[all-commits] [llvm/llvm-project] 5c0d61: [LTO] Reduce memory usage for import lists (#106772)

[Kazu Hirata via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 5c0d61e318a77434487fcec8361d8110fb06e59d
      https://github.com/llvm/llvm-project/commit/5c0d61e318a77434487fcec8361d8110fb06e59d
  Author: Kazu Hirata <kazu at google.com>
  Date:   2024-09-01 (Sun, 01 Sep 2024)

  Changed paths:
    M clang/lib/CodeGen/BackendUtil.cpp
    M llvm/include/llvm/Transforms/IPO/FunctionImport.h
    M llvm/lib/LTO/LTO.cpp
    M llvm/lib/Transforms/IPO/FunctionImport.cpp
    M llvm/tools/llvm-link/llvm-link.cpp

  Log Message:
  -----------
  [LTO] Reduce memory usage for import lists (#106772)

This patch reduces the memory usage for import lists by employing
memory-efficient data structures.

With this patch, an import list for a given destination module is
basically DenseSet<uint32_t> with each element indexing into the
deduplication table containing tuples of:

  {SourceModule, GUID, Definition/Declaration}

In one of our large applications, the peak memory usage goes down by
9.2% from 6.120GB to 5.555GB during the LTO indexing step.

This patch addresses several sources of space inefficiency associated
with std::unordered_map:

- std::unordered_map<GUID, ImportKind> takes up 16 bytes because of
  padding even though ImportKind only carries one bit of information.

- std::unordered_map uses pointers to elements, both in the hash table
  proper and for collision chains.

- We allocate an instance of std::unordered_map for each
  {Destination Module, Source Module} pair for which we have at least
  one import.  Most import lists have less than 10 imports, so the
  metadata like the size of std::unordered_map and the pointer to the
  hash table costs a lot relative to the actual contents.



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications