Conversation
LevDenisov
force-pushed
the
benchmark-artifact
branch
from
95bee2d to
2bf047c
Compare
|
|
||
| deplex::utils::savePointCloudCSV(test_duration.cast<float>().transpose(), (data_dir / "process_cloud.csv").string()); | ||
|
|
||
| long long elapsed_time_mean = std::accumulate(test_duration.begin(), test_duration.end(), 0) / NUMBER_OF_RUNS; |
There was a problem hiding this comment.
Why do you store mean in integer (long long) type? Shouldn't it be float/double instead?
There was a problem hiding this comment.
No, because this method returns data with the type long long
There was a problem hiding this comment.
On the right hand you have not only method, but an expression with division operator. In C++ terms since both std::accumulate (in your case) and number of runs return integer type you will get an integer. But we know, that mean result won't be integer for sure. Cast either result of accumulate or number of runs to double so we can get floating point division.
| end_time = std::chrono::high_resolution_clock::now(); | ||
|
|
||
| time = std::chrono::duration_cast<std::chrono::microseconds>(end_time - start_time).count(); | ||
| test_duration[i] = (int)time; |
There was a problem hiding this comment.
Don't use C-style type casting. In C++ it's better to use static_cast(..)
| std::filesystem::path config_path = data_dir / "config/TUM_fr3_long_val.ini"; | ||
|
|
||
| auto start_time = std::chrono::high_resolution_clock::now(); | ||
| auto end_time = std::chrono::high_resolution_clock::now(); |
There was a problem hiding this comment.
Strange formatting... Do you use clang-format?
| double standard_error = standard_deviation / sqrt(NUMBER_OF_RUNS); | ||
|
|
||
| // 95% confidence interval | ||
| double lower_bound = (double)elapsed_time_mean - 1.96 * (standard_error); |
There was a problem hiding this comment.
What is the meaning of 1.96? Instead of using hard-coded constants it's better to create a variable with clear name and than use the variable :)
There was a problem hiding this comment.
Why do you need two different formats for C++ and Python? I understand that C++ code works only with the version without commas, but you can also parse same file from Python
There was a problem hiding this comment.
Do you really need orb trajectory in your code?
| def graphic(): | ||
| data = np.genfromtxt(Path(data_dir) / Path('process_sequence_50_snapshot.csv'), delimiter=',') | ||
|
|
||
| # Создаем boxplot |
There was a problem hiding this comment.
Mix of russian and english
|
|
||
| plt.xticks([0], ['stable']) | ||
|
|
||
| plt.ylabel("Время (мс.)") |
There was a problem hiding this comment.
Same, everyting shoud be in english
| .idea | ||
|
|
||
| #macOS | ||
| .DS_Store |
There was a problem hiding this comment.
I still can see .DS_Store in config directory
2 participants
Implementation of the benchmark for C++ and Python versions of the code.
Measurements are carried out in the following groups:
Image reading, point cloud translation, image segmentation
For each stage of image segmentation