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

benchmarking: tuned blocksizes; slightly improved performance; mostly improved standard deviation
Some checks are pending
CI / Julia ${{ matrix.version }} - ${{ matrix.os }} - ${{ matrix.arch }} - ${{ github.event_name }} (x64, ubuntu-latest, 1.10) (push) Waiting to run
Details
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

Browse Source
This commit is contained in:
Wiplinger Daniel - s2310454043
2025-04-12 13:20:50 +02:00
parent c6e2ce47aa
commit 101ccef67b
5 changed files with 8 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
package/src/Transpiler.jl
View File

@ -73,7 +73,7 @@ function evaluate(expressions::Vector{Expr}, variables::Matrix{Float32}, paramet
# execute each kernel (also try doing this with Threads.@threads. Since we can have multiple grids, this might improve performance)
for kernel in kernels
# config = launch_configuration(kernels[i])
threads = min(variableCols, 256)
threads = min(variableCols, 96)
blocks = cld(variableCols, threads)
cudacall(kernel, (CuPtr{Float32},CuPtr{Float32},CuPtr{Float32}), cudaVars, cudaParams, cudaResults; threads=threads, blocks=blocks)