Merge pull request #7 from foldfelis/gpu

support GPU

Project.toml

@@ -4,19 +4,25 @@ authors = ["JingYu Ning <[email protected]> and contributors"]
 version = "0.1.0"
 
 [deps]
+CUDA = "052768ef-5323-5732-b1bb-66c8b64840ba"
+CUDAKernels = "72cfdca4-0801-4ab0-bf6a-d52aa10adc57"
 DataDeps = "124859b0-ceae-595e-8997-d05f6a7a8dfe"
 FFTW = "7a1cc6ca-52ef-59f5-83cd-3a7055c09341"
 Fetch = "bb354801-46f6-40b6-9c3d-d42d7a74c775"
 Flux = "587475ba-b771-5e3f-ad9e-33799f191a9c"
+KernelAbstractions = "63c18a36-062a-441e-b654-da1e3ab1ce7c"
 MAT = "23992714-dd62-5051-b70f-ba57cb901cac"
 Tullio = "bc48ee85-29a4-5162-ae0b-a64e1601d4bc"
 Zygote = "e88e6eb3-aa80-5325-afca-941959d7151f"
 
 [compat]
+CUDA = "3.3"
+CUDAKernels = "0.3"
 DataDeps = "0.7"
 FFTW = "1.4"
 Fetch = "0.1"
 Flux = "0.12"
+KernelAbstractions = "0.7"
 MAT = "0.10"
 Tullio = "0.3"
 Zygote = "0.6"

docs/Manifest.toml

@@ -59,7 +59,7 @@ uuid = "ca575930-c2e3-43a9-ace4-1e988b2c1908"
 
 [[NeuralOperators]]
 path = ".."
-uuid = "9ab867d4-5049-4b07-85bc-95379d8d6d9c"
+uuid = "ea5c82af-86e5-48da-8ee1-382d6ad7af4b"
 version = "0.1.0"
 
 [[Parsers]]

docs/Project.toml

@@ -1,3 +1,3 @@
 [deps]
 Documenter = "e30172f5-a6a5-5a46-863b-614d45cd2de4"
-NeuralOperators = "9ab867d4-5049-4b07-85bc-95379d8d6d9c"
+NeuralOperators = "ea5c82af-86e5-48da-8ee1-382d6ad7af4b"

example/burgers.jl

@@ -1,14 +1,14 @@
 using NeuralOperators
 using Flux
-# using CUDA
+using CUDA
 
-# if has_cuda()
-#     @info "CUDA is on"
-#     device = gpu
-#     CUDA.allowscalar(false)
-# else
+if has_cuda()
+    @info "CUDA is on"
+    device = gpu
+    CUDA.allowscalar(false)
+else
     device = cpu
-# end
+end
 
 m = FourierNeuralOperator() |> device
 loss(𝐱, 𝐲) = sum(abs2, 𝐲 .- m(𝐱)) / size(𝐱)[end]
@@ -24,13 +24,14 @@ n_test = 40
 loader_test = Flux.DataLoader((𝐱_test, 𝐲_test), batchsize=20, shuffle=false)
 
 function loss_test()
-    l = 0
+    l = 0f0
     for (𝐱, 𝐲) in loader_test
+        𝐱, 𝐲 = device(𝐱), device(𝐲)
         l += loss(𝐱, 𝐲)
     end
     @info "loss: $(l/length(loader_test))"
 end
 
 data = [(𝐱, 𝐲) for (𝐱, 𝐲) in loader_train] |> device
 opt = Flux.Optimiser(WeightDecay(1f-4), Flux.ADAM(1f-3))
-Flux.@epochs 500 @time(Flux.train!(loss, params(m), data, opt, cb=Flux.throttle(loss_test, 10)))
+Flux.@epochs 500 @time(Flux.train!(loss, params(m), data, opt, cb=Flux.throttle(loss_test, 5)))

src/NeuralOperators.jl

@@ -6,6 +6,9 @@ module NeuralOperators
     using Flux
     using FFTW
     using Tullio
+    using CUDA
+    using CUDAKernels
+    using KernelAbstractions
     using Zygote
 
     function __init__()

src/fourier.jl

@@ -1,7 +1,6 @@
 export
     SpectralConv1d,
-    FourierOperator,
-    FNO
+    FourierOperator
 
 struct SpectralConv1d{T, S}
     weight::T
@@ -28,26 +27,31 @@ function SpectralConv1d(
 
     return Chain(
         x -> Zygote.hook(real, x),
-        SpectralConv1d(weights, in_chs, out_chs, modes, σ)
+        SpectralConv1d(weights, in_chs, out_chs, modes, σ),
     )
 end
 
 Flux.@functor SpectralConv1d
 
+t(𝐱) = @tullio 𝐱ᵀ[i, j, k] := 𝐱[j, i, k]
+ein_mul(𝐱₁, 𝐱₂) = @tullio 𝐲[m, o, b] := 𝐱₁[m, i, b] * 𝐱₂[o, i, m]
+
 function (m::SpectralConv1d)(𝐱::AbstractArray)
-    𝐱_fft = fft(𝐱, 2) # [in_chs, x, batch]
-    𝐱_selected = 𝐱_fft[:, 1:m.modes, :] # [in_chs, modes, batch]
+    𝐱ᵀ = t(𝐱) # [x, in_chs, batch] <- [in_chs, x, batch]
+    𝐱_fft = fft(𝐱ᵀ, 1) # [x, in_chs, batch]
+    𝐱_selected = 𝐱_fft[1:m.modes, :, :] # [modes, in_chs, batch]
 
-    # [out_chs, modes, batch] <- [in_chs, modes, batch] [out_chs, in_chs, modes]
-    @tullio 𝐱_weighted[o, m, b] := 𝐱_selected[i, m, b] * m.weight[o, i, m]
+    # [modes, out_chs, batch] <- [modes, in_chs, batch] * [out_chs, in_chs, modes]
+    𝐱_weighted = ein_mul(𝐱_selected, m.weight)
 
-    s = size(𝐱_weighted)
-    d = size(𝐱, 2) - m.modes
-    𝐱_padded = cat(𝐱_weighted, zeros(ComplexF32, s[1], d, s[3:end]...), dims=2)
+    s = size(𝐱_weighted)[2:end]
+    d = size(𝐱ᵀ, 1) - m.modes
+    𝐱_padded = cat(𝐱_weighted, zeros(ComplexF32, d, s...), dims=1)
 
-    𝐱_out = ifft(𝐱_padded, 2)
+    𝐱_out = ifft(𝐱_padded, 1) # [x, out_chs, batch]
+    𝐱_outᵀ = t(𝐱_out) # [out_chs, x, batch] <- [x, out_chs, batch]
 
-    return m.σ.(real(𝐱_out))
+    return m.σ.(real(𝐱_outᵀ))
 end
 
 function FourierOperator(