Skip to content

Commit 43ef860

Browse files
KristofferCKristofferC
authored
Merge pull request #569 from JuliaDiff/kc/revert_simd
Revert SIMD.jl usage
2 parents 01a056d + 7dbd072 commit 43ef860

File tree

6 files changed

+23
-55
lines changed

6 files changed

+23
-55
lines changed

‎.github/workflows/ci.yml‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -16,6 +16,7 @@ jobs:
1616
fail-fast: false
1717
matrix:
1818
version:
19+
- '1.0'
1920
- '1'
2021
- 'nightly'
2122
os:

‎Project.toml‎

Lines changed: 1 addition & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -12,7 +12,6 @@ NaNMath = "77ba4419-2d1f-58cd-9bb1-8ffee604a2e3"
1212
Preferences = "21216c6a-2e73-6563-6e65-726566657250"
1313
Printf = "de0858da-6303-5e67-8744-51eddeeeb8d7"
1414
Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
15-
SIMD = "fdea26ae-647d-5447-a871-4b548cad5224"
1615
SpecialFunctions = "276daf66-3868-5448-9aa4-cd146d93841b"
1716
StaticArrays = "90137ffa-7385-5640-81b9-e52037218182"
1817

@@ -25,10 +24,9 @@ DiffTests = "0.0.1, 0.1"
2524
LogExpFunctions = "0.3"
2625
NaNMath = "0.2.2, 0.3"
2726
Preferences = "1"
28-
SIMD = "3"
2927
SpecialFunctions = "0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.10, 1.0"
3028
StaticArrays = "0.8.3, 0.9, 0.10, 0.11, 0.12, 1.0"
31-
julia = "1.6"
29+
julia = "1"
3230

3331
[extras]
3432
Calculus = "49dc2e85-a5d0-5ad3-a950-438e2897f1b9"

‎src/ForwardDiff.jl‎

Lines changed: 0 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -8,21 +8,13 @@ if VERSION >= v"1.6"
88
end
99
using Random
1010
using LinearAlgebra
11-
import SIMD: Vec
1211

1312
import Printf
1413
import NaNMath
1514
import SpecialFunctions
1615
import LogExpFunctions
1716
import CommonSubexpressions
1817

19-
const SIMDFloat = Union{Float64, Float32}
20-
const SIMDInt = Union{
21-
Int128, Int64, Int32, Int16, Int8,
22-
UInt128, UInt64, UInt32, UInt16, UInt8,
23-
}
24-
const SIMDType = Union{SIMDFloat, SIMDInt}
25-
2618
include("prelude.jl")
2719
include("partials.jl")
2820
include("dual.jl")

‎src/dual.jl‎

Lines changed: 0 additions & 20 deletions
Original file line numberDiff line numberDiff line change
@@ -541,16 +541,6 @@ end
541541
# fma #
542542
#-----#
543543

544-
@inline function calc_fma_xyz(x::Dual{T,V,N},
545-
y::Dual{T,V,N},
546-
z::Dual{T,V,N}) where {T, V<:SIMDFloat,N}
547-
xv, yv, zv = value(x), value(y), value(z)
548-
rv = fma(xv, yv, zv)
549-
N == 0 && return Dual{T}(rv)
550-
xp, yp, zp = Vec(partials(x).values), Vec(partials(y).values), Vec(partials(z).values)
551-
parts = Tuple(fma(xv, yp, fma(yv, xp, zp)))
552-
Dual{T}(rv, parts)
553-
end
554544
@generated function calc_fma_xyz(x::Dual{T,<:Any,N},
555545
y::Dual{T,<:Any,N},
556546
z::Dual{T,<:Any,N}) where {T,N}
@@ -593,16 +583,6 @@ end
593583
# muladd #
594584
#--------#
595585

596-
@inline function calc_muladd_xyz(x::Dual{T,V,N},
597-
y::Dual{T,V,N},
598-
z::Dual{T,V,N}) where {T, V<:SIMDType,N}
599-
xv, yv, zv = value(x), value(y), value(z)
600-
rv = muladd(xv, yv, zv)
601-
N == 0 && return Dual{T}(rv)
602-
xp, yp, zp = Vec(partials(x).values), Vec(partials(y).values), Vec(partials(z).values)
603-
parts = Tuple(muladd(xv, yp, muladd(yv, xp, zp)))
604-
Dual{T}(rv, parts)
605-
end
606586
@generated function calc_muladd_xyz(x::Dual{T,<:Any,N},
607587
y::Dual{T,<:Any,N},
608588
z::Dual{T,<:Any,N}) where {T,N}

‎src/partials.jl‎

Lines changed: 20 additions & 23 deletions
Original file line numberDiff line numberDiff line change
@@ -141,13 +141,6 @@ end
141141
@inline _mul_partials(a::Partials{0,A}, b::Partials{N,B}, afactor, bfactor) where {N,A,B} = bfactor * b
142142
@inline _mul_partials(a::Partials{N,A}, b::Partials{0,B}, afactor, bfactor) where {N,A,B} = afactor * a
143143

144-
const SIMDFloat = Union{Float64, Float32}
145-
const SIMDInt = Union{
146-
Int128, Int64, Int32, Int16, Int8,
147-
UInt128, UInt64, UInt32, UInt16, UInt8,
148-
}
149-
const SIMDType = Union{SIMDFloat, SIMDInt}
150-
151144
##################################
152145
# Generated Functions on NTuples #
153146
##################################
@@ -171,7 +164,6 @@ end
171164
@inline rand_tuple(::AbstractRNG, ::Type{Tuple{}}) = tuple()
172165
@inline rand_tuple(::Type{Tuple{}}) = tuple()
173166

174-
iszero_tuple(tup::NTuple{N,V}) where {N, V<:SIMDType} = sum(Vec(tup) != zero(V)) == 0
175167
@generated function iszero_tuple(tup::NTuple{N,V}) where {N,V}
176168
ex = Expr(:&&, [:(z == tup[$i]) for i=1:N]...)
177169
return quote
@@ -205,24 +197,29 @@ end
205197
return tupexpr(i -> :(rand(V)), N)
206198
end
207199

208-
const NT{N,T} = NTuple{N,T}
200+
@generated function scale_tuple(tup::NTuple{N}, x) where N
201+
return tupexpr(i -> :(tup[$i] * x), N)
202+
end
209203

210-
# SIMD implementation
211-
@inline add_tuples(a::NT{N,T}, b::NT{N,T}) where {N, T<:SIMDType} = Tuple(Vec(a) + Vec(b))
212-
@inline sub_tuples(a::NT{N,T}, b::NT{N,T}) where {N, T<:SIMDType} = Tuple(Vec(a) - Vec(b))
213-
@inline scale_tuple(tup::NT{N,T}, x::T) where {N, T<:SIMDType} = Tuple(Vec(tup) * x)
214-
@inline div_tuple_by_scalar(tup::NT{N,T}, x::T) where {N, T<:SIMDFloat} = Tuple(Vec(tup) / x)
215-
@inline minus_tuple(tup::NT{N,T}) where {N, T<:SIMDType} = Tuple(-Vec(tup))
216-
@inline mul_tuples(a::NT{N,T}, b::NT{N,T}, af::T, bf::T) where {N, T<:SIMDType} = Tuple(muladd(af, Vec(a), bf * Vec(b)))
204+
@generated function div_tuple_by_scalar(tup::NTuple{N}, x) where N
205+
return tupexpr(i -> :(tup[$i] / x), N)
206+
end
217207

208+
@generated function add_tuples(a::NTuple{N}, b::NTuple{N}) where N
209+
return tupexpr(i -> :(a[$i] + b[$i]), N)
210+
end
218211

219-
# Fallback implementations
220-
@generated add_tuples(a::NT{N}, b::NT{N}) where N = tupexpr(i -> :(a[$i] + b[$i]), N)
221-
@generated sub_tuples(a::NT{N}, b::NT{N}) where N = tupexpr(i -> :(a[$i] - b[$i]), N)
222-
@generated scale_tuple(tup::NT{N}, x) where N = tupexpr(i -> :(tup[$i] * x), N)
223-
@generated div_tuple_by_scalar(tup::NT{N}, x) where N = tupexpr(i -> :(tup[$i] / x), N)
224-
@generated minus_tuple(tup::NT{N}) where N = tupexpr(i -> :(-tup[$i]), N)
225-
@generated mul_tuples(a::NT{N}, b::NT{N}, af, bf) where N = tupexpr(i -> :(muladd(af, a[$i], bf * b[$i])), N)
212+
@generated function sub_tuples(a::NTuple{N}, b::NTuple{N}) where N
213+
return tupexpr(i -> :(a[$i] - b[$i]), N)
214+
end
215+
216+
@generated function minus_tuple(tup::NTuple{N}) where N
217+
return tupexpr(i -> :(-tup[$i]), N)
218+
end
219+
220+
@generated function mul_tuples(a::NTuple{N}, b::NTuple{N}, afactor, bfactor) where N
221+
return tupexpr(i -> :((afactor * a[$i]) + (bfactor * b[$i])), N)
222+
end
226223

227224
###################
228225
# Pretty Printing #

‎test/PartialsTest.jl‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -114,7 +114,7 @@ for N in (0, 3), T in (Int, Float32, Float64)
114114

115115
if N > 0
116116
@test ForwardDiff._div_partials(PARTIALS, PARTIALS2, X, Y) == ForwardDiff._mul_partials(PARTIALS, PARTIALS2, inv(Y), -X/(Y^2))
117-
@test all(isapprox.(ForwardDiff._mul_partials(PARTIALS, PARTIALS2, X, Y).values, map((a, b) -> (X * a) + (Y * b), VALUES, VALUES2)))
117+
@test ForwardDiff._mul_partials(PARTIALS, PARTIALS2, X, Y).values == map((a, b) -> (X * a) + (Y * b), VALUES, VALUES2)
118118
@test ForwardDiff._mul_partials(ZERO_PARTIALS, PARTIALS, X, Y) == Y * PARTIALS
119119
@test ForwardDiff._mul_partials(PARTIALS, ZERO_PARTIALS, X, Y) == X * PARTIALS
120120

0 commit comments

Comments
 (0)