diff options
| author | skal <pascal.massimino@gmail.com> | 2026-02-07 08:54:04 +0100 |
|---|---|---|
| committer | skal <pascal.massimino@gmail.com> | 2026-02-07 08:54:04 +0100 |
| commit | da16d32f1e1adab7e461cd228e5fcecc60e1a41c (patch) | |
| tree | 285da93841c2e00f49c9f98e45350d2dcad9b6ce /scripts/build_win.sh | |
| parent | f28aef533e039def7c356ecac2dd0f45c6114209 (diff) | |
docs: Add Task #67 for DCT/FFT performance benchmarking
Adds low-priority task to measure and compare DCT/IDCT performance:
**Goal:** Quantify performance differences between implementations
- Reference O(N²) naive DCT/IDCT
- Current FFT-based O(N log N) implementation
- Future SIMD-optimized versions (when written)
**Location:** test_dct.cc or test_fft.cc
**Measurements:**
- Average time per transform (microseconds)
- Throughput (transforms per second)
- Speedup factor vs reference
- Multiple test sizes (128, 256, 512, 1024) for scaling analysis
**Implementation:**
- std::chrono::high_resolution_clock for timing
- 1000+ iterations to reduce noise
- Min/avg/max statistics
- Guarded with !STRIP_ALL for zero production impact
**Benefits:**
- Validate FFT speedup claims (O(N log N) vs O(N²))
- Quantify SIMD optimization gains when implemented
- Detect performance regressions in CI
**Priority:** Very Low (informational, not blocking any features)
This will be useful when optimizing audio performance in Phase 2.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
Diffstat (limited to 'scripts/build_win.sh')
0 files changed, 0 insertions, 0 deletions
