2-arg atan

Author: christiangnrd

src/device/intrinsics/math.jl

@@ -53,6 +53,10 @@ using Base.Math: throw_complex_domainerror
 @device_override Base.atan(x::Float32) = ccall("extern air.atan.f32", llvmcall, Cfloat, (Cfloat,), x)
 @device_override Base.atan(x::Float16) = ccall("extern air.atan.f16", llvmcall, Float16, (Float16,), x)
 
+@device_override FastMath.atan_fast(x::Float32, y::Float32) = ccall("extern air.fast_atan2.f32", llvmcall, Cfloat, (Cfloat, Cfloat), x, y)
+@device_override Base.atan(x::Float32, y::Float32) = ccall("extern air.atan2.f32", llvmcall, Cfloat, (Cfloat, Cfloat), x, y)
+@device_override Base.atan(x::Float16, y::Float16) = ccall("extern air.atan2.f16", llvmcall, Float16, (Float16, Float16), x, y)
+
 @device_override FastMath.atanh_fast(x::Float32) = ccall("extern air.fast_atanh.f32", llvmcall, Cfloat, (Cfloat,), x)
 @device_override Base.atanh(x::Float32) = ccall("extern air.atanh.f32", llvmcall, Cfloat, (Cfloat,), x)
 @device_override Base.atanh(x::Float16) = ccall("extern air.atanh.f16", llvmcall, Float16, (Float16,), x)

test/device/intrinsics.jl

@@ -151,7 +151,7 @@ MATH_INTR_FUNCS_2_ARG = [
     # step, # T step(T edge, T x) Returns 0.0 if x < edge, otherwise it returns 1.0
 
     # float math
-    # atan2, # T atan2(T x, T y) Compute arc tangent of y over x.
+    atan, # T atan2(T x, T y) Compute arc tangent of y over x.
     # fdim, # T fdim(T x, T y)
     max, # T [f]max(T x, T y)
     min, # T [f]min(T x, T y)