keepalive_pipelining greenvalidity pass

Tier 5 · rps · 16-pipeline keepalive · OS linux

Validity gate

passpass — perf claims allowed when ratio is green

Source: latest.csv:perf_present

Wall-clock row exists only — no explicit correctness signal yet.

Ratio vs best competitor
0.836 (83.6% of lighttpd speed) (table SOTA: lighttpd — best competitor; charts use lighttpd = 1.0)
Best competitor value
6528.24 req/s

Numeric oracle

No analytical verify rows in ingest yet. Run tier-1 with --verify so latest.csv exports verify_ulps / verify_within_1ulp.

Problem size
16-pipeline keepalive (family keepalive_pipelining)
Category
http
Pillar
server
Package
lis
Li / catalog oracle (mean ± σ)
5457.48 req/s (n=3) / (nginx)
Ratio vs catalog oracle
0.0273×
Best competitor
lighttpd (6528.24 req/s)
Li relative speed vs SOTA
0.836 (1.0 = lighttpd speed)
Validity
validity pass
Threshold
1×
Source
lis/benchmarks/workloads/tier5_http/scenarios/keepalive_pipelining

PH ids: PH-H

Performance vs best competitor

Compare oracle: nginx

Relative speed vs best competitor (lighttpd) — SOTA = 1.0, higher is better. Absolute req/s values are in the table below.

Absolute measurements

Language comparison
LangMean ± σRunsUnitVariantOS
li5457.48req/sreleaseunknown
nginx149.21req/sreleaseunknown
apache1426.32req/sreleaseunknown
lighttpd6528.24req/sreleaseunknown
node2844.99req/sreleaseunknown
Metric: rps (value = mean of timed runs)

Host OS

Measurement hosts by language
ScopeOS
Row aggregatelinux
li (release)unknown
nginx (release)unknown
apache (release)unknown
lighttpd (release)unknown
node (release)unknown

← Overview