pt-three-ways
pt-three-ways copied to clipboard
mattgodbolt
Sharing benchmarks
Not an issue but here my results. A bit more consistent as opposed to what you mentioned in the talk.
Intel(R) Core(TM) i7-7700HQ CPU @ 2.80GHz (governor performance) Linux 5.1.11-300.fc30.x86_64 gcc 9.1.1
| Test | 1 Thread | 6 Threads |
|---|---|---|
| cornell | ||
| oo | 72 s | 26 s |
| fp | 76 s | 28 s |
| dod | 64 s | 22 s |
| owl | ||
| oo | 96 s | 21 s |
| fp | 117 s | 30 s |
| dod | 66 s | 15 s |
| suzanne | ||
| oo | 126 s | 193 s |
| fp | 102 s | 173 s |
| dod | 78 s | 129 s |
P.S. You should add the link to your talk to the Readme, and maybe instruction on how to build. This worked for me (even without using clion):
git clone https://github.com/mattgodbolt/pt-three-ways.git
cd pt-three-ways
mkdir cmake-build-release
cd cmake-build-release
cmake .. -DCMAKE_BUILD_TYPE=Release
cd ..
scripts/bench-6t-cornell.sh
EDIT: suzanne st and 6t values were. Maybe we should use same sampling?
My results were also a bit more consistent.
Intel(R) Core(TH) i5-4590 CPU @ 3.30GHz Windows 10 Education Version 1803 MSVC 19.23.2806.4
| Test | 1 Thread | 6 Threads | 6 Threads |
|---|---|---|---|
| cornell | 32 spp | 48 spp | 32 spp |
| oo | 113s | 42s | 29s |
| fp | 120s | 49s | 36s |
| dod | 88s | 32s | 22s |
| bbc-owl | 128 spp | 48 spp | 128 spp |
| oo | 161s | 14s | 38s |
| fp | 365s | 40s | 100s |
| dod | 97s | 9s | 23s |
| suzanne | 8 spp | 48 spp | 8 spp |
| oo | 196s | 315s | 58s |
| fp | 202s | 359s | 84s |
| dod | 125s | 197s | 36s |
Thanks both. The code as currently in the repo has the "fix" applied: there's a branch somewhere with the branch predictor thing still problematic. I will collate these and update the README.
