ben-manes / caffeine
100%
master: 100%

LAST BUILD BRANCH: v3.dev
DEFAULT BRANCH: master
Repo Added
Files 79
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst

LAST BUILD ON BRANCH v3.dev
branch: SELECT
CHANGE BRANCH
x

coverage: 99.988% (+0.04%) from 99.953%
#5686

push

github

ben-manes 
Adapt the window by hit density with starvation-guarded probes

Large caches size the admission window by the within-sample hit density
of the two regions, ln(windowDensity / mainDensity), taking proportional
steps that are immune to the cross-sample hit rate swings of a phasey
workload. The signal is resident-only, so a region earning ~nothing is
its blind state: it cannot see what a different split would earn, and
holding such a position can pin the window at an extreme (constructible
steady-state workloads pinned it ~28pp below LRU at any scale).

Starved samples at a blind corner therefore probe by hit rate. Walks
are bounded by a sample budget, by a reversal that crosses back through
the probe's start, and by an absolute hit rate veto that undoes damage;
the density signal adjudicates each endpoint under strict confirmation.
A crash re-arms the refractory without doubling, since an exogenous
workload shift is indistinguishable from probe damage and mispricing it
would starve the re-exploration the shift calls for. Escalating
commitment resolves the stray-exit dilemma: first-round probes exit
cheaply when the watched region earns without density confirmation
(stray and transferred hits scale with a region's size, and thin-signal
workloads correctly at the floor must not pay deep-walk displacement),
while each adjudicated failure lengthens the refractory ladder and its
deeper rungs commit the walk past the stray zone, converting a deep
silent reuse band from an absorbing pin into a bounded, temporary dip.
Deep rungs also stride at 2x and 4x (capped at the 30% max step): the
ladder previously bought a committed walk permission to pass the stray
zone but not speed, so deep reuse bands were reached only by outlasting
the budget, and stray walls calibrated wider than a flat stride's reach
were absorbing. Failed walks still undo in full, which is what
distinguishes this from the rejected travel-budget exits. The sample
period is capped by the sketch's e... (continued)

4305 of 4315 branches covered (99.77%)

196 of 196 new or added lines in 6 files covered. (100.0%)

8601 of 8602 relevant lines covered (99.99%)

1.0 hits per line

Relevant lines Covered
8602 RELEVANT LINES 8601 COVERED LINES
1.0 HITS PER LINE
Source Files on master

Recent builds

Builds Branch Commit Type Ran Committer Via Coverage
#5686 v3.dev Adapt the window by hit density with starvation-guarded probes Large caches size the admission window by the within-sample hit density of the two regions, ln(windowDensity / mainDensity), taking proportional steps that are immune to the cross-sam... push ben-manes github
99.99
#5685 v3.dev Adapt the window by hit density with starvation-guarded probes Large caches size the admission window by the within-sample hit density of the two regions, ln(windowDensity / mainDensity), taking proportional steps that are immune to the cross-sam... push ben-manes github
99.99
#5684 v3.dev Adapt the window by hit density with starvation-guarded probes Large caches size the admission window by the within-sample hit density of the two regions, ln(windowDensity / mainDensity), taking proportional steps that are immune to the cross-sam... push ben-manes github
99.99
#5683 v3.dev Adapt the window by hit density with starvation-guarded probes Large caches size the admission window by the within-sample hit density of the two regions, ln(windowDensity / mainDensity), taking proportional steps that are immune to the cross-sam... push ben-manes github
99.91
#5682 master Align the 2Q in-queue default to the paper's recommendation Johnson and Shasha recommend Kin ~= 25% (Kout ~= 50%); the default was 0.20 / 0.50. Match the paper. 2Q is insensitive to Kin in this range, so the hit rate is unchanged on the bundled t... push ben-manes github
99.95
#5681 v3.dev Align the 2Q in-queue default to the paper's recommendation Johnson and Shasha recommend Kin ~= 25% (Kout ~= 50%); the default was 0.20 / 0.50. Match the paper. 2Q is insensitive to Kin in this range, so the hit rate is unchanged on the bundled t... push ben-manes github
99.95
#5680 v3.dev Harden the simulator combined-report and chart steps CombinedCsvReport matched --metric against the CSV headers case- sensitively and swallowed a miss into empty cells, which later made the chart step crash on all-null data. Resolve the metric ig... push ben-manes github
99.95
#5679 v3.dev Key Tencent photo variants by format and size A photo is stored as a distinct blob per (id, format, size category), but the reader keyed by (id, return_size), using the byte count as a proxy for the variant. Two variants of a photo that return th... push ben-manes github
99.95
#5678 v3.dev Materialize the clairvoyant look-ahead to a shared file opt.Clairvoyant and admission.Clairvoyant each held the whole trace in memory to compute Belady's next-access times. Materialize them once, up front, to a fixed-width temp file that every cl... push ben-manes github
99.95
#5677 master Materialize the clairvoyant look-ahead to a shared file opt.Clairvoyant and admission.Clairvoyant each held the whole trace in memory to compute Belady's next-access times. Materialize them once, up front, to a fixed-width temp file that every cl... push ben-manes github
99.95
See All Builds (4709)