From 973b879de24a81fcf3b4435757397540b5104811 Mon Sep 17 00:00:00 2001
From: burgerguy <ryan.gehling@yahoo.com>
Date: Sat, 29 Apr 2023 23:42:59 -0400
Subject: [PATCH] correctly use no_std, get rid of macro in exchange for zips
 and maps, bump ver

---
 Cargo.toml          |  2 +-
 src/lib.rs          | 10 +++-------
 src/simd/consts.rs  | 22 ----------------------
 src/simd/float.rs   | 25 ++++++++++++++++++++-----
 src/simd/int.rs     |  7 +++----
 src/simd/mod.rs     |  1 -
 src/test/compile.rs |  6 +++---
 src/test/mod.rs     |  4 +++-
 8 files changed, 33 insertions(+), 44 deletions(-)
 delete mode 100644 src/simd/consts.rs

diff --git a/Cargo.toml b/Cargo.toml
index f5bf393..de38fb3 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -1,7 +1,7 @@
 [package]
 name = "fath"
 authors = [ "burgerindividual", "duplexsystem" ]
-version = "0.1.5"
+version = "0.1.6"
 edition = "2021"
 license = "LGPL-3.0"
 repository = "https://github.com/burgerindividual/fath"
diff --git a/src/lib.rs b/src/lib.rs
index 590bdf2..da5d2e9 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -1,9 +1,5 @@
-#![feature(core_intrinsics, portable_simd)]
-// #![no_std]
-
-pub extern crate std as core;
-
-extern crate alloc;
+#![feature(core_intrinsics, portable_simd, array_zip)]
+#![no_std]
 
 pub mod scalar;
 pub mod shared;
@@ -11,4 +7,4 @@ pub mod shared;
 pub mod simd;
 
 #[cfg(test)]
-pub mod test;
\ No newline at end of file
+pub mod test;
diff --git a/src/simd/consts.rs b/src/simd/consts.rs
deleted file mode 100644
index 8632cae..0000000
--- a/src/simd/consts.rs
+++ /dev/null
@@ -1,22 +0,0 @@
-#[macro_export]
-macro_rules! wrap_auto_vectorize {
-    ($func:expr, $lanes:expr, $($x:ident),+) => {
-        {
-            let mut vec_uninit: core::mem::MaybeUninit<Simd<_, $lanes>> = core::mem::MaybeUninit::uninit();
-            let vec_ptr = vec_uninit.as_mut_ptr();
-
-            let mut i = 0;
-            while i < $lanes {
-                let evaluated = $func($($x[i]),+);
-                #[allow(unused_unsafe)]
-                unsafe {
-                    (*vec_ptr)[i] = evaluated;
-                }
-                i += 1;
-            }
-
-            #[allow(unused_unsafe)]
-            unsafe { vec_uninit.assume_init() }
-        }
-    }
-}
diff --git a/src/simd/float.rs b/src/simd/float.rs
index f72c344..ed3dd55 100644
--- a/src/simd/float.rs
+++ b/src/simd/float.rs
@@ -1,5 +1,4 @@
 use crate::shared::float::*;
-use crate::*;
 use core::simd::*;
 
 impl<const LANES: usize> FastApproxFloat for Simd<f32, LANES>
@@ -8,21 +7,37 @@ where
 {
     #[inline(always)]
     unsafe fn sin_fast_approx<const PRECISION: usize>(self) -> Self {
-        wrap_auto_vectorize!(sin_fast_approx::<PRECISION, false>, LANES, self)
+        Simd::from_array(
+            self.as_array()
+                .map(|e| sin_fast_approx::<PRECISION, false>(e)),
+        )
     }
 
     #[inline(always)]
     unsafe fn cos_fast_approx<const PRECISION: usize>(self) -> Self {
-        wrap_auto_vectorize!(sin_fast_approx::<PRECISION, true>, LANES, self)
+        Simd::from_array(
+            self.as_array()
+                .map(|e| sin_fast_approx::<PRECISION, true>(e)),
+        )
     }
 
     #[inline(always)]
     unsafe fn log_fast_approx<const PRECISION: usize>(self, base: Self) -> Self {
-        wrap_auto_vectorize!(log_fast_approx::<PRECISION>, LANES, self, base)
+        Simd::from_array(
+            self.as_array()
+                .zip(*base.as_array())
+                .map(|(self_elem, base_elem)| log_fast_approx::<PRECISION>(self_elem, base_elem)),
+        )
     }
 
     #[inline(always)]
     unsafe fn log_fast_approx_const_base<const PRECISION: usize>(self, base: Self) -> Self {
-        wrap_auto_vectorize!(log_fast_approx_const_base::<PRECISION>, LANES, self, base)
+        Simd::from_array(
+            self.as_array()
+                .zip(*base.as_array())
+                .map(|(self_elem, base_elem)| {
+                    log_fast_approx_const_base::<PRECISION>(self_elem, base_elem)
+                }),
+        )
     }
 }
diff --git a/src/simd/int.rs b/src/simd/int.rs
index 4fae199..58279ca 100644
--- a/src/simd/int.rs
+++ b/src/simd/int.rs
@@ -1,9 +1,8 @@
 use crate::shared::int::*;
 use crate::*;
 
+use core::mem::size_of;
 use core::simd::*;
-use mem::size_of;
-use std::mem;
 
 macro_rules! unsigned_impl {
     ($u:ty,$s:ty,$f:ty,$mant_bits:expr) => {
@@ -31,10 +30,10 @@ macro_rules! unsigned_impl {
                     let unsigned_mask = Mask::from_int_unchecked(
                         self.cast::<$s>() >> Simd::splat(UNSIGNED_LOG2 as $s),
                     );
-                    
+
                     // need to get rid of bits that could cause a round-up
                     let adjusted = (self & !(self >> Simd::splat($mant_bits + 1))).cast::<$s>();
-                    
+
                     let exponent = (adjusted.cast::<$f>().to_bits() >> Simd::splat($mant_bits))
                         - Simd::splat((1 << ((size_of::<$f>() * 8) - 2 - $mant_bits)) - 1);
 
diff --git a/src/simd/mod.rs b/src/simd/mod.rs
index 4bf4b3b..a89042e 100644
--- a/src/simd/mod.rs
+++ b/src/simd/mod.rs
@@ -1,3 +1,2 @@
-pub mod consts;
 pub mod float;
 pub mod int;
diff --git a/src/test/compile.rs b/src/test/compile.rs
index 11e5874..fe31a4c 100644
--- a/src/test/compile.rs
+++ b/src/test/compile.rs
@@ -1,8 +1,8 @@
-use crate::shared::int::*;
+use crate::shared::float::FastApproxFloat;
 use core::simd::*;
 
 #[inline(never)]
 #[allow(dead_code)]
-pub fn test(x: u32x8) -> u32x8 {
-    unsafe { x.ilog_const_base_unchecked::<2>() }
+pub fn test(x: f32x8, base: f32x8) -> f32x8 {
+    unsafe { x.log_fast_approx::<0>(base) }
 }
diff --git a/src/test/mod.rs b/src/test/mod.rs
index 7d02b94..29eef7a 100644
--- a/src/test/mod.rs
+++ b/src/test/mod.rs
@@ -1,2 +1,4 @@
+#[cfg(test)]
 pub mod checks;
-pub mod compile;
\ No newline at end of file
+
+pub mod compile;