AVX512-VP2INTERSECT implementations

crates/core_arch/src/x86/avx512vp2intersect.rs

@@ -0,0 +1,77 @@
+//! Vector Pair Intersection to a Pair of Mask Registers (VP2INTERSECT)
+//!
+//! The intrinsics here correspond to those in the `immintrin.h` C header.
+//!
+//! The reference is [Intel® Architecture
+//! Instruction Set Extensions and Future Features
+//! Programming Reference][intel64_ext_ref].
+//!
+//! [intel64_ext_ref]: https://software.intel.com/sites/default/files/managed/c5/15/architecture-instruction-set-extensions-programming-reference.pdf
+
+use crate::core_arch::simd::i64x8;
+use crate::core_arch::simd::{i32x16, i32x4, i32x8, i64x2, i64x4};
+
+use crate::core_arch::x86::m128iExt;
+
+use crate::core_arch::x86::{__m128i, __mmask32, __mmask8};
+
+use core::ptr;
+
+#[cfg(test)]
+use stdarch_test::assert_instr;
+
+#[allow(improper_ctypes)]
+extern "C" {
+    #[link_name = "llvm.x86.avx512.vp2intersect.d.128"]
+    fn vp2intersectd_128(a: i32x4, b: i32x4);
+    #[link_name = "llvm.x86.avx512.vp2intersect.q.128"]
+    fn vp2intersectq_128(a: i64x2, b: i64x2);
+
+    #[link_name = "llvm.x86.avx512.vp2intersect.d.256"]
+    fn vp2intersectd_256(a: i32x8, b: i32x8);
+    #[link_name = "llvm.x86.avx512.vp2intersect.q.256"]
+    fn vp2intersectq_256(a: i64x4, b: i64x4);
+
+    #[link_name = "llvm.x86.avx512.vp2intersect.d.512"]
+    fn vp2intersectd_512(a: i32x16, b: i32x16);
+    #[link_name = "llvm.x86.avx512.vp2intersect.q.512"]
+    fn vp2intersectq_512(a: i64x8, b: i64x8);
+}
+
+/// Compute intersection of packed 32-bit integer vectors a and b,
+/// and store indication of match in the corresponding bit of two mask registers
+/// specified by k1 and k2. A match in corresponding elements of a and b is
+/// indicated by a set bit in the corresponding bit of the mask registers.
+///
+/// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_2intersect_epi32&expand=0)
+#[inline]
+#[target_feature(enable = "avx512vp2intersect,avx512vl")]
+#[cfg_attr(test, assert_instr(vp2intersectd))]
+pub unsafe fn _mm_2intersect_epi32(
+    a: __m128i,
+    b: __m128i,
+    k1: *const __mmask8,
+    k2: *const __mmask8,
+) {
+    let k = __mmask32::MIN;
+    vp2intersectd_128(a.as_i32x4(), b.as_i32x4());
+    let interim = (k >> 4) << 4;
+    ptr::write(k1 as *mut __mmask8, interim as u8);
+    ptr::write(k2 as *mut __mmask8, ((interim >> 4) << 4) as u8);
+}
+
+#[cfg(test)]
+mod tests {
+    use stdarch_test::simd_test;
+
+    use crate::core_arch::simd::i32x4;
+    use crate::core_arch::x86::*;
+
+    #[simd_test(enable = "avx512vp2intersect,avx512vl")]
+    unsafe fn test_mm_2intersect_epi32() {
+        let k1 = u32::MIN as *const __mmask8;
+        let k2 = u32::MIN as *const __mmask8;
+
+        _mm_2intersect_epi32(__m128i(123, 123), __m128i(456, 456), k1, k2);
+    }
+}

crates/core_arch/src/x86/mod.rs

@@ -717,6 +717,9 @@ pub use self::avx512vaes::*;
 mod avx512vpclmulqdq;
 pub use self::avx512vpclmulqdq::*;
 
+mod avx512vp2intersect;
+pub use self::avx512vp2intersect::*;
+
 mod bt;
 pub use self::bt::*;