Add benchmarks for geometry transformations (rot, flip, transpose) #525
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
mloskot
approved these changes
Oct 22, 2020
| return res; | ||
| }); | ||
|
|
||
| #if GIL_ENABLE_UNROLLED |
There was a problem hiding this comment.
We should prefix all our macros with BOOST_GIL_*, see #410
I'd suggest to rename this to BOOST_GIL_BENCHMARK_ENABLE_UNROLLED or similar.
mloskot
added
cat/performance
Both temp_red and max_color are float32_t so the result of the subraction already has the right type. This should also remove the only dependency on Boost.NumericConversion
The "cxxstd" json field is being added to each Boost library's meta json information for libraries whose minumum C++ standard compilation level is C++11 on up. The value of this field matches one of the values for 'cxxstd' in Boost.Build. The purpose of doing this is to provide information for the Boost website documentation for each library which will specify the minimum C++ standard compilation that an end-user must employ in order to use the particular library. This will aid end-users who want to know if they can successfully use a Boost library based on their C++ compiler's compilation level, without having to search the library's documentation to find this out. https://lists.boost.org/Archives/boost/2020/12/250543.php
Discussion at conan-io/conan-center-index#4151 (comment) Fixes Azure Pipelines builds for boostorg#535
Helping to investigate boostorg/build#704
…ts (boostorg#542) If `Element` is a reference, then Element v{} is ill-formed. Refines boostorg#273 which aimed to correctly value-initialize channel and pixel value members
The output type must be floating point, thus a check was added to make sure it is the case. Though I had to add specific cases for float32_t as std::is_floating_point does not consider it a floating point type The accumulate part was wrong, it multiplied by delta_t on every sum, which is wrong Use 8 way nabla compute. This is just different discretization of Laplace operator https://en.wikipedia.org/wiki/Discrete_Laplace_operator Laplace stencils are now the same as in mathematical notation The new function will provide a uniform way to generate stencils by making sure directions are indexed properly Add only required stencil points: The 5 points Laplace stencil is now adding only required points and not assuming that others are zero
A new histogram class proposed with close suport for gil image constructs. Shift the stl support implmentation to extension to serve as example for overloading fill_histogram. Add cumulative histogram and histogram normalization. Co-authored-by: debabrata1 <[email protected]>
Co-authored-by: codejaeger <dhabalm1@>
Allow non-equal dims along stack dim Width can be different in hstack, height can be different in vstack
Add basic GitHub Actions configuration based on mp11
Remove Actions jobs using GCC 4.7 and 4.8 - unsupported compilers
Run b2 with --abbreviate-paths on Windows
The -std=c++1z is broken for clang-4.0 but no need to test it
Add -mbig-obj to GCC on Windows
- That is to avoid string table overflow and file too big
Define _GLIBCXX_USE_CXX11_ABI=0 for clang 3.5, 3.6, 3.7
- Should help avoid linker error:
`undefined reference to std::ios_base::failure::failure(char const*, std::error_code const&)`
Disable certain check in algorithm_channel_relation test for clang<3.8
Support construction from step_size, step_count, and a function for angles Implement angle and radious version of Hough line transform and adds a demo with static line that goes over secondary diagonal. Implement incremental line raster Implement naive line raster Implement Bresenham line raster Leave only Bresenham line rasterization Naive and incremental algorithms were removed because they are supposed to produce the same results anyway. The reason for diverging results is inaccuracy of floating point numbers Add circle rendering through trigonometric functions, using arctan(1 / (radius + 1)) as minimal angle step. Trigonometric circle rasterizer does not follow circle equation, but still produces very round shapes. A new testing methodology needs to be devised for this rasterizer. The new version accepts start and points inclusively and tries to use canonic representation during computations. Slope decided to be is (diff_y + 1) / (diff_x + 1).
Follows: - boostorg#544 Outstanding tasks: - boostorg#549 - boostorg#548
Most uses of cxxflags are wrong. Use b2 warnings instead, e.g. warnings=pedantic
Completing boostorg#545
Patch via e-mail from @sdarwin to refine PR boostorg#719 Thank you Sam!
sdebionne
force-pushed
the
add-benchmark-geom-xform
branch
from
20a098a
to
9398fd9
Compare
This commit adds support for benchmark definition and running, currently supporting only cmake. Configuration was tested using GIL_USE_CONAN on, on Windows. One benchmark set was run successfully
The commit adds celero into conanfile and cmakelists, also provides one incorrectly implemented benchmark, which shows results that are completely inconsistent with google-benchmark
This commit adds a guide on how to write, build and execute benchmarks with supported frameworks (google-benchmark, celero). It also changes celero benchmark to crash instead of outputting wrong results.
This way we have a bit more control on release of dependencies what solves some annoying issues (e.g. Conan is missing latest Celero fixed for VS2019).
Benchmark of various pixel iteration and access methods for a contiguous images composed of homogeneous pixels with single 8-bit integer channel (grayscale).
sdebionne
force-pushed
the
add-benchmark-geom-xform
branch
from
9398fd9
to
0a520f4
Compare
sdebionne
force-pushed
the
add-benchmark-geom-xform
branch
from
0a520f4
to
30d6c6c
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description
Add benchmarks for geometry transformations with additional benchmarks for reference libraries such as OpenCV, IPP or Blaze for the transpose transformation.
References
Preliminary results were discussed on Slack
Tasklist