2269
100%

Ran 11 May 2018 09:40AM UTC

Jobs 1

Files 66

Run time 4s

Badge

Embed ▾

pending completion

Build # 2269

Build Type

push

travis-ci

Committed by

ben-manes

Commit Message

Added a compute time benchmark based on the traces

The goal of this benchmark is to simulate the performance benefit of
switching algorithms on a fixed workload. If the hit rate increases,
one would expect to see the execution time decrease as the cost of a
miss goes up. However, if the eviction policy's overhead outweighs
the workload's costs then the optimization is not worthwhile.

Therefore the goal is to show if the improvements by a more advanced
algorithm outweigh its costs. For example LRU is extremely cheap to
execute and, with a small penalty, TinyLFU can be added to increase
its hit rate. In a workload like ARC's DS1 trace (database), this is
clear cut. But if we introduced a new policy that was very expensive
to operate, it wouldn't be feasible in practice.

This isn't expected to be used as a microbenchmark to performance tune
with. Rather its an indicator of whether a technique, such as mini-sim
sampling, is feasible in practice.

Coverage Stats

5720 of 6091 relevant lines covered (93.91%)

0.94 hits per line

Jobs

ID	Job ID	Ran	Files	Coverage
2	2269.2 (GROUP=tests)	11 May 2018 09:40AM UTC	0	93.91	Travis Job 2269.2

Source Files on build 2269

Detailed source file information is not available for this build.

ben-manes / caffeine / 2269
100%

README BADGES
x

Markdown

Textile

RDoc

HTML

Rst

Jobs

Source Files on build 2269

ben-manes / caffeine / 2269 100%

README BADGES x

Markdown

Textile

RDoc

HTML

Rst

Jobs

Source Files on build 2269

ben-manes / caffeine / 2269
100%

README BADGES
x