gistfile1.txt

Julia v0.7's stack-allocation optimizations and escape analysis 

https://github.com/JuliaLang/julia/pull/22684

Watch it in action: stack-allocated mutable static sized arrays. Example thanks to Valentin Churavy.

julia> function f(x::Int64)
         A = MArray{Tuple{3}}(1, 2, 3)
         A[2] += x
         return A[2]-A[1] 
       end
f (generic function with 1 method)

julia> @code_llvm f(2)

; Function f
; Location: REPL[13]
define i64 @julia_f_63256(i64) #0 {
L23:
; Location: REPL[13]:4
  %1 = add i64 %0, 1
  ret i64 %1
}

################################################

v0.6

julia> function f(x::Int64)
                A = MArray{Tuple{3}}(1, 2, 3)
                A[2] += x
                return A[2]-A[1] 
              end
f (generic function with 1 method)

julia> @code_llvm f(2)

define i64 @julia_f_60721(i64) #0 !dbg !5 {
L25:
  %ptls_i8 = call i8* asm "movq %fs:0, $0;\0Aaddq $$-10888, $0", "=r,~{dirflag},~{fpsr},~{flags}"() #5
  %ptls = bitcast i8* %ptls_i8 to i8****
  %1 = alloca [4 x i8**], align 8
  %.sub = getelementptr inbounds [4 x i8**], [4 x i8**]* %1, i64 0, i64 0
  %2 = getelementptr [4 x i8**], [4 x i8**]* %1, i64 0, i64 2
  %3 = bitcast [4 x i8**]* %1 to i64*
  %4 = bitcast i8*** %2 to i8*
  call void @llvm.memset.p0i8.i64(i8* %4, i8 0, i64 16, i32 8, i1 false)
  store i64 4, i64* %3, align 8
  %5 = getelementptr [4 x i8**], [4 x i8**]* %1, i64 0, i64 1
  %6 = bitcast i8* %ptls_i8 to i64*
  %7 = load i64, i64* %6, align 8
  %8 = bitcast i8*** %5 to i64*
  store i64 %7, i64* %8, align 8
  store i8*** %.sub, i8**** %ptls, align 8
  %9 = call i8** @jl_gc_pool_alloc(i8* %ptls_i8, i32 1456, i32 32)
  %10 = getelementptr i8*, i8** %9, i64 -1
  %11 = bitcast i8** %10 to i8***
  store i8** inttoptr (i64 140077945663888 to i8**), i8*** %11, align 8
  store i8** %9, i8*** %2, align 8
  %.repack = bitcast i8** %9 to i64*
  store i64 1, i64* %.repack, align 16
  %.repack2 = getelementptr i8*, i8** %9, i64 1
  %12 = bitcast i8** %.repack2 to i64*
  %.repack3 = getelementptr inbounds i8*, i8** %9, i64 2
  %13 = bitcast i8** %.repack3 to i64*
  store i64 3, i64* %13, align 16
  %14 = add i64 %0, 2
  store i64 %14, i64* %12, align 1
  %15 = add i64 %0, 1
  %16 = load i64, i64* %8, align 8
  store i64 %16, i64* %6, align 8
  ret i64 %15
}