16777255093

Committed 06 Aug 2025 12:43PM UTC coverage: 84.036% (+0.007%) from 84.029%

Build # 16777255093

Build Type

Pull #4138

github

Committed by

web-flow

Commit Message

Merge c628d6ed8 into 773d4ec96

Pull Request Pull Request #4138: fix: `into_iter_multithread` buffered futures usage

Run Details

0 of 4 new or added lines in 1 file covered. (0.0%)

1 existing line in 1 file now uncovered.

48340 of 57523 relevant lines covered (84.04%)

520681.07 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

0.0

/vortex-scan/src/multi_thread.rs

// SPDX-License-Identifier: Apache-2.0
// SPDX-FileCopyrightText: Copyright the Vortex contributors

use std::iter;
use std::sync::{Arc, LazyLock};

use futures::{StreamExt, stream};
use tokio::runtime::{Builder, Runtime};
use vortex_array::ArrayRef;
use vortex_array::iter::{ArrayIterator, ArrayIteratorAdapter};
use vortex_error::{VortexExpect, VortexResult, vortex_err};

use crate::ScanBuilder;

/// We create an internal Tokio runtime used exclusively for orchestrating work-stealing
/// of CPU-bound work for multithreaded scans.
///
/// It is intentionally not exposed to the user, not configurable, and does not enable I/O or
/// timers.
static CPU_RUNTIME: LazyLock<Runtime> = LazyLock::new(|| {
    Builder::new_multi_thread()
        .thread_name("vortex-multithread-scan")
        .build()
        .vortex_expect("Failed to create a new Tokio runtime")
});

impl ScanBuilder<ArrayRef> {
    /// Execute the scan on multiple worker threads.
    pub fn into_array_iter_multithread(self) -> VortexResult<impl ArrayIterator + Send + 'static> {
        let dtype = self.dtype()?;
        Ok(ArrayIteratorAdapter::new(
            dtype,
            self.into_iter_multithread(|a| a)?,
        ))
    }

    /// Execute the scan on multiple worker threads.
    ///
    /// A `map_fn` can be passed to further transform the results of the scan while still running
    /// on the thread pool.
    pub fn into_iter_multithread<T, F>(
        self,
        map_fn: F,
    ) -> VortexResult<impl Iterator<Item = T> + Send + 'static>
    where
        T: 'static + Send,
        F: Fn(VortexResult<ArrayRef>) -> T + Send + Sync + 'static,
    {
        let concurrency = self.concurrency;
        let num_workers = CPU_RUNTIME.metrics().num_workers();

        let tasks = self.build()?;
        // We need to clone and send the map_fn into each task.
        let map_fn = Arc::new(map_fn);

        let mut stream = stream::iter(tasks)
            // TODO(ngates): this is very crude indeed. This buffered call essentially controls how
            //  many splits we have in-flight at any given time. We multiple workers by concurrency
            //  to configure per-thread concurrency, which essentially means each thread can make
            //  progress on one split while waiting for the I/O of another split to complete.
            //  In an ideal world, the number of in-flight tasks would be dynamically adjusted
            //  based on how much I/O the tasks _actually_ require. For example, all pruning tasks
            //  could be spawned immediately since they all use a single segment, this would allow
            //  head-room to run ahead and figure out the I/O demands of subsequent tasks.
            .buffered(num_workers * concurrency);

        Ok(iter::from_fn(move || {
            tokio::task::block_in_place(|| CPU_RUNTIME.handle().block_on(stream.next()))
        })
        .filter_map(move |result| match result {
            Ok(Some(array)) => Some(map_fn(Ok(array))),
            Ok(None) => None,
            Err(e) => Some(map_fn(Err(e))),
        }))
    }
}

1	// SPDX-License-Identifier: Apache-2.0
2	// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4	use std::iter;
5	use std::sync::{Arc, LazyLock};
6
7	use futures::{StreamExt, stream};
8	use tokio::runtime::{Builder, Runtime};
9	use vortex_array::ArrayRef;
10	use vortex_array::iter::{ArrayIterator, ArrayIteratorAdapter};
11	use vortex_error::{VortexExpect, VortexResult, vortex_err};
12
13	use crate::ScanBuilder;
14
15	/// We create an internal Tokio runtime used exclusively for orchestrating work-stealing
16	/// of CPU-bound work for multithreaded scans.
17	///
18	/// It is intentionally not exposed to the user, not configurable, and does not enable I/O or
19	/// timers.
20	static CPU_RUNTIME: LazyLock<Runtime> = LazyLock::new(\|\| {	×
21	Builder::new_multi_thread()	×
22	.thread_name("vortex-multithread-scan")	×
23	.build()	×
24	.vortex_expect("Failed to create a new Tokio runtime")	×
25	});	×
26
27	impl ScanBuilder<ArrayRef> {
28	/// Execute the scan on multiple worker threads.
29	pub fn into_array_iter_multithread(self) -> VortexResult<impl ArrayIterator + Send + 'static> {	×
30	let dtype = self.dtype()?;	×
31	Ok(ArrayIteratorAdapter::new(	×
32	dtype,	×
33	self.into_iter_multithread(\|a\| a)?,	×
34	))
35	}	×
36
37	/// Execute the scan on multiple worker threads.
38	///
39	/// A `map_fn` can be passed to further transform the results of the scan while still running
40	/// on the thread pool.
41	pub fn into_iter_multithread<T, F>(	×
42	self,	×
43	map_fn: F,	×
44	) -> VortexResult<impl Iterator<Item = T> + Send + 'static>	×
45	where	×
46	T: 'static + Send,	×
47	F: Fn(VortexResult<ArrayRef>) -> T + Send + Sync + 'static,	×
48	{
49	let concurrency = self.concurrency;	×
50	let num_workers = CPU_RUNTIME.metrics().num_workers();	×
51
52	let tasks = self.build()?;	×
53	// We need to clone and send the map_fn into each task.
54	let map_fn = Arc::new(map_fn);	×
55
UNCOV 56	let mut stream = stream::iter(tasks)	×
57	// TODO(ngates): this is very crude indeed. This buffered call essentially controls how
58	// many splits we have in-flight at any given time. We multiple workers by concurrency
59	// to configure per-thread concurrency, which essentially means each thread can make
60	// progress on one split while waiting for the I/O of another split to complete.
61	// In an ideal world, the number of in-flight tasks would be dynamically adjusted
62	// based on how much I/O the tasks _actually_ require. For example, all pruning tasks
63	// could be spawned immediately since they all use a single segment, this would allow
64	// head-room to run ahead and figure out the I/O demands of subsequent tasks.
65	.buffered(num_workers * concurrency);	×
66
67	Ok(iter::from_fn(move \|\| {	×
68	tokio::task::block_in_place(\|\| CPU_RUNTIME.handle().block_on(stream.next()))	×
69	})	×
NEW 70	.filter_map(move \|result\| match result {	×
NEW 71	Ok(Some(array)) => Some(map_fn(Ok(array))),	×
NEW 72	Ok(None) => None,	×
NEW 73	Err(e) => Some(map_fn(Err(e))),	×
74	}))	×
75	}	×
76	}

vortex-data / vortex / 16777255093

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous