
%result = phi i32 [ %b, %case_b ],
[ %c, %case_c ],
[ %d, %case_d ],
[ %e, %case_e ]
ret i32 %result
}
!8 = !{!"branch_weights", i32 4, i32 4, i32 64, i32 4, i32 4}
define i32 @test_unreachable_with_switch_prof2(i32 %i, i32 %a, i32 %b, i32 %c, i32 %d, i32 %e) {
; CHECK: Printing analysis {{.*}} for function 'test_unreachable_with_switch_prof2'
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 !9
[BPI] Reduce the probability of unreachable edge to minimal value greater than 0 The probability of edge coming to unreachable block should be as low as possible. The change reduces the probability to minimal value greater than zero. The bug https://bugs.llvm.org/show_bug.cgi?id=32214 show the example when the probability of edge coming to unreachable block is greater than for edge coming to out of the loop and it causes incorrect loop rotation. Please note that with this change the behavior of unreachable heuristic is a bit different than others. Specifically, before this change the sum of probabilities coming to unreachable blocks have the same weight for all branches (it was just split over all edges of this block coming to unreachable blocks). With this change it might be slightly different but not to much due to probability of taken branch to unreachable block is really small. Reviewers: chandlerc, sanjoy, vsk, congh, junbuml, davidxl, dexonsmith Reviewed By: chandlerc, dexonsmith Subscribers: reames, llvm-commits Differential Revision: https://reviews.llvm.org/D30633 llvm-svn: 303327
2017-05-18 14:11:56 +08:00
; CHECK: edge entry -> case_a probability is 0x00000001 / 0x80000000 = 0.00%
; CHECK: edge entry -> case_b probability is 0x00000001 / 0x80000000 = 0.00%
; CHECK: edge entry -> case_c probability is 0x6aaaaaa9 / 0x80000000 = 83.33% [HOT edge]
; CHECK: edge entry -> case_d probability is 0x0aaaaaa9 / 0x80000000 = 8.33%
; CHECK: edge entry -> case_e probability is 0x0aaaaaa9 / 0x80000000 = 8.33%
[BPI] Use metadata info before any other heuristics Metadata potentially is more precise than any heuristics we use, so it makes sense to use first metadata info if it is available. However it makes sense to examine it against other strong heuristics like unreachable one. If edge coming to unreachable block has higher probability then it is expected by unreachable heuristic then we use heuristic and remaining probability is distributed among other reachable blocks equally. An example where metadata might be more strong then unreachable heuristic is as follows: it is possible that there are two branches and for the branch A metadata says that its probability is (0, 2^25). For the branch B the probability is (1, 2^25). So the expectation is that first edge of B is hotter than first edge of A because first edge of A did not executed at least once. If first edge of A points to the unreachable block then using the unreachable heuristics we'll set the probability for A to (1, 2^20) and now edge of A becomes hotter than edge of B. This is unexpected behavior. This fixed the biggest part of https://bugs.llvm.org/show_bug.cgi?id=32214 Reviewers: sanjoy, junbuml, vsk, chandlerc Reviewed By: chandlerc Subscribers: llvm-commits, reames, davidxl Differential Revision: https://reviews.llvm.org/D30631 llvm-svn: 300440
2017-04-17 12:33:04 +08:00
case_a:
unreachable
case_b:
unreachable
case_c:
br label %exit
; CHECK: edge case_c -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
case_d:
br label %exit
; CHECK: edge case_d -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
case_e:
br label %exit
; CHECK: edge case_e -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
exit:
%result = phi i32 [ %c, %case_c ],
[ %d, %case_d ],
[ %e, %case_e ]
ret i32 %result
}
!9 = !{!"branch_weights", i32 4, i32 4, i32 64, i32 4, i32 4}
define i32 @test_unreachable_with_switch_prof3(i32 %i, i32 %a, i32 %b, i32 %c, i32 %d, i32 %e) {
; CHECK: Printing analysis {{.*}} for function 'test_unreachable_with_switch_prof3'
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 !10
; CHECK: edge entry -> case_a probability is 0x00000000 / 0x80000000 = 0.00%
[BPI] Reduce the probability of unreachable edge to minimal value greater than 0 The probability of edge coming to unreachable block should be as low as possible. The change reduces the probability to minimal value greater than zero. The bug https://bugs.llvm.org/show_bug.cgi?id=32214 show the example when the probability of edge coming to unreachable block is greater than for edge coming to out of the loop and it causes incorrect loop rotation. Please note that with this change the behavior of unreachable heuristic is a bit different than others. Specifically, before this change the sum of probabilities coming to unreachable blocks have the same weight for all branches (it was just split over all edges of this block coming to unreachable blocks). With this change it might be slightly different but not to much due to probability of taken branch to unreachable block is really small. Reviewers: chandlerc, sanjoy, vsk, congh, junbuml, davidxl, dexonsmith Reviewed By: chandlerc, dexonsmith Subscribers: reames, llvm-commits Differential Revision: https://reviews.llvm.org/D30633 llvm-svn: 303327
2017-05-18 14:11:56 +08:00
; CHECK: edge entry -> case_b probability is 0x00000001 / 0x80000000 = 0.00%
; CHECK: edge entry -> case_c probability is 0x6e08fb82 / 0x80000000 = 85.96% [HOT edge]
; CHECK: edge entry -> case_d probability is 0x08fb823e / 0x80000000 = 7.02%
; CHECK: edge entry -> case_e probability is 0x08fb823e / 0x80000000 = 7.02%
[BPI] Use metadata info before any other heuristics Metadata potentially is more precise than any heuristics we use, so it makes sense to use first metadata info if it is available. However it makes sense to examine it against other strong heuristics like unreachable one. If edge coming to unreachable block has higher probability then it is expected by unreachable heuristic then we use heuristic and remaining probability is distributed among other reachable blocks equally. An example where metadata might be more strong then unreachable heuristic is as follows: it is possible that there are two branches and for the branch A metadata says that its probability is (0, 2^25). For the branch B the probability is (1, 2^25). So the expectation is that first edge of B is hotter than first edge of A because first edge of A did not executed at least once. If first edge of A points to the unreachable block then using the unreachable heuristics we'll set the probability for A to (1, 2^20) and now edge of A becomes hotter than edge of B. This is unexpected behavior. This fixed the biggest part of https://bugs.llvm.org/show_bug.cgi?id=32214 Reviewers: sanjoy, junbuml, vsk, chandlerc Reviewed By: chandlerc Subscribers: llvm-commits, reames, davidxl Differential Revision: https://reviews.llvm.org/D30631 llvm-svn: 300440
2017-04-17 12:33:04 +08:00
case_a:
unreachable
case_b:
unreachable
case_c:
br label %exit
; CHECK: edge case_c -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
case_d:
br label %exit
; CHECK: edge case_d -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
case_e:
br label %exit
; CHECK: edge case_e -> exit probability is 0x80000000 / 0x80000000 = 100.00% [HOT edge]
exit:
%result = phi i32 [ %c, %case_c ],
[ %d, %case_d ],
[ %e, %case_e ]
ret i32 %result
}
!10 = !{!"branch_weights", i32 0, i32 4, i32 64, i32 4, i32 4}
define i32 @test_unreachable_with_switch_prof4(i32 %i, i32 %a, i32 %b, i32 %c, i32 %d, i32 %e) {
; CHECK: Printing analysis {{.*}} for function 'test_unreachable_with_switch_prof4'
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 !11
; CHECK: edge entry -> case_a probability is 0x1999999a / 0x80000000 = 20.00%
; CHECK: edge entry -> case_b probability is 0x1999999a / 0x80000000 = 20.00%
; CHECK: edge entry -> case_c probability is 0x1999999a / 0x80000000 = 20.00%
; CHECK: edge entry -> case_d probability is 0x1999999a / 0x80000000 = 20.00%
; CHECK: edge entry -> case_e probability is 0x1999999a / 0x80000000 = 20.00%
case_a:
unreachable
case_b:
unreachable
case_c:
unreachable
case_d:
unreachable
case_e:
unreachable
}
!11 = !{!"branch_weights", i32 0, i32 4, i32 64, i32 4, i32 4}