[BOLT] Fix counts aggregation in merge-fdata (#119652)

merge-fdata used to consider misprediction count as part of "signature",
or the aggregation key. This prevented it from collapsing profile lin

[BOLT] Fix counts aggregation in merge-fdata (#119652)

merge-fdata used to consider misprediction count as part of "signature",
or the aggregation key. This prevented it from collapsing profile lines
with different misprediction counts, which resulted in duplicate
`(from, to)` pairs with different misprediction and execution counts.

Fix that by splitting out misprediction count and accumulating it
separately.

Test Plan: updated bolt/test/merge-fdata-lbr-mode.test

show more ...

[BOLT][NFC] Speedup merge-fdata (#119942)

Eliminate splitting the buffer into lines, and use `std::getline`
directly. Simplify no_lbr and boltedcollection handling as well.

Test Plan: For a larg

[BOLT][NFC] Speedup merge-fdata (#119942)

Eliminate splitting the buffer into lines, and use `std::getline`
directly. Simplify no_lbr and boltedcollection handling as well.

Test Plan: For a large fdata file (200MB), fstream version is ~10%
faster.

show more ...

[BOLT][merge-fdata] Fix basic sample profile aggregation without LBR info (#118481)

When a basic sample profile is gathered without LBR info, the generated
profile contains a "no-lbr" tag in the fi

[BOLT][merge-fdata] Fix basic sample profile aggregation without LBR info (#118481)

When a basic sample profile is gathered without LBR info, the generated
profile contains a "no-lbr" tag in the first line of the fdata file.
This PR fixes merge-fdata to recognize and save this tag to the output
file.

show more ...

[BOLT][merge-fdata] Initialize YAML profile header (#109613)

While merging profiles, some fields in the input header, e.g.
HashFunction, could be uninitialized leading to a UMR. Initialize merged

[BOLT][merge-fdata] Initialize YAML profile header (#109613)

While merging profiles, some fields in the input header, e.g.
HashFunction, could be uninitialized leading to a UMR. Initialize merged
header with the first input header.

Fixes #109592

show more ...

[BOLT][YAML] Allow unknown keys in the input (#100824)

This ensures forward compatibility, where old BOLT versions can consume
the profile created by newer versions with extra keys.

Test Plan: a

[BOLT][YAML] Allow unknown keys in the input (#100824)

This ensures forward compatibility, where old BOLT versions can consume
the profile created by newer versions with extra keys.

Test Plan: added yaml-unknown-keys.test

show more ...

Rename llvm::ThreadPool -> llvm::DefaultThreadPool (NFC) (#83702)

The base class llvm::ThreadPoolInterface will be renamed
llvm::ThreadPool in a subsequent commit.

This is a breaking change: cli

Rename llvm::ThreadPool -> llvm::DefaultThreadPool (NFC) (#83702)

The base class llvm::ThreadPoolInterface will be renamed
llvm::ThreadPool in a subsequent commit.

This is a breaking change: clients who use to create a ThreadPool must
now create a DefaultThreadPool instead.

show more ...

Rename `ThreadPool::getThreadCount()` to `getMaxConcurrency()` (NFC) (#82296)

This is addressing a long-time TODO to rename this misleading API. The
old one is preserved for now but marked deprecat

Rename `ThreadPool::getThreadCount()` to `getMaxConcurrency()` (NFC) (#82296)

This is addressing a long-time TODO to rename this misleading API. The
old one is preserved for now but marked deprecated.

show more ...

[BOLT] Fix unconditional output of boltedcollection in merge-fdata (#78653)

Fix the bug where merge-fdata unconditionally outputs boltedcollection
line, regardless of whether input files have it set

[BOLT] Fix unconditional output of boltedcollection in merge-fdata (#78653)

Fix the bug where merge-fdata unconditionally outputs boltedcollection
line, regardless of whether input files have it set.

Test Plan:
Added bolt/test/X86/merge-fdata-nobat-mode.test which fails without this
fix.

show more ...

Revert "[BOLT] Fix unconditional output of boltedcollection in merge-fdata (#78653)"

This reverts commit 82bc33ea3f1a539be50ed46919dc53fc6b685da9.

Accidentally pushed unrelated changes.

[BOLT] Fix unconditional output of boltedcollection in merge-fdata (#78653)

Fix the bug where merge-fdata unconditionally outputs boltedcollection
line, regardless of whether input files have it s

[BOLT] Fix unconditional output of boltedcollection in merge-fdata (#78653)

Fix the bug where merge-fdata unconditionally outputs boltedcollection
line, regardless of whether input files have it set.

Test Plan:
Added bolt/test/X86/merge-fdata-nobat-mode.test which fails without this
fix.

show more ...

[BOLT] Use StringRef::{starts,ends}_with (NFC)

This patch replaces uses of StringRef::{starts,ends}with with
StringRef::{starts,ends}_with for consistency with
std::{string,string_view}::{starts,end

[BOLT] Use StringRef::{starts,ends}_with (NFC)

This patch replaces uses of StringRef::{starts,ends}with with
StringRef::{starts,ends}_with for consistency with
std::{string,string_view}::{starts,ends}_with in C++20.

I'm planning to deprecate and eventually remove
StringRef::{starts,ends}with.

show more ...

Reland^2 "[BOLT] Parallelize legacy profile merging"

Resovled the issue that when number of tasks is fewer than cores, we end
up creating as many threads as the number of cores, making the
performan

Reland^2 "[BOLT] Parallelize legacy profile merging"

Resovled the issue that when number of tasks is fewer than cores, we end
up creating as many threads as the number of cores, making the
performance worse than the single thread version.

show more ...

Revert "Reland "[BOLT] Parallelize legacy profile merging""

This reverts commit 611fb179b19857ffb87df81c926902fc7e3412ab.

Broken tests

Reland "[BOLT] Parallelize legacy profile merging"

This reverts commit 78d8d016490909ac759c6f76c5f8679bc7a58b2e.

Revert "[BOLT] Parallelize legacy profile merging"

This reverts commit 35af20d9e036deeed250b73fd3ae86d6455173c5.

The patch caused a test failure.

[BOLT] Parallelize legacy profile merging

Merging profiles is quite expensive, but easily paralleizable.

8359 profiles on n2d-standard-128:
single-thread: 808s
multi-thread: 200s (~75% speed up)

D

[BOLT] Parallelize legacy profile merging

Merging profiles is quite expensive, but easily paralleizable.

8359 profiles on n2d-standard-128:
single-thread: 808s
multi-thread: 200s (~75% speed up)

Differential Revision: https://reviews.llvm.org/D149014

show more ...

[BOLT][NFC] Simplify code using std::optional

Use std::optional instead of tracking if it is the first profile seen.

Differential Revision: https://reviews.llvm.org/D147308

[BOLT][NFC] Use llvm::make_second_range

Reviewed By: #bolt, rafauler

Differential Revision: https://reviews.llvm.org/D143019

[BOLT][NFC] Use range-based STL wrappers

Replace `std::` algorithms taking begin/end iterators with `llvm::` counterparts
accepting ranges.

Reviewed By: rafauler

Differential Revision: https://rev

[BOLT][NFC] Use range-based STL wrappers

Replace `std::` algorithms taking begin/end iterators with `llvm::` counterparts
accepting ranges.

Reviewed By: rafauler

Differential Revision: https://reviews.llvm.org/D128154

show more ...

[BOLT] Add `-o` option to merge-fdata

Differential Revision: https://reviews.llvm.org/D126788

[BOLT] Allow merge-fdata to take a directory as input

and recursively merge all files under said directory. This is similar
to `llvm-profdata merge`.

Differential Revision: https://reviews.llvm.org

[BOLT] Allow merge-fdata to take a directory as input

and recursively merge all files under said directory. This is similar
to `llvm-profdata merge`.

Differential Revision: https://reviews.llvm.org/D126695

show more ...

[BOLT][NFC] Don't over-specify the size of SmallVector

This is the recommended way, should make merging profiles ever so
slightly faster.

[BOLT] Fix merge-fdata handling of BAT profiles

When a profile is collected in a BOLTed binary, the generated
profile is tagged with a header string "boltedcollection" in the first
line of the fdata

[BOLT] Fix merge-fdata handling of BAT profiles

When a profile is collected in a BOLTed binary, the generated
profile is tagged with a header string "boltedcollection" in the first
line of the fdata file. Fix merge-fdata to recognize this header
string and preserve it into the output.

Reviewed By: Amir

Differential Revision: https://reviews.llvm.org/D125591

show more ...

[BOLT] Compact legacy profiles

Merging multiple legacy profiles (produced by instrumentation BOLT) can
easily reach GiBs. Let merge-fdata compact the profiles during merge to
significantly reduce sp

[BOLT] Compact legacy profiles

Merging multiple legacy profiles (produced by instrumentation BOLT) can
easily reach GiBs. Let merge-fdata compact the profiles during merge to
significantly reduce space usage.

Differential Revision: https://reviews.llvm.org/D123513

show more ...

[BOLT][NFC] Fix braces usage in the rest of the codebase

Summary:
Refactor remaining bolt sources to follow the braces rule for if/else/loop from
[LLVM Coding Standards](https://llvm.org/docs/Coding

[BOLT][NFC] Fix braces usage in the rest of the codebase

Summary:
Refactor remaining bolt sources to follow the braces rule for if/else/loop from
[LLVM Coding Standards](https://llvm.org/docs/CodingStandards.html).

(cherry picked from FBD33345885)

show more ...