use num_cpus by default for parallelism

Author: cosmicexplorer

Cargo.toml

@@ -48,6 +48,7 @@ zstd = { version = "0.13", optional = true, default-features = false }
 zopfli = { version = "0.8", optional = true }
 deflate64 = { version = "0.1.9", optional = true }
 lzma-rs = { version = "0.3", default-features = false, optional = true }
+num_cpus = { version = "1.16", optional = true }
 
 [target.'cfg(any(all(target_arch = "arm", target_pointer_width = "32"), target_arch = "mips", target_arch = "powerpc"))'.dependencies]
 crossbeam-utils = "0.8.20"
@@ -83,7 +84,7 @@ deflate-zopfli = ["zopfli", "_deflate-any"]
 lzma = ["lzma-rs/stream"]
 unreserved = []
 xz = ["lzma-rs/raw_decoder"]
-parallelism = ["libc"]
+parallelism = ["libc", "num_cpus"]
 default = [
     "aes-crypto",
     "bzip2",

benches/extract.rs

@@ -12,6 +12,9 @@ use zip::ZipArchive;
 #[cfg(all(feature = "parallelism", feature = "bzip2", unix))]
 use zip::read::{split_extract, ExtractionParameters};
 
+#[cfg(feature = "parallelism")]
+use num_cpus;
+
 /* This archive has a set of entries repeated 20x:
  * - 200K random data, stored uncompressed (CompressionMethod::Stored)
  * - 246K text data (the project gutenberg html version of king lear)
@@ -47,9 +50,6 @@ fn extract_basic(bench: &mut Bencher) {
     });
 }
 
-#[cfg(all(feature = "parallelism", feature = "bzip2", unix))]
-const DECOMPRESSION_THREADS: usize = 8;
-
 #[cfg(all(feature = "parallelism", feature = "bzip2", unix))]
 fn extract_split(bench: &mut Bencher) {
     let readable_archive = get_test_archive().unwrap();
@@ -60,7 +60,7 @@ fn extract_split(bench: &mut Bencher) {
         .unwrap();
 
     let params = ExtractionParameters {
-        decompression_threads: DECOMPRESSION_THREADS,
+        decompression_threads: num_cpus::get(),
         ..Default::default()
     };
 

src/read/pipelining.rs

@@ -601,6 +601,7 @@ pub mod handle_creation {
 #[cfg(unix)]
 pub mod split_extraction {
     use displaydoc::Display;
+    use num_cpus;
     use thiserror::Error;
 
     use std::fs;
@@ -710,7 +711,7 @@ pub mod split_extraction {
     pub struct ExtractionParameters {
         /// Number of threads used for decompression.
         ///
-        /// Default value: 4.
+        /// Default value: number of available cpus via [`num_cpus::get()`].
         ///
         /// Note that multiple times this many threads will be spawned by [`split_extract()`] as
         /// part of the pipelined process. Only this many threads will be used to perform
@@ -748,7 +749,10 @@ pub mod split_extraction {
     impl Default for ExtractionParameters {
         fn default() -> Self {
             Self {
-                decompression_threads: 4,
+                /* NB: this will perform a syscall. We still probably want to call this dynamically
+                 * instead of globally caching the call, in order to respect the dynamic value of
+                 * e.g. sched affinity. */
+                decompression_threads: num_cpus::get(),
                 decompression_copy_buffer_length: 1024 * 1024,
                 file_range_copy_buffer_length: 1024 * 1024,
                 #[cfg(not(target_os = "linux"))]