Add cr_hypot from core-math

Author: tgross35

libm-test/src/generate/case_list.rs

@@ -18,7 +18,6 @@ pub struct TestCase<Op: MathOp> {
 }
 
 impl<Op: MathOp> TestCase<Op> {
-    #[expect(dead_code)]
     fn append_inputs(v: &mut Vec<Self>, l: &[Op::RustArgs]) {
         v.extend(l.iter().copied().map(|input| Self {
             input,
@@ -479,7 +478,27 @@ fn frexpf128_cases() -> Vec<TestCase<op::frexpf128::Routine>> {
 }
 
 fn hypot_cases() -> Vec<TestCase<op::hypot::Routine>> {
-    vec![]
+    let mut v = vec![];
+    TestCase::append_inputs(
+        &mut v,
+        &[
+            // Cases that can overflow exponent if wrapping arithmetic is not used
+            (
+                hf64!("-0x1.800f800f80100p+1023"),
+                hf64!("0x1.8354835473720p+996"),
+            ),
+            (hf64!("0x1.201b201b201c0p+0"), hf64!("0x1.b028b028b02a0p-1")),
+            (
+                hf64!("-0x1.e538e538e564p+980"),
+                hf64!("-0x1.c4dfc4dfc508p+983"),
+            ),
+            (
+                hf64!("-0x1.2f22e4f77aa58p+983"),
+                hf64!("-0x1.44c9f5524c8ccp+980"),
+            ),
+        ],
+    );
+    v
 }
 
 fn hypotf_cases() -> Vec<TestCase<op::hypotf::Routine>> {

libm-test/src/precision.rs

@@ -46,6 +46,7 @@ pub fn default_ulp(ctx: &CheckCtx) -> u32 {
 
         // Operations that aren't required to be exact, but our implementations are.
         Bn::Cbrt => 0,
+        Bn::Hypot if ctx.fn_ident == Id::Hypot => 0,
 
         // Bessel functions have large inaccuracies.
         Bn::J0 | Bn::J1 | Bn::Y0 | Bn::Y1 | Bn::Jn | Bn::Yn => 8_000_000,
@@ -112,6 +113,7 @@ pub fn default_ulp(ctx: &CheckCtx) -> u32 {
             Id::Fdim => ulp = 2,
             Id::Exp2f => ulp = 1,
             Id::Expf => ulp = 1,
+            Id::Hypot => ulp = 1,
             Id::Sincosf => ulp = 500,
             Id::Tgamma => ulp = 20,
             _ => (),