Daniel/master-thesis
1
0
Fork 0
Code Issues 3 Pull Requests Actions Packages Projects 1 Releases Wiki Activity

benchmarking: fixed bugs introduced by modification of transpiler
Some checks are pending
CI / Julia ${{ matrix.version }} - ${{ matrix.os }} - ${{ matrix.arch }} - ${{ github.event_name }} (x64, ubuntu-latest, 1.6) (push) Waiting to run
Details
CI / Julia ${{ matrix.version }} - ${{ matrix.os }} - ${{ matrix.arch }} - ${{ github.event_name }} (x64, ubuntu-latest, pre) (push) Waiting to run
Details
CI / Julia ${{ matrix.version }} - ${{ matrix.os }} - ${{ matrix.arch }} - ${{ github.event_name }} (x64, ubuntu-latest, 1.10) (push) Waiting to run
Details

Browse Source
This commit is contained in:
Daniel
2025-05-19 12:29:05 +02:00
parent e29199d865
commit a9ffd5da63
2 changed files with 5 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
package/src/ExpressionExecutorCuda.jl
View File

@ -61,14 +61,13 @@ function evaluate_gpu(expressions::Vector{Expr}, X::Matrix{Float32}, p::Vector{V
@inbounds Threads.@threads for i in eachindex(expressions) @inbounds Threads.@threads for i in eachindex(expressions)
ex = ExpressionProcessing.expr_to_postfix(expressions[i]) ex = ExpressionProcessing.expr_to_postfix(expressions[i])
ptxKernels[i] = Transpiler.transpile(ex, variableSetSize, largestParameterSetSize, numVariableSets, i-1, kernelName) # i-1 because julia is 1-based but PTX needs 0-based indexing ptxKernels[i] = Transpiler.transpile(ex, variableSetSize, largestParameterSetSize, numVariableSets, i-1, kernelName) # i-1 because julia is 1-based but PTX needs 0-based indexing
# compiledKernels[i] = Transpiler.CompileKernel(ptxKernel, kernelName)
end end
results = Matrix{Float32}(undef, numVariableSets, length(exprs)) results = Matrix{Float32}(undef, numVariableSets, length(expressions))
for i in 1:repetitions # Simulate parameter tuning -> local search (X remains the same, p gets changed in small steps and must be performed sequentially, which it is with this impl) for i in 1:repetitions # Simulate parameter tuning -> local search (X remains the same, p gets changed in small steps and must be performed sequentially, which it is with this impl)
# evaluate # evaluate
# results = Transpiler.evaluate(exprs, variables, numVariableSets, variableSetSize, p) # results = Transpiler.evaluate(exprs, variables, numVariableSets, variableSetSize, p)
results = Transpiler.evaluate(ptxKernels, variables, variableSetSize, p, kernelName) results = Transpiler.evaluate(ptxKernels, variables, numVariableSets, p, kernelName)
end end
return results return results

6
package/src/Transpiler.jl
View File

@ -33,20 +33,20 @@ function evaluate(expressions::Vector{ExpressionProcessing.PostfixType}, cudaVar
end end
" "
A simplified version of the evaluate function. It takes a list of already compiled kernels to be executed. This should yield better performance, where the same expressions should be evaluated multiple times i.e. for parameter optimisation. A simplified version of the evaluate function. It takes a list of already transpiled kernels to be executed. This should yield better performance, where the same expressions should be evaluated multiple times i.e. for parameter optimisation.
" "
function evaluate(kernels::Vector{String}, cudaVars::CuArray{Float32}, nrOfVariableSets::Integer, parameters::Vector{Vector{Float32}}, kernelName::String)::Matrix{Float32} function evaluate(kernels::Vector{String}, cudaVars::CuArray{Float32}, nrOfVariableSets::Integer, parameters::Vector{Vector{Float32}}, kernelName::String)::Matrix{Float32}
cudaParams = Utils.create_cuda_array(parameters, NaN32) # maybe make constant (see PerformanceTests.jl for more info) cudaParams = Utils.create_cuda_array(parameters, NaN32) # maybe make constant (see PerformanceTests.jl for more info)
# each expression has nr. of variable sets (nr. of columns of the variables) results and there are n expressions # each expression has nr. of variable sets (nr. of columns of the variables) results and there are n expressions
cudaResults = CuArray{Float32}(undef, nrOfVariableSets, length(expressions)) cudaResults = CuArray{Float32}(undef, nrOfVariableSets, length(kernels))
threads = min(nrOfVariableSets, 256) threads = min(nrOfVariableSets, 256)
blocks = cld(nrOfVariableSets, threads) blocks = cld(nrOfVariableSets, threads)
@inbounds Threads.@threads for i in eachindex(kernels) @inbounds Threads.@threads for i in eachindex(kernels)
compiledKernel = CompileKernel(kernel[i], kernelName) compiledKernel = CompileKernel(kernels[i], kernelName)
cudacall(compiledKernel, (CuPtr{Float32},CuPtr{Float32},CuPtr{Float32}), cudaVars, cudaParams, cudaResults; threads=threads, blocks=blocks) cudacall(compiledKernel, (CuPtr{Float32},CuPtr{Float32},CuPtr{Float32}), cudaVars, cudaParams, cudaResults; threads=threads, blocks=blocks)
end end