Slice Benchmark Report
Generated on 2025-05-05 04:54:08
System Information
| Property | Value |
|---|---|
| OS | Darwin |
| OS Version | 23.5.0 |
| Architecture | arm64 |
| Python Version | 3.12.2 |
| Date | 2025-05-05 04:54:08 |
| CPU | Apple M1 |
Executive Summary
This benchmark report compares different implementations of the 'slice' utility, with a focus on comparing the C implementations (slice1.c, slice2.c, and slice3.c).
C Implementation Comparison
| Scenario | Fastest Implementation | Avg. Time (s) | Throughput (MB/s) |
|---|---|---|---|
| full_file | C-slice3 | 0.006108 | 3884.64 |
| large_chunk | C-slice3 | 0.004951 | 4039.91 |
| small_start | C-slice3 | 0.003196 | 313.11 |
Benchmark Visualizations
C Implementation Full File
C Implementation Large Chunk
C Implementation Small Start
Small Slice Performance
Throughput Comparison
Time By File Size
C Implementation Details
The benchmark compares three C implementations:
- slice1.c: Basic implementation using low-level system calls and chunked reading.
- slice2.c: More robust implementation with improved error handling and safety features, but allocates the entire buffer at once.
- slice3.c: Enhanced implementation that combines the performance advantages of slice1.c with the safety features of slice2.c, using dynamic chunk sizing based on file size.
Scenario: full_file
| File Size (MB) | C-slice1 (s) | C-slice2 (s) | C-slice3 (s) |
|---|---|---|---|
| 1 | 0.007074 | 0.004507 | 0.004864 |
| 5 | 0.009209 | 0.003978 | 0.003536 |
| 10 | 0.017107 | 0.004834 | 0.004066 |
| 20 | 0.041724 | 0.006553 | 0.005189 |
| 50 | 0.065837 | 0.010855 | 0.007451 |
| 100 | 0.148108 | 0.018468 | 0.011542 |
Scenario: small_start
| File Size (MB) | C-slice1 (s) | C-slice2 (s) | C-slice3 (s) |
|---|---|---|---|
| 5 | 0.006980 | 0.003222 | 0.003217 |
| 10 | 0.004742 | 0.003307 | 0.003050 |
| 20 | 0.007412 | 0.003502 | 0.003278 |
| 50 | 0.006365 | 0.003358 | 0.003283 |
| 100 | 0.008634 | 0.003451 | 0.003153 |
Scenario: large_chunk
| File Size (MB) | C-slice1 (s) | C-slice2 (s) | C-slice3 (s) |
|---|---|---|---|
| 50 | 0.039727 | 0.006393 | 0.005018 |
| 100 | 0.041164 | 0.006194 | 0.004885 |
Raw Benchmark Data
| impl | scenario | file_size_MB | start_offset | slice_size | iterations | avg_time | min_time | max_time | stdev | throughput_MBps |
|---|---|---|---|---|---|---|---|---|---|---|
| C-slice1 | full_file | 1 | 0 | 1048564 | 5 | 0.007074 | 0.003532 | 0.020089 | 0.007278 | 141.361119 |
| C-slice2 | full_file | 1 | 0 | 1048564 | 5 | 0.004507 | 0.003759 | 0.005731 | 0.000952 | 221.874541 |
| C-slice3 | full_file | 1 | 0 | 1048564 | 5 | 0.004864 | 0.003314 | 0.009182 | 0.002492 | 205.589752 |
| Go | full_file | 1 | 0 | 1048564 | 5 | 0.005895 | 0.004075 | 0.011783 | 0.003305 | 169.633343 |
| Python | full_file | 1 | 0 | 1048564 | 5 | 0.024427 | 0.022679 | 0.029014 | 0.002755 | 40.937837 |
| Ruby | full_file | 1 | 0 | 1048564 | 5 | 0.049139 | 0.038352 | 0.087905 | 0.021682 | 20.350202 |
| Shell | full_file | 1 | 0 | 1048564 | 5 | 1.135014 | 1.098187 | 1.201039 | 0.042428 | 0.881036 |
| C-slice1 | small_start | 5 | 0 | 1048576 | 5 | 0.006980 | 0.002978 | 0.022489 | 0.008671 | 143.266476 |
| C-slice2 | small_start | 5 | 0 | 1048576 | 5 | 0.003222 | 0.002937 | 0.003996 | 0.000445 | 310.366232 |
| C-slice3 | small_start | 5 | 0 | 1048576 | 5 | 0.003217 | 0.002829 | 0.004143 | 0.000547 | 310.848617 |
| Go | small_start | 5 | 0 | 1048576 | 5 | 0.003780 | 0.003594 | 0.004086 | 0.000214 | 264.550265 |
| Python | small_start | 5 | 0 | 1048576 | 5 | 0.020993 | 0.020825 | 0.021089 | 0.000117 | 47.634926 |
| Ruby | small_start | 5 | 0 | 1048576 | 5 | 0.039839 | 0.036753 | 0.050483 | 0.005966 | 25.101032 |
| Shell | small_start | 5 | 0 | 1048576 | 5 | 1.099666 | 1.086396 | 1.109370 | 0.009182 | 0.909367 |
| C-slice1 | full_file | 5 | 0 | 5242864 | 5 | 0.009209 | 0.004151 | 0.029244 | 0.011199 | 542.945460 |
| C-slice2 | full_file | 5 | 0 | 5242864 | 5 | 0.003978 | 0.003814 | 0.004062 | 0.000099 | 1256.909186 |
| C-slice3 | full_file | 5 | 0 | 5242864 | 5 | 0.003536 | 0.003377 | 0.003687 | 0.000136 | 1414.022834 |
| Go | full_file | 5 | 0 | 5242864 | 5 | 0.004865 | 0.004754 | 0.005045 | 0.000112 | 1027.746093 |
| Python | full_file | 5 | 0 | 5242864 | 5 | 0.021809 | 0.021639 | 0.021938 | 0.000122 | 229.262449 |
| Ruby | full_file | 5 | 0 | 5242864 | 5 | 0.038637 | 0.037219 | 0.041942 | 0.001887 | 129.409238 |
Showing 20 rows out of 91 total. See the CSV file for complete data.
Conclusions
Based on the benchmark results, we can draw the following conclusions:
- slice3.c is on average 13.6% faster than slice2.c across all tests.
- The largest improvement was seen in the full_file scenario with 100MB files, where slice3.c was 37.5% faster.
- slice3.c outperformed slice1.c in 13 out of 13 test cases (100.0%).
- The dynamic chunk sizing approach in slice3.c successfully balances performance and memory usage.
- For small files, the performance difference between implementations is minimal.
- For larger files, the benefits of slice3.c become more pronounced.
Report generated automatically by slice/bench/generate_report.py