Add intrinsics for float arithmetic with `fast` flag enabled

`fast` a.k.a UnsafeAlgebra is the flag for enabling all "unsafe"
(according to llvm) float optimizations.

See LangRef for more information http://llvm.org/docs/LangRef.html#fast-math-flags

Providing these operations with less precise associativity rules (for
example) is useful to numerical applications.

For example, the summation loop:

    let sum = 0.;
    for element in data {
        sum += *element;
    }

Using the default floating point semantics, this loop expresses the
floats must be added in a sequence, one after another. This constraint
is usually completely unintended, and it means that no autovectorization
is possible.

1 files changed, 26 insertions, 0 deletions

diff --git a/src/libcore/intrinsics.rs b/src/libcore/intrinsics.rs
index 0417ef84163..69cfd0368d6 100644
--- a/src/libcore/intrinsics.rs
+++ b/src/libcore/intrinsics.rs
@@ -539,6 +539,32 @@ extern "rust-intrinsic" {
     /// Returns the nearest integer to an `f64`. Rounds half-way cases away from zero.
     pub fn roundf64(x: f64) -> f64;
 
+    /// Float addition that allows optimizations based on algebraic rules.
+    /// May assume inputs are finite.
+    #[cfg(not(stage0))]
+    pub fn fadd_fast<T>(a: T, b: T) -> T;
+
+    /// Float subtraction that allows optimizations based on algebraic rules.
+    /// May assume inputs are finite.
+    #[cfg(not(stage0))]
+    pub fn fsub_fast<T>(a: T, b: T) -> T;
+
+    /// Float multiplication that allows optimizations based on algebraic rules.
+    /// May assume inputs are finite.
+    #[cfg(not(stage0))]
+    pub fn fmul_fast<T>(a: T, b: T) -> T;
+
+    /// Float division that allows optimizations based on algebraic rules.
+    /// May assume inputs are finite.
+    #[cfg(not(stage0))]
+    pub fn fdiv_fast<T>(a: T, b: T) -> T;
+
+    /// Float remainder that allows optimizations based on algebraic rules.
+    /// May assume inputs are finite.
+    #[cfg(not(stage0))]
+    pub fn frem_fast<T>(a: T, b: T) -> T;
+
+
     /// Returns the number of bits set in an integer type `T`
     pub fn ctpop<T>(x: T) -> T;