8000 perf(dep_graph): Avoid allocating a set on when the number reads are … by Marwes · Pull Request #69778 · rust-lang/rust · GitHub
[go: up one dir, main page]
Skip to content

perf(dep_graph): Avoid allocating a set on when the number reads are … #69778

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Mar 23, 2020
Merged

perf(dep_graph): Avoid allocating a set on when the number reads are … #69778

Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
c6f1215
perf(dep_graph): Avoid allocating a set on when the number reads are …
Mar 6, 2020
4168c25
refactor: Add an alias for the SmallVec in dep_graph
Mar 7, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Next Next commit
perf(dep_graph): Avoid allocating a set on when the number reads are … 
…small

`reserve_and_rehash` takes up 1.4% of the runtime on the `packed-simd`
benchmark which I believe is due to the number of reads are very low in
many cases (see #50565 for
instance).

This avoids allocating the set until we start allocating the `reads`
`SmallVec` but it is possible that a lower limit might be better (not
tested since the improvement will be hard to spot either way).
  • Loading branch information
Markus Westerlind committed Mar 6, 2020
commit c6f1215d925561021eec296d9e88aa0305d0a18d
19 changes: 17 additions & 2 deletions src/librustc/dep_graph/graph.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1128,11 +1128,25 @@ impl DepGraphData {
let icx = if let Some(icx) = icx { icx } else { return };
if let Some(task_deps) = icx.task_deps {
let mut task_deps = task_deps.lock();
let task_deps = &mut *task_deps;
if cfg!(debug_assertions) {
self.current.total_read_count.fetch_add(1, Relaxed);
}
if task_deps.read_set.insert(source) {

// As long as we only have a low number of reads we can avoid doing a hash
// insert and potentially allocating/reallocating the hashmap
let new_read = if task_deps.reads.len() < TASK_DEPS_READS_CAP {
task_deps.reads.iter().all(|other| *other != source)
} else {
task_deps.read_set.insert(source)
};
if new_read {
task_deps.reads.push(source);
if task_deps.reads.len() == TASK_DEPS_READS_CAP {
// Fill `read_set` with what we have so far so we can use the hashset next
// time
task_deps.read_set.extend(task_deps.reads.iter().copied());
}

#[cfg(debug_assertions)]
{
Expand All @@ -1154,10 +1168,11 @@ impl DepGraphData {
}
}

const TASK_DEPS_READS_CAP: usize = 8;
pub struct TaskDeps {
#[cfg(debug_assertions)]
node: Option<DepNode>,
reads: SmallVec<[DepNodeIndex; 8]>,
reads: SmallVec<[DepNodeIndex; TASK_DEPS_READS_CAP]>,
read_set: FxHashSet<DepNodeIndex>,
}

Expand Down
0