Use native fmin/fmax in aarch64

base/math.jl

@@ -851,26 +851,44 @@ minmax(x::T, y::T) where {T<:AbstractFloat} = min(x, y), max(x, y)
 
 _isless(x::Float16, y::Float16) = signbit(widen(x) - widen(y))
 
+Base.@assume_effects :total @inline llvm_min(x::Float64, y::Float64) = ccall("llvm.minimum.f64", llvmcall, Float64, (Float64, Float64), x, y)
+Base.@assume_effects :total @inline llvm_min(x::Float32, y::Float32) = ccall("llvm.minimum.f32", llvmcall, Float32, (Float32, Float32), x, y)
+Base.@assume_effects :total @inline llvm_max(x::Float64, y::Float64) = ccall("llvm.maximum.f64", llvmcall, Float64, (Float64, Float64), x, y)
+Base.@assume_effects :total @inline llvm_max(x::Float32, y::Float32) = ccall("llvm.maximum.f32", llvmcall, Float32, (Float32, Float32), x, y)
+
+const has_native_fminmax = Sys.ARCH === :aarch64
 function min(x::T, y::T) where {T<:Union{Float32,Float64}}
+    if has_native_fminmax
+        return llvm_min(x,y)
+    else
     diff = x - y
     argmin = ifelse(signbit(diff), x, y)
     anynan = isnan(x)|isnan(y)
-    ifelse(anynan, diff, argmin)
+    return ifelse(anynan, diff, argmin)
+    end
 end
 
 function max(x::T, y::T) where {T<:Union{Float32,Float64}}
+    if has_native_fminmax
+        return llvm_max(x,y)
+    else
     diff = x - y
     argmax = ifelse(signbit(diff), y, x)
     anynan = isnan(x)|isnan(y)
-    ifelse(anynan, diff, argmax)
+    return ifelse(anynan, diff, argmax)
+    end
 end
 
 function minmax(x::T, y::T) where {T<:Union{Float32,Float64}}
+    if has_native_fminmax
+        return llvm_min(x, y), llvm_max(x, y)
+    else
     diff = x - y
     sdiff = signbit(diff)
     min, max = ifelse(sdiff, x, y), ifelse(sdiff, y, x)
     anynan = isnan(x)|isnan(y)
-    ifelse(anynan, diff, min), ifelse(anynan, diff, max)
+    return ifelse(anynan, diff, min), ifelse(anynan, diff, max)
+    end
 end
 
 """