...
- open a discussion with rholang team about shape of incoming data
- get lock free store implementation part of dev
- investigate the slowdown observed in last experiment
- Implement a version with KeyRange and compare.
- Investigate further with dupe contract (create new bench)
Analysis of RSpace behaviour with 'wide' contract.
Replace event log syncvar with a concurrent structure. Jira Legacy server System JIRA serverId 50130123-f232-3df4-bccb-c16e7d83cd3e key CORE-1386
AtomicAny was used to store the event log.
The comparison shows that the change is insignificant.
Proposed ways forward:
- the event log does not need to be gathered inside rspace lock, some much needed refactor could result in the matcher returning refs to events
Figure out SpatialMatcher impact on rspace.
Jira Legacy | ||||||
---|---|---|---|---|---|---|
|
Jira Legacy | ||||||
---|---|---|---|---|---|---|
|
- while running comparisons of wide the following observations were made
- usually a run of produce (while the tuplespace holds consumes) takes about 0.9ms
- a simple match (one consume in the tuplespace) takes about 0.35ms
- this merits a further dig into optimisation of the macher
Analysis of serialization impact.
Jira Legacy | ||||||
---|---|---|---|---|---|---|
|
...