mirror of
https://github.com/tdurieux/anonymous_github.git
synced 2026-11-14 00:07:00 +01:00
1 line
1.8 KiB
HTML
1 line
1.8 KiB
HTML
<!doctype html><html><head><meta charset="utf-8"><meta name="viewport" content="width=device-width,initial-scale=1"><style>body{margin:0;font:14px/1.5 var(--font-sans,system-ui);color:var(--foreground)}h2{font-size:21px;margin:0 0 6px}p{margin:8px 0;color:var(--muted-foreground)}table{border-collapse:collapse;width:100%;margin:16px 0}th,td{text-align:right;padding:10px 6px;border-bottom:1px solid var(--border)}th:first-child,td:first-child{text-align:left;padding-left:0}th{color:var(--muted-foreground);font-weight:500}strong{color:var(--foreground)}.bar{height:12px;border-radius:3px;background:var(--chart-1,#748cff);margin:5px 0 14px}.after{background:var(--success,#35a879);width:10.74%}.label{display:flex;justify-content:space-between}.note{font-size:12px}</style></head><body><h2>ZIP peak memory: 89% lower</h2><p>Same repository snapshot · 2,017 files · Go before and after</p><div class="label"><span>Before</span><strong>664 MiB</strong></div><div class="bar"></div><div class="label"><span>After</span><strong>71 MiB</strong></div><div class="bar after"></div><table><thead><tr><th>Repeated download</th><th>Before</th><th>After</th></tr></thead><tbody><tr><td>Total time</td><td>39.3 s</td><td>36.0 s</td></tr><tr><td>First body bytes</td><td>149 ms</td><td>39 ms</td></tr><tr><td>Contents verified</td><td colspan="2">2,017 / 2,017 match</td></tr></tbody></table><p>MongoDB’s zstd buffers dominated the heap. Snappy reduces that cost; ZIP metadata now arrives in batches and entry copies reuse one buffer.</p><p class="note">One initial and one repeated request per build, sequential loopback runs. Memory is peak process-tree RSS during the repeated request; excludes database, Redis and client. Timing is indicative. ZIP directory metadata still grows with entry count.</p></body></html> |