Add no no_std support (#74)

* Add no `no_std` support

# Conflicts:
#	.github/workflows/build_test.yml
#	Cargo.toml

* More readme

---------

Co-authored-by: Moritz Mœller <[email protected]>

Cargo.toml

@@ -13,6 +13,8 @@ categories = ["algorithms", "data-structures", "no-std"]
 exclude = ["article/*"]
 
 [features]
+default = ["std"]
+std = []
 # Only relevant for throughput benchmarks
 bench-csv = []
 bench-md = []

README.md

@@ -35,6 +35,9 @@ Thanks to this, GxHash passes all [SMHasher](https://github.com/rurban/smhasher)
 
 Check out the [paper](https://github.com/ogxd/gxhash-rust/blob/main/article/article.pdf) for more technical details.
 
+### 0 Dependencies 📦
+GxHash has 0 cargo dependency. The `Hasher` and `Hashset`/`Hashmap` convenience types require the standard library, enabled by default with the `std` feature.
+
 ## Portability
 
 > **Important**
@@ -50,6 +53,22 @@ GxHash is compatible with:
 ### Hashes Stability
 All generated hashes for a given version of GxHash are stable, meaning that for a given input the output hash will be the same across all supported platforms.
 
+### `no_std`
+
+The `std` feature flag enables the `HashMap`/`HashSet` container convenience type aliases. This is on by default. Disable to make the crate `no_std`:
+
+```toml
+[dependencies.gxhash]
+...
+default-features = false
+```
+
+### `hybrid`
+
+The `hybrid` feature flag enables a hybrid implementation of GxHash. This is disabled by default. When `hybrid` feature is enabled and for CPUs that supports it, GxHash will use wider registers and instructions (`VAES` + `AVX2`), which can lead to a significant performance improvement for large inputs. This preserves hashes stability, meaning that hashes generated with or without the `hybrid` feature are the same for a given input and seed.
+
+*Note: Even without this feature enabled GxHash is already the fastest option out there. We recommend enabling this feature only when inputs can be larger than a few hundred bytes, see the benchmarks below.*
+
 ## Benchmarks
 
 [![Benchmark](https://github.com/ogxd/gxhash/actions/workflows/bench.yml/badge.svg)](https://github.com/ogxd/gxhash/actions/workflows/bench.yml)  

src/gxhash/platform/arm.rs

@@ -27,7 +27,7 @@ pub unsafe fn get_partial_safe(data: *const State, len: usize) -> State {
     // Temporary buffer filled with zeros
     let mut buffer = [0i8; VECTOR_SIZE];
     // Copy data into the buffer
-    std::ptr::copy(data as *const i8, buffer.as_mut_ptr(), len);
+    core::ptr::copy(data as *const i8, buffer.as_mut_ptr(), len);
     // Load the buffer into a __m256i vector
     let partial_vector = vld1q_s8(buffer.as_ptr());
     vaddq_s8(partial_vector, vdupq_n_s8(len as i8))

src/gxhash/platform/mod.rs

@@ -8,7 +8,7 @@ mod platform;
 
 pub use platform::*;
 
-use std::mem::size_of;
+use core::mem::size_of;
 
 pub(crate) const VECTOR_SIZE: usize = size_of::<State>();
 // 4KiB is the default page size for most systems, and conservative for other systems such as MacOS ARM (16KiB)

src/gxhash/platform/x86.rs

@@ -27,7 +27,7 @@ pub unsafe fn get_partial_safe(data: *const State, len: usize) -> State {
     // Temporary buffer filled with zeros
     let mut buffer = [0i8; VECTOR_SIZE];
     // Copy data into the buffer
-    std::ptr::copy(data as *const i8, buffer.as_mut_ptr(), len);
+    core::ptr::copy(data as *const i8, buffer.as_mut_ptr(), len);
     // Load the buffer into a __m256i vector
     let partial_vector = _mm_loadu_si128(buffer.as_ptr() as *const State);
     _mm_add_epi8(partial_vector, _mm_set1_epi8(len as i8))

src/lib.rs

@@ -1,3 +1,4 @@
+#![cfg_attr(not(feature = "std"), no_std)]
 // Hybrid SIMD width usage currently requires unstable 'stdsimd'
 #![cfg_attr(feature = "hybrid", feature(stdarch_x86_avx512))]
 
@@ -9,7 +10,9 @@ compile_error!{"Gxhash requires aes intrinsics. Make sure the processor supports
 
 #[rustfmt::skip]
 mod gxhash;
-mod hasher;
-
 pub use crate::gxhash::*;
+
+#[cfg(feature = "std")]
+mod hasher;
+#[cfg(feature = "std")]
 pub use crate::hasher::*;