Fix build on x86 and bump cargo doc version as packed_simd fails on docs.rs version

Author: alecmocatta

Cargo.toml

@@ -6,7 +6,7 @@ authors = ["Alec Mocatta <[email protected]>"]
 categories = ["data-structures","algorithms","science"]
 keywords = ["streaming-algorithm","probabilistic","sketch","data-structure","hyperloglog"]
 description = """
-Performant implementations of various streaming algorithms, including Count–min sketch; Top k; HyperLogLog; Reservoir sampling.
+SIMD-accelerated implementations of various streaming algorithms, including Count–min sketch, Top k, HyperLogLog, Reservoir sampling.
 """
 repository = "https://github.com/alecmocatta/streaming_algorithms"
 homepage = "https://github.com/alecmocatta/streaming_algorithms"
@@ -26,17 +26,4 @@ twox-hash = "1.1"
 serde_derive = "1.0"
 serde = "1.0"
 rand = "0.5"
-bytecount = "0.3"
-# faster = { git = "https://github.com/alecmocatta/faster" } #"0.5"
-# faster = { path = "./faster" } #"0.5"
 packed_simd = { version = "0.3", features = ["into_bits"] }
-
-[features]
-avx-accel = ["bytecount/avx-accel"]
-simd-accel = ["bytecount/simd-accel"]
-
-# [replace]
-# "vektor:0.2.0" = { git = "https://github.com/alecmocatta/vektor" }
-
-[profile.release]
-debug = 2

README.md

@@ -8,7 +8,7 @@
 
 [Docs](https://docs.rs/streaming_algorithms/0.1.0)
 
-Performant implementations of various [streaming algorithms](https://en.wikipedia.org/wiki/Streaming_algorithm).
+SIMD-accelerated implementations of various [streaming algorithms](https://en.wikipedia.org/wiki/Streaming_algorithm).
 
 This library is a work in progress. PRs are very welcome! Currently implemented algorithms include:
 
@@ -17,7 +17,13 @@ This library is a work in progress. PRs are very welcome! Currently implemented
  * HyperLogLog
  * Reservoir sampling
 
-A goal of this library is to enable composition of these algorithms; for example Top k + HyperLogLog to enable roughly `SELECT key FROM table GROUP BY key ORDER BY COUNT(DISTINCT value) DESC LIMIT k`.
+A goal of this library is to enable composition of these algorithms; for example Top k + HyperLogLog to enable an approximate version of something akin to `SELECT key FROM table GROUP BY key ORDER BY COUNT(DISTINCT value) DESC LIMIT k`.
+
+Run your application with `RUSTFLAGS="-C target-cpu=native"` to benefit from the SIMD-acceleration like so:
+
+```bash
+	RUSTFLAGS="-C target-cpu=native" cargo run --release
+```
 
 See [this gist](https://gist.github.com/debasishg/8172796) for a good list of further algorithms to be implemented. Other resources are [Probabilistic data structures – Wikipedia](https://en.wikipedia.org/wiki/Category:Probabilistic_data_structures), [DataSketches – A similar Java library originating at Yahoo](https://datasketches.github.io/), and [Algebird  – A similar Java library originating at Twitter](https://github.com/twitter/algebird).
 

src/distinct.rs

@@ -392,29 +392,33 @@ mod simd_types {
 use self::simd_types::*;
 
 struct Sad<X>(PhantomData<fn(X)>);
+#[cfg(any(target_arch = "x86", target_arch = "x86_64"))]
+mod x86 {
+	#[cfg(target_arch = "x86")]
+	pub use std::arch::x86::*;
+	#[cfg(target_arch = "x86_64")]
+	pub use std::arch::x86_64::*;
+}
 // TODO
 // #[cfg(target_feature = "avx512bw")]
 // impl Sad<packed_simd::u8x64> {
 // 	#[inline]
 // 	#[target_feature(enable = "avx512bw")]
 // 	unsafe fn sad(a: packed_simd::u8x64, b: packed_simd::u8x64) -> packed_simd::u64x8 {
-// 		use std::{arch::x86_64::_mm512_sad_epu8, mem::transmute};
-// 		packed_simd::Simd(transmute(_mm512_sad_epu8(transmute(a.0), transmute(b.0))))
+// 		use std::mem::transmute;
+// 		packed_simd::Simd(transmute(x86::_mm512_sad_epu8(transmute(a.0), transmute(b.0))))
 // 	}
 // }
-mod x86 {
-	#[cfg(target_arch = "x86")]
-	pub use std::arch::x86::*;
-	#[cfg(target_arch = "x86_64")]
-	pub use std::arch::x86_64::*;
-}
 #[cfg(target_feature = "avx2")]
 impl Sad<packed_simd::u8x32> {
 	#[inline]
 	#[target_feature(enable = "avx2")]
 	unsafe fn sad(a: packed_simd::u8x32, b: packed_simd::u8x32) -> packed_simd::u64x4 {
 		use std::mem::transmute;
-		packed_simd::Simd(transmute(x86::_mm256_sad_epu8(transmute(a.0), transmute(b.0))))
+		packed_simd::Simd(transmute(x86::_mm256_sad_epu8(
+			transmute(a.0),
+			transmute(b.0),
+		)))
 	}
 }
 #[cfg(target_feature = "sse2")]

src/lib.rs

@@ -1,4 +1,4 @@
-//! Performant implementations of various [streaming algorithms](https://en.wikipedia.org/wiki/Streaming_algorithm).
+//! SIMD-accelerated implementations of various [streaming algorithms](https://en.wikipedia.org/wiki/Streaming_algorithm).
 //!
 //! **[Crates.io](https://crates.io/crates/streaming_algorithms) │ [Repo](https://github.com/alecmocatta/streaming_algorithms)**
 //!
@@ -9,7 +9,13 @@
 //!  * HyperLogLog
 //!  * Reservoir sampling
 //!
-//! A goal of this library is to enable composition of these algorithms; for example Top k + HyperLogLog to enable roughly `SELECT key FROM table GROUP BY key ORDER BY COUNT(DISTINCT value) DESC LIMIT k`.
+//! A goal of this library is to enable composition of these algorithms; for example Top k + HyperLogLog to enable an approximate version of something akin to `SELECT key FROM table GROUP BY key ORDER BY COUNT(DISTINCT value) DESC LIMIT k`.
+//!
+//! Run your application with `RUSTFLAGS="-C target-cpu=native"` to benefit from the SIMD-acceleration like so:
+//!
+//! ```bash
+//! 	RUSTFLAGS="-C target-cpu=native" cargo run --release
+//! ```
 //!
 //! See [this gist](https://gist.github.com/debasishg/8172796) for a good list of further algorithms to be implemented. Other resources are [Probabilistic data structures – Wikipedia](https://en.wikipedia.org/wiki/Category:Probabilistic_data_structures), [DataSketches – A similar Java library originating at Yahoo](https://datasketches.github.io/), and [Algebird  – A similar Java library originating at Twitter](https://github.com/twitter/algebird).
 //!
@@ -54,11 +60,9 @@
 extern crate twox_hash;
 #[macro_use]
 extern crate serde_derive;
-// extern crate bytecount;
+extern crate packed_simd;
 extern crate rand;
 extern crate serde;
-// extern crate faster;
-extern crate packed_simd;
 
 mod count_min;
 mod distinct;

src/top.rs

@@ -354,7 +354,7 @@ mod test {
 		// println!("{:#?}", x);
 	}
 
-	// #[ignore] // takes too long on CI
+	#[ignore] // takes too long on CI
 	#[test]
 	fn many() {
 		let start = time::Instant::now();