Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

add depthwise_conv* overloads for CUDA #22

Open
wants to merge 2 commits into
base: master
Choose a base branch
from
Open

add depthwise_conv* overloads for CUDA #22

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
5343eca
add depthwise_conv* overloads for CUDA
DhairyaLGandhi Jul 17, 2021
c1c6154
Merge branch 'master' into dg/depth
DhairyaLGandhi Jul 18, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
37 changes: 26 additions & 11 deletions src/cudnn/conv.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@

using NNlib: DenseConvDims
using NNlib: DenseConvDims, DepthwiseConvDims
import NNlib: conv!, ∇conv_filter!, ∇conv_data!, conv_bias_act!
import NNlib: depthwise_conv!, ∇depthwise_conv_filter!, ∇depthwise_conv_data!

using CUDA.CUDNN: scalingParameter, CUDNN_CONVOLUTION, convdims,
cudnnConvolutionDescriptor, cudnnConvolutionBwdDataAlgoPerf,
Expand All @@ -10,8 +10,8 @@ using CUDA.CUDNN: scalingParameter, CUDNN_CONVOLUTION, convdims,

const CUDNNFloat = Union{Float16,Float32,Float64}

function cudnnConvolutionDescriptor(cdims::DenseConvDims, x::DenseCuArray{T}) where T
mode=(NNlib.flipkernel(cdims) ? CUDNN_CROSS_CORRELATION : CUDNN_CONVOLUTION)
function cudnnConvolutionDescriptor(cdims::ConvDims, x::DenseCuArray{T}) where T
mode = (NNlib.flipkernel(cdims) ? CUDNN_CROSS_CORRELATION : CUDNN_CONVOLUTION)
cudnnConvolutionDescriptor(convdims(nnlibPadding(cdims),size(x),0),
convdims(NNlib.stride(cdims),size(x),1),
convdims(NNlib.dilation(cdims),size(x),1),
Expand All @@ -22,8 +22,8 @@ function cudnnConvolutionDescriptor(cdims::DenseConvDims, x::DenseCuArray{T}) wh
Cint(NNlib.groupcount(cdims)))
end

function conv!(y::DenseCuArray{T}, x::DenseCuArray{T}, w::DenseCuArray{T}, cdims::DenseConvDims;
alpha=1, beta=0, algo=-1) where T<:CUDNNFloat
function conv!(y::DenseCuArray{T}, x::DenseCuArray{T}, w::DenseCuArray{T}, cdims::ConvDims;
alpha = 1, beta = 0, algo = -1) where T<:CUDNNFloat
if cudnnversion() < v"6"
all(x -> x == 1, dilation(cdims)) || error("Only dilation = 1 is supported in cuDNN version < 6")
end
Expand All @@ -34,9 +34,9 @@ function conv!(y::DenseCuArray{T}, x::DenseCuArray{T}, w::DenseCuArray{T}, cdims
cudnnConvolutionForward!(y, w, x, d; alpha, beta, z=y)
end

function conv_bias_act!(y::DenseCuArray{T}, x::DenseCuArray{T}, w::DenseCuArray{T},
cdims::DenseConvDims, bias::DenseCuArray{T}, σ=identity;
z::DenseCuArray{T}=y, alpha=1, beta=0, algo=-1) where T<:CUDNNFloat
function conv_bias_act!(y::DenseCuArray{T}, x::DenseCuArray{T}, w::DenseCuArray{T},
cdims::ConvDims, bias::DenseCuArray{T}, σ = identity;
z::DenseCuArray{T} = y, alpha = 1, beta = 0, algo = -1) where T <: CUDNNFloat
if cudnnversion() < v"6"
all(x -> x == 1, dilation(cdims)) || error("Only dilation = 1 is supported in cuDNN version < 6")
end
Expand All @@ -54,7 +54,7 @@ function conv_bias_act!(y::DenseCuArray{T}, x::DenseCuArray{T}, w::DenseCuArray{
end

function ∇conv_data!(dx::DenseCuArray{T}, dy::DenseCuArray{T}, w::DenseCuArray{T},
cdims::DenseConvDims; alpha=1, beta=0, algo=-1) where T<:CUDNNFloat
cdims::ConvDims; alpha = 1, beta = 0, algo = -1) where T <: CUDNNFloat
if cudnnversion() < v"6"
all(x -> x == 1, dilation(cdims)) || error("Only dilation = 1 is supported in cuDNN version < 6")
end
Expand All @@ -72,7 +72,7 @@ function ∇conv_data!(dx::DenseCuArray{T}, dy::DenseCuArray{T}, w::DenseCuArray
end

function ∇conv_filter!(dw::DenseCuArray{T}, x::DenseCuArray{T}, dy::DenseCuArray{T},
cdims::DenseConvDims; alpha=1, beta=0, algo=-1) where T<:CUDNNFloat
cdims::ConvDims; alpha = 1, beta = 0, algo = -1) where T <: CUDNNFloat
if cudnnversion() < v"6"
all(x -> x == 1, dilation(cdims)) || error("Only dilation = 1 is supported in cuDNN version < 6")
end
Expand All @@ -95,3 +95,18 @@ function ∇conv_bias!(db::DenseCuArray{T}, dy::DenseCuArray{T}; alpha=1, beta=0
cudnnConvolutionBackwardBias(handle(), alpha, yDesc, dy, beta, bDesc, db)
return db
end

function depthwise_conv!(y::DenseCuArray{T}, x::DenseCuArray{T}, w::DenseCuArray{T}, cdims::DepthwiseConvDims;
alpha = 1, beta = 0, algo = -1) where T <: CUDNNFloat
conv!(y, x, w, cims; alpha, beta, algo)
end

function ∇depthwise_conv_filter!(dw::DenseCuArray{T}, x::DenseCuArray{T}, dy::DenseCuArray{T},
cdims::ConvDims; alpha = 1, beta = 0, algo = -1) where T <: CUDNNFloat
∇conv_filter!(dw, x, dy, cdims; alpha, beta, algo)
end

function ∇depthwise_conv_data!(dx::DenseCuArray{T}, dy::DenseCuArray{T}, w::DenseCuArray{T},
cdims::ConvDims; alpha = 1, beta = 0, algo = -1) where T <: CUDNNFloat
∇conv_data!(dx, dy, w, cdims; alpha, beta, algo)
end
Comment on lines +99 to +112
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

these don't have to be cuda specific, we can add them to NNlib and remove the specific implementations (after a performance comparison)

Copy link
Member Author

@DhairyaLGandhi DhairyaLGandhi Jul 17, 2021 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Add what to nnlib, sorry? This package is specific to GPU functionality.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

exactly these methods, with AbstractArray arguments, i.e. fallback on conv

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Umm, we probably want to retain the cpu kernels anyway. Without explicitly having and launching Julia with many threads, grouped convolutions would scale with the number of groups.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

this would be true for any implementation, specialized or not

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more. 

julia> x′ = rand(Float32, 28, 28, 4, 2);

julia> w′ = rand(Float32, 3, 3, 4, 30);

julia> cdims = DenseConvDims(x′, w′, groups = 4)

julia> @btime conv($x′, $w′, $cdims);
 362.792 μs (86 allocations: 736.36 KiB) # -t1
 236.368 μs (94 allocations: 831.89 KiB) # -t2
 232.137 μs (94 allocations: 831.89 KiB) # -t4

julia> @btime depthwiseconv($x′, $(permutedims(w′, (1,2,4,3))));
 348.914 μs (42 allocations: 731.03 KiB) # -t1
 156.558 μs (47 allocations: 826.53 KiB) # -t2
 161.059 μs (47 allocations: 826.53 KiB) # -t4

This is with https://github.com/DhairyaLGandhi/NNlib.jl#dg/g2 which has a couple of fixes pending a PR.