Update links to concept-collection location
5 changed files+8−8
README.mdmodified+3−3View file
@@ -2,9 +2,9 @@
22
33 A benchmarking framework for evaluating compression algorithms on scientific data arrays.
44
5-Latest benchmark results: https://magland.github.io/benchcompress/
5+Latest benchmark results: https://concept-collection.github.io/benchcompress/
66
7-Paper (WIP): https://magland.github.io/benchcompress/paper
7+Paper (WIP): https://concept-collection.github.io/benchcompress/paper
88
99 ## Overview
1010
@@ -29,7 +29,7 @@ Benchcompress is a comprehensive benchmarking framework for evaluating compressi
2929 - Caching is based on algorithm and dataset versions
3030
3131 4. **Web Interface**
32- - Interactive visualization at https://magland.github.io/benchcompress/
32+ - Interactive visualization at https://concept-collection.github.io/benchcompress/
3333 - Filter and sort results by dataset or algorithm
3434 - Visual charts for comparing performance metrics
3535 - Export results to CSV for further analysis
benchcompress/src/benchcompress/run_benchmarks/collect_info.pymodified+2−2View file
@@ -2,8 +2,8 @@ from typing import List, Dict, Any
22 from ._memobin import construct_dataset_url
33 from ..types import Algorithm
44
5-GITHUB_ALGORITHMS_PREFIX = "https://github.com/magland/benchcompress/blob/main/benchcompress/src/benchcompress/algorithms/"
6-GITHUB_DATASETS_PREFIX = "https://github.com/magland/benchcompress/blob/main/benchcompress/src/benchcompress/datasets/"
5+GITHUB_ALGORITHMS_PREFIX = "https://github.com/concept-collection/benchcompress/blob/main/benchcompress/src/benchcompress/algorithms/"
6+GITHUB_DATASETS_PREFIX = "https://github.com/concept-collection/benchcompress/blob/main/benchcompress/src/benchcompress/datasets/"
77
88
99 def collect_algorithm_info(algorithms: List[Dict[str, Algorithm]]) -> List[Dict[str, Any]]:
paper/paper.mdmodified+1−1View file
@@ -121,7 +121,7 @@ Results are presented through an interactive web interface where users can explo
121121
122122 We first report on compression ratio performance, and then we will move on to encoding and decoding speeds.
123123
124-For the synthetic datasets that involve i.i.d. samples, ANS demonstrates superior compression performance compared to other algorithms, achieving compression ratios close to the theoretical entropy bounds. For example, in the Bernoulli sequence with $p=0.1$ ([https://magland.github.io/benchcompress/dataset/bernoulli-0.1](https://magland.github.io/benchcompress/dataset/bernoulli-0.1)), ANS achieves a compression ratio of ~16.9 whereas the next best algorithms (bzip2, lzma, btrotli-11, zstd-19, and zstd-22) achieve ratios around 13.9. Zlib struggles for this dataset, with a compression ratio of only ~12.2 at the highest compression level (9), and much lower values for lower levels. Brotli and Zstandard at the lower compression levels also perform poorly (for example ~9.6 for zstd-7). The results show a similar trend for the other Bernoulli datasets with different probabilities, but the spread is less pronounced as the entropy increases. The story is similar for the quantized Gaussian datasets, with ANS outperforming the other algorithms, especially for lower standard deviations (e.g., $\sigma=3$, [https://magland.github.io/benchcompress/dataset/gaussian-q3](https://magland.github.io/benchcompress/dataset/gaussian-q3)).
124+For the synthetic datasets that involve i.i.d. samples, ANS demonstrates superior compression performance compared to other algorithms, achieving compression ratios close to the theoretical entropy bounds. For example, in the Bernoulli sequence with $p=0.1$ ([https://concept-collection.github.io/benchcompress/dataset/bernoulli-0.1](https://concept-collection.github.io/benchcompress/dataset/bernoulli-0.1)), ANS achieves a compression ratio of ~16.9 whereas the next best algorithms (bzip2, lzma, btrotli-11, zstd-19, and zstd-22) achieve ratios around 13.9. Zlib struggles for this dataset, with a compression ratio of only ~12.2 at the highest compression level (9), and much lower values for lower levels. Brotli and Zstandard at the lower compression levels also perform poorly (for example ~9.6 for zstd-7). The results show a similar trend for the other Bernoulli datasets with different probabilities, but the spread is less pronounced as the entropy increases. The story is similar for the quantized Gaussian datasets, with ANS outperforming the other algorithms, especially for lower standard deviations (e.g., $\sigma=3$, [https://concept-collection.github.io/benchcompress/dataset/gaussian-q3](https://concept-collection.github.io/benchcompress/dataset/gaussian-q3)).
125125
126126
127127
web-ui/src/content/home-content.ymlmodified+1−1View file
@@ -44,7 +44,7 @@ sections:
4444 description:
4545 Explore the GitHub repository to view the source code, contribute to the
4646 project, or run your own benchmarks locally.
47- link: https://github.com/magland/benchcompress
47+ link: https://github.com/concept-collection/benchcompress
4848 linkText: View on GitHub
4949 external: true
5050
web-ui/src/pages/Home.tsxmodified+1−1View file
@@ -178,7 +178,7 @@ export default function Home() {
178178 )}
179179 Released under{" "}
180180 <a
181- href="https://github.com/magland/benchcompress/blob/main/LICENSE"
181+ href="https://github.com/concept-collection/benchcompress/blob/main/LICENSE"
182182 target="_blank"
183183 rel="noopener noreferrer"
184184 style={{ color: "#666", textDecoration: "underline" }}