diff options
| author | Paul Buetow <paul@buetow.org> | 2025-06-16 23:23:26 +0300 |
|---|---|---|
| committer | Paul Buetow <paul@buetow.org> | 2025-06-16 23:23:26 +0300 |
| commit | 111fb5753d416214c680abb288d31c595dcdcea1 (patch) | |
| tree | 5bd23cb0cfa7add488c57936be9502de42073e5d /scripts | |
| parent | 7bc889acb28d8613944551d2129f0d9a7b65ecc3 (diff) | |
implement true Profile-Guided Optimization with Go compiler -pgo flag
- Refactor PGO script to use actual Go compiler PGO instead of just profiling
- Add proper baseline vs PGO-optimized binary comparison
- Break script into maintainable functions for better organization
- Update Makefile and documentation to reflect PGO process
- Generate comprehensive performance reports with before/after analysis
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
Diffstat (limited to 'scripts')
| -rw-r--r-- | scripts/pgo_report.txt | 68 |
1 files changed, 68 insertions, 0 deletions
diff --git a/scripts/pgo_report.txt b/scripts/pgo_report.txt new file mode 100644 index 0000000..d6e1a83 --- /dev/null +++ b/scripts/pgo_report.txt @@ -0,0 +1,68 @@ +=== PROFILE GUIDED OPTIMIZATION REPORT === +Generated: Mon 16 Jun 23:18:37 EEST 2025 + +BASELINE (without PGO): +Baseline performance (5 iterations): +real 0m3.040s +real 0m3.029s +real 0m3.032s +real 0m3.030s +real 0m3.031s + +PGO-OPTIMIZED: +PGO-optimized performance (5 iterations): +real 0m3.035s +real 0m3.033s +real 0m3.033s +real 0m3.034s +real 0m3.031s + +DETAILED ANALYSIS: + +Baseline CPU Profile: +File: dgrep +Build ID: c4f25989f74683061bfabfc72b383431d1aeeb23 +Type: cpu +Time: 2025-06-16 23:17:42 EEST +Duration: 3.20s, Total samples = 8.73s (272.51%) +Showing nodes accounting for 7.32s, 83.85% of 8.73s total +Dropped 174 nodes (cum <= 0.04s) + flat flat% sum% cum cum% + 2.23s 25.54% 25.54% 2.23s 25.54% internal/runtime/syscall.Syscall6 + 0.37s 4.24% 29.78% 1.01s 11.57% runtime.selectgo + +PGO-Optimized CPU Profile: +File: dgrep_pgo +Build ID: 106bf00e9fe2a0beaaf9b0e80a5e7e14aae84c40 +Type: cpu +Time: 2025-06-16 23:18:34 EEST +Duration: 3.11s, Total samples = 8.66s (278.78%) +Showing nodes accounting for 7.41s, 85.57% of 8.66s total +Dropped 152 nodes (cum <= 0.04s) + flat flat% sum% cum cum% + 2.17s 25.06% 25.06% 2.17s 25.06% internal/runtime/syscall.Syscall6 + 0.51s 5.89% 30.95% 1.31s 15.13% runtime.selectgo + +Baseline Memory Profile: +File: dgrep +Build ID: c4f25989f74683061bfabfc72b383431d1aeeb23 +Type: inuse_space +Time: 2025-06-16 23:17:45 EEST +Showing nodes accounting for 66.08MB, 100% of 66.08MB total + flat flat% sum% cum cum% + 33MB 49.94% 49.94% 60.84MB 92.06% time.NewTimer + 27.83MB 42.12% 92.06% 27.83MB 42.12% time.newTimer + 1.72MB 2.61% 94.67% 1.72MB 2.61% runtime/pprof.StartCPUProfile + 1.50MB 2.27% 96.94% 1.50MB 2.27% runtime.allocm + +PGO-Optimized Memory Profile: +File: dgrep_pgo +Build ID: 106bf00e9fe2a0beaaf9b0e80a5e7e14aae84c40 +Type: inuse_space +Time: 2025-06-16 23:18:37 EEST +Showing nodes accounting for 80.57MB, 100% of 80.57MB total + flat flat% sum% cum cum% + 42.35MB 52.57% 52.57% 42.35MB 52.57% time.newTimer + 32.50MB 40.34% 92.91% 74.86MB 92.91% time.NewTimer + 2MB 2.49% 95.39% 2MB 2.49% runtime.allocm + 1.16MB 1.44% 96.83% 1.16MB 1.44% runtime/pprof.StartCPUProfile |