Skip to content

Commit

Permalink
implemented wrappers for Find* and Get*
Browse files Browse the repository at this point in the history
  • Loading branch information
@gartangh
gartangh committed Jul 27, 2020
1 parent a73a069 commit 904f3fb
Show file tree
Hide file tree
Showing 2 changed files with 229 additions and 2 deletions.
219 changes: 219 additions & 0 deletions lib/cudnn/conv.jl
View file
Expand Up @@ -59,6 +59,17 @@ end

# wrappers

# Forward

function cudnnGetConvolutionForwardAlgorithmMaxCount()
count=@argout(
cudnnGetConvolutionForwardAlgorithmMaxCount(
handle(),
out(Ref{Cint}()))
)[]
return count
end

function cudnnGetConvolutionForwardAlgorithm(y::CuArray{T,N}, x::CuArray{T,N}, w::CuArray{T,N},
cdims::DenseConvDims; preference=1, memoryLimitInBytes=1<<32) where {T,N}
algo=@argout(
Expand All @@ -73,6 +84,64 @@ function cudnnGetConvolutionForwardAlgorithm(y::CuArray{T,N}, x::CuArray{T,N}, w
return algo
end

function cudnnGetConvolutionForwardAlgorithm_v7(y::CuArray{T,N}, x::CuArray{T,N}, w::CuArray{T,N},
cdims::DenseConvDims; count=-1) where {T,N}
if count < 0
count = cudnnGetConvolutionForwardAlgorithmMaxCount()
end
perfResults = Array{cudnnConvolutionFwdAlgoPerf_t, 1}(undef, count)
returnedAlgoCount=@argout(
cudnnGetConvolutionForwardAlgorithm_v7(
handle(), TensorDesc(x),
FilterDesc(w), ConvDesc(T, cdims),
TensorDesc(y),
Cint(count),
out(Ref{Cint}()),
perfResults)
)[]
return returnedAlgoCount, perfResults
end

function cudnnFindConvolutionForwardAlgorithm(y::CuArray{T,N}, x::CuArray{T,N}, w::CuArray{T,N},
cdims::DenseConvDims; count=-1) where {T,N}
if count < 0
count = cudnnGetConvolutionForwardAlgorithmMaxCount()
end
perfResults = Array{cudnnConvolutionFwdAlgoPerf_t, 1}(undef, count)
returnedAlgoCount=@argout(
cudnnFindConvolutionForwardAlgorithm(
handle(), TensorDesc(x),
FilterDesc(w), ConvDesc(T, cdims),
TensorDesc(y),
Cint(count),
out(Ref{Cint}()),
perfResults)
)[]
return returnedAlgoCount, perfResults
end

function cudnnFindConvolutionForwardAlgorithmEx(y::CuArray{T,N}, x::CuArray{T,N}, w::CuArray{T,N},
cdims::DenseConvDims; count=-1, workspacesize=1<<32) where {T,N}
if count < 0
count = cudnnGetConvolutionForwardAlgorithmMaxCount()
end
@workspace size=workspacesize workspace->begin
perfResults = Array{cudnnConvolutionFwdAlgoPerf_t, 1}(undef, count)
returnedAlgoCount=@argout(
cudnnFindConvolutionForwardAlgorithmEx(
handle(), TensorDesc(x), x,
FilterDesc(w), w, ConvDesc(T, cdims),
TensorDesc(y), y,
Cint(count),
out(Ref{Cint}()),
perfResults,
workspace,
workspacesize)
)[]
return returnedAlgoCount, perfResults
end
end

function cudnnConvolutionForward(y::CuArray{T,N}, x::CuArray{T,N}, w::CuArray{T,N},
cdims::DenseConvDims; algo=0, alpha=1, beta=0) where {T,N}
@workspace size=@argout(
Expand Down Expand Up @@ -110,6 +179,78 @@ function cudnnConvolutionBiasActivationForward(y::CuArray{T,N}, x::CuArray{T,N},
return y
end

# Backward data

function cudnnGetConvolutionBackwardDataAlgorithmMaxCount()
count=@argout(
cudnnGetConvolutionBackwardDataAlgorithmMaxCount(
handle(),
out(Ref{Cint}()))
)[]
return count
end

function cudnnGetConvolutionBackwardDataAlgorithm(dx::CuArray{T,N}, w::CuArray{T,N}, dy::CuArray{T,N},
cdims::DenseConvDims; preference=1, memoryLimitInBytes=1<<32) where {T,N}
algo=@argout(
cudnnGetConvolutionBackwardDataAlgorithm(
handle(), FilterDesc(w), TensorDesc(dy), ConvDesc(T, cdims),
TensorDesc(dx), cudnnConvolutionBwdDataPreference_t(preference),
Csize_t(memoryLimitInBytes), out(Ref{cudnnConvolutionBwdDataAlgo_t}()))
)[]
return algo
end

function cudnnGetConvolutionBackwardDataAlgorithm_v7(dx::CuArray{T,N}, w::CuArray{T,N}, dy::CuArray{T,N},
cdims::DenseConvDims; count=-1) where {T,N}
if count < 0
count = cudnnGetConvolutionBackwardDataAlgorithmMaxCount()
end
perfResults = Array{cudnnConvolutionBwdDataAlgoPerf_t, 1}(undef, count)
returnedAlgoCount=@argout(
cudnnGetConvolutionBackwardDataAlgorithm_v7(
handle(), FilterDesc(w), TensorDesc(dy),
ConvDesc(T, cdims), TensorDesc(dx),
Cint(count),
out(Ref{Cint}()), perfResults)
)[]
return returnedAlgoCount, perfResults
end

function cudnnFindConvolutionBackwardDataAlgorithm(dx::CuArray{T,N}, w::CuArray{T,N}, dy::CuArray{T,N},
cdims::DenseConvDims; count=-1) where {T,N}
if count < 0
count = cudnnGetConvolutionBackwardDataAlgorithmMaxCount()
end
perfResults = Array{cudnnConvolutionBwdDataAlgoPerf_t, 1}(undef, count)
returnedAlgoCount=@argout(
cudnnFindConvolutionBackwardDataAlgorithm(
handle(), FilterDesc(w), TensorDesc(dy),
ConvDesc(T, cdims), TensorDesc(dx),
Cint(count),
out(Ref{Cint}()), perfResults)
)[]
return returnedAlgoCount, perfResults
end

function cudnnFindConvolutionBackwardDataAlgorithmEx(dx::CuArray{T,N}, w::CuArray{T,N}, dy::CuArray{T,N},
cdims::DenseConvDims; count=-1, workspacesize=1<<32) where {T,N}
if count < 0
count = cudnnGetConvolutionBackwardDataAlgorithmMaxCount()
end
perfResults = Array{cudnnConvolutionBwdDataAlgoPerf_t, 1}(undef, count)
returnedAlgoCount=@argout(
cudnnFindConvolutionBackwardDataAlgorithmEx(
handle(), FilterDesc(w), w, TensorDesc(dy), y,
ConvDesc(T, cdims), TensorDesc(dx), x,
Cint(count),
out(Ref{Cint}()),
perfResults, workspace,
workspacesize)
)[]
return returnedAlgoCount, perfResults
end

function cudnnConvolutionBackwardData(dx::CuArray{T,N}, w::CuArray{T,N}, dy::CuArray{T,N},
cdims::DenseConvDims; algo=0, alpha=1, beta=0) where {T,N}
@workspace size=@argout(
Expand All @@ -129,6 +270,82 @@ function cudnnConvolutionBackwardData(dx::CuArray{T,N}, w::CuArray{T,N}, dy::CuA
return dx
end

# Backward filter

function cudnnGetConvolutionBackwardFilterAlgorithmMaxCount()
count=@argout(
cudnnGetConvolutionBackwardFilterAlgorithmMaxCount(
handle(),
out(Ref{Cint}()))
)[]
return count
end

function cudnnGetConvolutionBackwardFilterAlgorithm(dw::CuArray{T,N}, x::CuArray{T,N}, dy::CuArray{T,N},
cdims::DenseConvDims; preference=1, memoryLimitInBytes=1<<32) where {T,N}
algo=@argout(
cudnnGetConvolutionBackwardFilterAlgorithm(
handle(), TensorDesc(x), TensorDesc(dy),
ConvDesc(T, cdims), FilterDesc(dw), cudnnConvolutionBwdFilterPreference_t(preference),
Csize_t(memoryLimitInBytes), out(Ref{cudnnConvolutionBwdFilterAlgo_t}()))
)[]
return algo
end

function cudnnGetConvolutionBackwardFilterAlgorithm_v7(dw::CuArray{T,N}, x::CuArray{T,N}, dy::CuArray{T,N},
cdims::DenseConvDims; count=-1) where {T,N}
if count < 0
count = cudnnGetConvolutionBackwardFilterAlgorithmMaxCount()
end
perfResults = Array{cudnnConvolutionBwdFilterAlgoPerf_t, 1}(undef, count)
returnedAlgoCount=@argout(
cudnnGetConvolutionBackwardFilterAlgorithm_v7(
handle(), TensorDesc(x), TensorDesc(dy),
ConvDesc(T, cdims), FilterDesc(dw),
Cint(count),
out(Ref{Cint}()),
perfResults)
)[]
return returnedAlgoCount, perfResults
end

function cudnnFindConvolutionBackwardFilterAlgorithm(dw::CuArray{T,N}, x::CuArray{T,N}, dy::CuArray{T,N},
cdims::DenseConvDims; count=-1) where {T,N}
if count < 0
count = cudnnGetConvolutionBackwardFilterAlgorithmMaxCount()
end
perfResults = Array{cudnnConvolutionBwdFilterAlgoPerf_t, 1}(undef, count)
returnedAlgoCount=@argout(
cudnnFindConvolutionBackwardFilterAlgorithm(
handle(), TensorDesc(x), TensorDesc(dy),
ConvDesc(T, cdims), FilterDesc(dw),
Cint(count),
out(Ref{Cint}()),
perfResults)
)[]
return returnedAlgoCount, perfResults
end

function cudnnFindConvolutionBackwardFilterAlgorithmEx(dw::CuArray{T,N}, x::CuArray{T,N}, dy::CuArray{T,N},
cdims::DenseConvDims; count=-1, workspacesize=1<<32) where {T,N}
if count < 0
count = cudnnGetConvolutionBackwardFilterAlgorithmMaxCount()
end
@workspace size=workspacesize workspace->begin
perfResults = Array{cudnnConvolutionBwdFilterAlgoPerf_t, 1}(undef, count)
returnedAlgoCount=@argout(
cudnnFindConvolutionBackwardFilterAlgorithmEx(
handle(), TensorDesc(x), x, TensorDesc(dy),
dy, ConvDesc(T, cdims), FilterDesc(dw), dw,
Cint(count),
out(Ref{Cint}()),
perfResults, workspace,
workspacesize)
)[]
return returnedAlgoCount, perfResults
end
end

function cudnnConvolutionBackwardFilter(dw::CuArray{T,N}, x::CuArray{T,N}, dy::CuArray{T,N},
cdims::DenseConvDims; algo=0, alpha=1, beta=0) where {T,N}
@workspace size=@argout(
Expand All @@ -149,6 +366,8 @@ function cudnnConvolutionBackwardFilter(dw::CuArray{T,N}, x::CuArray{T,N}, dy::C
return dw
end

# Backward bias

function cudnnConvolutionBackwardBias(db::CuArray{T,N}, dy::CuArray{T,N}; alpha=1, beta=0) where {T,N}
cudnnConvolutionBackwardBias(handle(), Ref(T(alpha)), TensorDesc(dy), dy, Ref(T(beta)), TensorDesc(db), db)
return db
Expand Down
12 changes: 10 additions & 2 deletions lib/cudnn/nnlib.jl
View file
Expand Up @@ -73,7 +73,11 @@ function conv_bias_act!(y::CuArray{T}, x::CuArray{T}, w::CuArray{T}, cdims::Dens
# only relu and identity are supported
if σ == NNlib.relu
if algo < 0
algo = UInt32(cudnnGetConvolutionForwardAlgorithm(fix1d(y), fix1d(x), fix1d(w), fix1d(cdims)))
# algo = UInt32(cudnnGetConvolutionForwardAlgorithm(fix1d(y), fix1d(x), fix1d(w), fix1d(cdims))) # will be removed in cuDNN 8
returnedAlgoCount, perfResults = cudnnGetConvolutionForwardAlgorithm_v7(fix1d(y), fix1d(x), fix1d(w), fix1d(cdims))
# returnedAlgoCount, perfResults = cudnnFindConvolutionForwardAlgorithm(fix1d(y), fix1d(x), fix1d(w), fix1d(cdims))
# returnedAlgoCount, perfResults = cudnnFindConvolutionForwardAlgorithmEx(fix1d(y), fix1d(x), fix1d(w), fix1d(cdims))
algo = UInt32(perfResults[1].algo)
end
cudnnConvolutionBiasActivationForward(fix1d(y), fix1d(x), fix1d(w), fix1d(z), fix1d(b),
fix1d(cdims), algo=algo, alpha1=alpha1, alpha2=alpha2,
Expand All @@ -97,7 +101,11 @@ function conv!(y::CuArray{T}, x::CuArray{T}, w::CuArray{T}, cdims::DenseConvDims
end

if algo < 0
algo = UInt32(cudnnGetConvolutionForwardAlgorithm(fix1d(y), fix1d(x), fix1d(w), fix1d(cdims)))
# algo = UInt32(cudnnGetConvolutionForwardAlgorithm(fix1d(y), fix1d(x), fix1d(w), fix1d(cdims))) # will be removed in cuDNN 8
returnedAlgoCount, perfResults = cudnnGetConvolutionForwardAlgorithm_v7(fix1d(y), fix1d(x), fix1d(w), fix1d(cdims))
# returnedAlgoCount, perfResults = cudnnFindConvolutionForwardAlgorithm(fix1d(y), fix1d(x), fix1d(w), fix1d(cdims))
# returnedAlgoCount, perfResults = cudnnFindConvolutionForwardAlgorithmEx(fix1d(y), fix1d(x), fix1d(w), fix1d(cdims))
algo = UInt32(perfResults[1].algo)
end

cudnnConvolutionForward(fix1d(y), fix1d(x), fix1d(w), fix1d(cdims), alpha=alpha, algo=algo)
Expand Down

0 comments on commit 904f3fb

Please sign in to comment.