Apply formatter

perf/flame.jl

@@ -10,21 +10,21 @@ include(joinpath(pkgdir(MBF), "test", "utils.jl"))
 has_cuda = CUDA.has_cuda()
 AType = has_cuda ? CUDA.CuArray : Array
 # arr_size = (prod((50,5,5,6,50)),)
-arr_size = (50,5,5,6,50)
+arr_size = (50, 5, 5, 6, 50)
 X = get_arrays(:x, arr_size, AType)
 Y = get_arrays(:y, arr_size, AType)
 
 function perf_kernel_fused!(X, Y)
-    (;x1,x2,x3,x4,x5,x6,x7,x8,x9,x10) = X
-    (;y1,y2,y3,y4,y5,y6,y7,y8,y9,y10) = Y
+    (; x1, x2, x3, x4, x5, x6, x7, x8, x9, x10) = X
+    (; y1, y2, y3, y4, y5, y6, y7, y8, y9, y10) = Y
     MBF.@fused begin
-        @. y1 = x1+x2+x3+x4
-        @. y2 = x2+x3+x4+x5
-        @. y3 = x3+x4+x5+x6
-        @. y4 = x4+x5+x6+x7
-        @. y5 = x5+x6+x7+x8
-        @. y6 = x6+x7+x8+x9
-        @. y7 = x7+x8+x9+x10
+        @. y1 = x1 + x2 + x3 + x4
+        @. y2 = x2 + x3 + x4 + x5
+        @. y3 = x3 + x4 + x5 + x6
+        @. y4 = x4 + x5 + x6 + x7
+        @. y5 = x5 + x6 + x7 + x8
+        @. y6 = x6 + x7 + x8 + x9
+        @. y7 = x7 + x8 + x9 + x10
     end
 end
 

src/MultiBroadcastFusion.jl

@@ -9,7 +9,7 @@ abstract type AbstractFusedMultiBroadcast end
 A mult-broadcast fusion object
 """
 struct FusedMultiBroadcast{T} <: AbstractFusedMultiBroadcast
-	pairs::T
+    pairs::T
 end
 
 # Base.@propagate_inbounds function rcopyto_at!(pair::Pair, i::CartesianIndex)
@@ -25,7 +25,7 @@ end
 #     return nothing
 # end
 Base.@propagate_inbounds function rcopyto_at!(pair::Pair, i...)
-    dest,src = pair.first, pair.second
+    dest, src = pair.first, pair.second
     @inbounds dest[i...] = src[i...]
     return nothing
 end

src/macro_utils.jl

@@ -1,83 +1,82 @@
 
 function materialize_args(expr::Expr)
-	@assert expr.head == :call
-	@assert expr.args[1] == :(Base.materialize!)
-	return (expr.args[2], expr.args[3])
+    @assert expr.head == :call
+    @assert expr.args[1] == :(Base.materialize!)
+    return (expr.args[2], expr.args[3])
 end
 
-function fused(expr)
-end
+function fused(expr) end
 
 macro fused(expr)
-	_pairs = gensym()
-	quote
-		$_pairs = $(esc(fused_pairs(expr)))
-		Base.copyto!(FusedMultiBroadcast($_pairs))
-	end
+    _pairs = gensym()
+    quote
+        $_pairs = $(esc(fused_pairs(expr)))
+        Base.copyto!(FusedMultiBroadcast($_pairs))
+    end
 end
 
 macro fused_pairs(expr)
-	esc(fused_pairs(expr))
+    esc(fused_pairs(expr))
 end
 
 function _fused_pairs(expr::Expr)
-	@assert expr.head == :block
-	exprs_out = []
-	for _expr in expr.args
-		# TODO: should we retain LineNumberNode?
-		_expr isa LineNumberNode && continue
-		@assert _expr isa Expr
-		if _expr.head == :macrocall && _expr.args[1] == Symbol("@__dot__")
-			se = code_lowered_single_expression(_expr)
-			margs = materialize_args(se)
-			push!(exprs_out, :(Pair($(margs[1]), $(margs[2]))))
-		end
-	end
-	if length(exprs_out) == 1
-		return "($(exprs_out[1]),)"
-	else
-		return "("*join(exprs_out, ",")*")"
-	end
+    @assert expr.head == :block
+    exprs_out = []
+    for _expr in expr.args
+        # TODO: should we retain LineNumberNode?
+        _expr isa LineNumberNode && continue
+        @assert _expr isa Expr
+        if _expr.head == :macrocall && _expr.args[1] == Symbol("@__dot__")
+            se = code_lowered_single_expression(_expr)
+            margs = materialize_args(se)
+            push!(exprs_out, :(Pair($(margs[1]), $(margs[2]))))
+        end
+    end
+    if length(exprs_out) == 1
+        return "($(exprs_out[1]),)"
+    else
+        return "(" * join(exprs_out, ",") * ")"
+    end
 end
 
 fused_pairs(expr::Expr) = Meta.parse(_fused_pairs(expr))
 
 macro fused_multibroadcast(expr)
-	esc(fused_multibroadcast("MultiBroadcastFusion.FusedMultiBroadcast", expr))
+    esc(fused_multibroadcast("MultiBroadcastFusion.FusedMultiBroadcast", expr))
 end
 
 macro fused_multibroadcast(fmb, expr)
-	esc(fused_multibroadcast(fmb, expr))
+    esc(fused_multibroadcast(fmb, expr))
 end
 fused_multibroadcast(fmb, expr::Expr) =
-	Meta.parse("$(fmb)($(_fused_pairs(expr)))")
+    Meta.parse("$(fmb)($(_fused_pairs(expr)))")
 
 function build_expr(s::String, code_remain)
-	n_subs = count("%", s)
-	if n_subs > 0
-		while n_subs > 0
-			regex = r"%[0-9]"
-			m = match(regex, s)
-			smatch = m.match
-			j = Meta.parse(smatch[2:end])
-			s = replace(s, smatch => string(code_remain[j]))
-			n_subs = count("%", s)
-		end
-	else
-		return s
-	end
-	return s
+    n_subs = count("%", s)
+    if n_subs > 0
+        while n_subs > 0
+            regex = r"%[0-9]"
+            m = match(regex, s)
+            smatch = m.match
+            j = Meta.parse(smatch[2:end])
+            s = replace(s, smatch => string(code_remain[j]))
+            n_subs = count("%", s)
+        end
+    else
+        return s
+    end
+    return s
 end
 
 build_expr(code::Vector) = build_expr(string(code[end]), code)
 
 function code_lowered_single_expression(expr)
-	code_lowered = Base.Meta.lower(Main, expr)
-	code_info = code_lowered.args[1]
-	code = code_info.code # vector
-	s = build_expr(code)
-	if startswith(s, "return ")
-		s = replace(s, "return " => "")
-	end
-	return Base.Meta.parse(s)
+    code_lowered = Base.Meta.lower(Main, expr)
+    code_info = code_lowered.args[1]
+    code = code_info.code # vector
+    s = build_expr(code)
+    if startswith(s, "return ")
+        s = replace(s, "return " => "")
+    end
+    return Base.Meta.parse(s)
 end

test/bm_fused_reads_vs_hard_coded.jl

@@ -7,16 +7,16 @@ include("utils.jl")
 perf_kernel_hard_coded!(X, Y) = perf_kernel_hard_coded!(X, Y, device(X.x1))
 
 function perf_kernel_hard_coded!(X, Y, ::CPU)
-    (;x1,x2,x3,x4,x5,x6,x7,x8,x9,x10) = X
-    (;y1,y2,y3,y4,y5,y6,y7,y8,y9,y10) = Y
+    (; x1, x2, x3, x4, x5, x6, x7, x8, x9, x10) = X
+    (; y1, y2, y3, y4, y5, y6, y7, y8, y9, y10) = Y
     @inbounds for i in eachindex(x1)
-        y1[i] = x1[i]+x2[i]+x3[i]+x4[i]
-        y2[i] = x2[i]+x3[i]+x4[i]+x5[i]
-        y3[i] = x3[i]+x4[i]+x5[i]+x6[i]
-        y4[i] = x4[i]+x5[i]+x6[i]+x7[i]
-        y5[i] = x5[i]+x6[i]+x7[i]+x8[i]
-        y6[i] = x6[i]+x7[i]+x8[i]+x9[i]
-        y7[i] = x7[i]+x8[i]+x9[i]+x10[i]
+        y1[i] = x1[i] + x2[i] + x3[i] + x4[i]
+        y2[i] = x2[i] + x3[i] + x4[i] + x5[i]
+        y3[i] = x3[i] + x4[i] + x5[i] + x6[i]
+        y4[i] = x4[i] + x5[i] + x6[i] + x7[i]
+        y5[i] = x5[i] + x6[i] + x7[i] + x8[i]
+        y6[i] = x6[i] + x7[i] + x8[i] + x9[i]
+        y7[i] = x7[i] + x8[i] + x9[i] + x10[i]
     end
 end
 function perf_kernel_hard_coded!(X, Y, ::GPU)
@@ -25,21 +25,25 @@ function perf_kernel_hard_coded!(X, Y, ::GPU)
     max_threads = 256 # can be higher if conditions permit
     nthreads = min(max_threads, nitems)
     nblocks = cld(nitems, nthreads)
-    CUDA.@cuda threads = (nthreads) blocks = (nblocks) knl_multi_copyto_hard_coded!(X, Y, Val(nitems))
+    CUDA.@cuda threads = (nthreads) blocks = (nblocks) knl_multi_copyto_hard_coded!(
+        X,
+        Y,
+        Val(nitems),
+    )
 end
 function knl_multi_copyto_hard_coded!(X, Y, ::Val{nitems}) where {nitems}
-    (;x1,x2,x3,x4,x5,x6,x7,x8,x9,x10) = X
-    (;y1,y2,y3,y4,y5,y6,y7,y8,y9,y10) = Y
+    (; x1, x2, x3, x4, x5, x6, x7, x8, x9, x10) = X
+    (; y1, y2, y3, y4, y5, y6, y7, y8, y9, y10) = Y
     gidx = CUDA.threadIdx().x + (CUDA.blockIdx().x - 1) * CUDA.blockDim().x
     if gidx < nitems
         idx = gidx
-        y1[idx] = x1[idx]+x2[idx]+x3[idx]+x4[idx]
-        y2[idx] = x2[idx]+x3[idx]+x4[idx]+x5[idx]
-        y3[idx] = x3[idx]+x4[idx]+x5[idx]+x6[idx]
-        y4[idx] = x4[idx]+x5[idx]+x6[idx]+x7[idx]
-        y5[idx] = x5[idx]+x6[idx]+x7[idx]+x8[idx]
-        y6[idx] = x6[idx]+x7[idx]+x8[idx]+x9[idx]
-        y7[idx] = x7[idx]+x8[idx]+x9[idx]+x10[idx]
+        y1[idx] = x1[idx] + x2[idx] + x3[idx] + x4[idx]
+        y2[idx] = x2[idx] + x3[idx] + x4[idx] + x5[idx]
+        y3[idx] = x3[idx] + x4[idx] + x5[idx] + x6[idx]
+        y4[idx] = x4[idx] + x5[idx] + x6[idx] + x7[idx]
+        y5[idx] = x5[idx] + x6[idx] + x7[idx] + x8[idx]
+        y6[idx] = x6[idx] + x7[idx] + x8[idx] + x9[idx]
+        y7[idx] = x7[idx] + x8[idx] + x9[idx] + x10[idx]
     end
     return nothing
 end
@@ -48,37 +52,37 @@ end
 
 has_cuda = CUDA.has_cuda()
 AType = has_cuda ? CUDA.CuArray : Array
-arr_size = (prod((50,5,5,6,50)),)
+arr_size = (prod((50, 5, 5, 6, 50)),)
 # arr_size = (50,5,5,6,50)
 X = get_arrays(:x, arr_size, AType);
 Y = get_arrays(:y, arr_size, AType);
 
 function perf_kernel_unfused!(X, Y)
-    (;x1,x2,x3,x4,x5,x6,x7,x8,x9,x10) = X
-    (;y1,y2,y3,y4,y5,y6,y7,y8,y9,y10) = Y
+    (; x1, x2, x3, x4, x5, x6, x7, x8, x9, x10) = X
+    (; y1, y2, y3, y4, y5, y6, y7, y8, y9, y10) = Y
     # 7 writes; 10 unique reads
     # 7 writes; 28 reads including redundant ones
-    @. y1 = x1+x2+x3+x4
-    @. y2 = x2+x3+x4+x5
-    @. y3 = x3+x4+x5+x6
-    @. y4 = x4+x5+x6+x7
-    @. y5 = x5+x6+x7+x8
-    @. y6 = x6+x7+x8+x9
-    @. y7 = x7+x8+x9+x10
+    @. y1 = x1 + x2 + x3 + x4
+    @. y2 = x2 + x3 + x4 + x5
+    @. y3 = x3 + x4 + x5 + x6
+    @. y4 = x4 + x5 + x6 + x7
+    @. y5 = x5 + x6 + x7 + x8
+    @. y6 = x6 + x7 + x8 + x9
+    @. y7 = x7 + x8 + x9 + x10
     return nothing
 end
 
 function perf_kernel_fused!(X, Y)
-    (;x1,x2,x3,x4,x5,x6,x7,x8,x9,x10) = X;
-    (;y1,y2,y3,y4,y5,y6,y7,y8,y9,y10) = Y;
+    (; x1, x2, x3, x4, x5, x6, x7, x8, x9, x10) = X
+    (; y1, y2, y3, y4, y5, y6, y7, y8, y9, y10) = Y
     MBF.@fused begin
-        @. y1 = x1+x2+x3+x4
-        @. y2 = x2+x3+x4+x5
-        @. y3 = x3+x4+x5+x6
-        @. y4 = x4+x5+x6+x7
-        @. y5 = x5+x6+x7+x8
-        @. y6 = x6+x7+x8+x9
-        @. y7 = x7+x8+x9+x10
+        @. y1 = x1 + x2 + x3 + x4
+        @. y2 = x2 + x3 + x4 + x5
+        @. y3 = x3 + x4 + x5 + x6
+        @. y4 = x4 + x5 + x6 + x7
+        @. y5 = x5 + x6 + x7 + x8
+        @. y6 = x6 + x7 + x8 + x9
+        @. y7 = x7 + x8 + x9 + x10
     end
 end
 

test/bm_fused_shared_reads.jl

@@ -5,41 +5,43 @@ using Revise; include(joinpath("test", "fused.jl"))
 include("utils.jl")
 
 function perf_kernel_shared_reads_unfused!(X, Y)
-    (;x1,x2,x3,x4,x5,x6,x7,x8,x9,x10) = X
-    (;y1,y2,y3,y4,y5,y6,y7,y8,y9,y10) = Y
-    @. y1 = x1+x2+x3+x4
-    @. y2 = x2+x3+x4+x5
-    @. y3 = x3+x4+x5+x6
-    @. y4 = x4+x5+x6+x7
-    @. y5 = x5+x6+x7+x8
-    @. y6 = x6+x7+x8+x9
-    @. y7 = x7+x8+x9+x10
+    (; x1, x2, x3, x4, x5, x6, x7, x8, x9, x10) = X
+    (; y1, y2, y3, y4, y5, y6, y7, y8, y9, y10) = Y
+    @. y1 = x1 + x2 + x3 + x4
+    @. y2 = x2 + x3 + x4 + x5
+    @. y3 = x3 + x4 + x5 + x6
+    @. y4 = x4 + x5 + x6 + x7
+    @. y5 = x5 + x6 + x7 + x8
+    @. y6 = x6 + x7 + x8 + x9
+    @. y7 = x7 + x8 + x9 + x10
 end
 
 function perf_kernel_shared_reads_fused!(X, Y)
-    (;x1,x2,x3,x4,x5,x6,x7,x8,x9,x10) = X
-    (;y1,y2,y3,y4,y5,y6,y7,y8,y9,y10) = Y
+    (; x1, x2, x3, x4, x5, x6, x7, x8, x9, x10) = X
+    (; y1, y2, y3, y4, y5, y6, y7, y8, y9, y10) = Y
     MBF.@fused begin
-        @. y1 = x1+x2+x3+x4
-        @. y2 = x2+x3+x4+x5
-        @. y3 = x3+x4+x5+x6
-        @. y4 = x4+x5+x6+x7
-        @. y5 = x5+x6+x7+x8
-        @. y6 = x6+x7+x8+x9
-        @. y7 = x7+x8+x9+x10
+        @. y1 = x1 + x2 + x3 + x4
+        @. y2 = x2 + x3 + x4 + x5
+        @. y3 = x3 + x4 + x5 + x6
+        @. y4 = x4 + x5 + x6 + x7
+        @. y5 = x5 + x6 + x7 + x8
+        @. y6 = x6 + x7 + x8 + x9
+        @. y7 = x7 + x8 + x9 + x10
     end
 end
 
 has_cuda = CUDA.has_cuda()
 AType = has_cuda ? CUDA.CuArray : Array
-arr_size = (prod((50,5,5,6,50)),)
+arr_size = (prod((50, 5, 5, 6, 50)),)
 X = get_arrays(:x, arr_size, AType)
 Y = get_arrays(:y, arr_size, AType)
 
 test_kernel!(;
     fused! = perf_kernel_shared_reads_fused!,
     unfused! = perf_kernel_shared_reads_unfused!,
-    X, Y)
+    X,
+    Y,
+)
 # Compile
 perf_kernel_shared_reads_unfused!(X, Y)
 perf_kernel_shared_reads_fused!(X, Y)