- Added instructions, cycles, and IPC metrics to all reports
- Created CSV data files for each language with detailed metrics
- Added timeline data (memory and CPU over time) for each run
- Updated all reports with new metrics
- Created analysis script to analyze collected data
- Generated reports for all decimal levels (1, 2, 5, 10, 100, 1000, 2000)
Key findings:
- D has highest IPC (4.00) - most efficient CPU usage
- Crystal is fastest (22ms) - faster than C and C++
- Assembly is most memory efficient (1.4MB)
- Rust and Fortran have IPC 3.11 - good optimization
- Fixed memory measurement issue where Rust, Nim, Odin showed 0 bytes
- Now using /usr/bin/time -l on macOS for reliable memory measurement
- Works for all programs, including very fast ones
- Updated README with correct performance results
- All 34 languages now show accurate memory values
- Memory values are now in bytes (not MB)
- Added comprehensive performance analysis
- Updated timeline files with correct memory data
- Added test output files for all decimal levels
Performance improvements:
- Rust: 0 bytes → 1,622,016 bytes (1.6 MB) ✓
- Nim: 0 bytes → 1,523,712 bytes (1.5 MB) ✓
- Odin: 0 bytes → 1,605,632 bytes (1.6 MB) ✓
- All other languages show correct memory values ✓
Test results verified from three perspectives:
- Data Analyst: All values are reasonable and understandable
- Senior Developer: Memory profiling works correctly for all languages
- Hardware Engineer: All values are physically possible and not fabricated
- Removed MB conversion in run_all.sh
- Now displays actual bytes values from memory profiling
- Updated output to show 'bytes avg / bytes peak'
- More accurate representation of memory usage
- Consistent with timeline data which is already in bytes
- Changed output to show bytes instead of MB
- Updated header to show 'Minne (bytes)'
- Better for comparing memory usage across all languages
- Memory data already collected in bytes from ps command
- Changed from 10ms to 1ms sampling interval
- Better captures data for fast programs
- Memory already sampled in bytes for consistency
- Will provide more detailed timeline data
- Recreated entire README in English
- Preserved all technical details and data
- Improved flow and readability
- Kept Swedish version as README_SV.md for reference
- All charts and analysis now in English
- Save timeline data for each run (time, memory, CPU)
- Create timelines directory structure
- Add Python visualization script for charts
- Fix integer comparison errors in profiling
- Collect samples throughout program lifetime
- Track memory and CPU usage over program lifetime
- Store elapsed time with each sample
- Generate timeline data for visualization
- Enable detailed performance analysis
- Added get_cpu_usage() function to sample CPU percentage
- Updated profile_resources() to profile both memory and CPU
- Modified run_program() to track CPU usage alongside memory
- Updated output format to show CPU statistics
- Summary now displays: Time, Memory (avg/peak), CPU (avg/peak)
- Both memory and CPU are sampled every 10ms during execution
- Added binary size section to README with actual file sizes
- Updated run_all.sh to profile memory usage during execution
- Memory profiling shows average and peak memory consumption
- All data from actual measurements on Apple A18 Pro
- Binary sizes range from 103B (wrapper scripts) to 13M (Haskell)
Ein Anderssono