- For a list of bug fixes, see: https://github.com/radical-cybertools/radical.analytics/issues?q=is%3Aissue+is%3Aclosed+sort%3Aupdated-desc
- For a list of open issues and known problems, see: https://github.com/radical-cybertools/radical.analytics/issues?q=is%3Aissue+is%3Aopen
- maintenance release
- add entity name
- add name filter
- avoid deprecation warning
- changed path to RP event documentation
- speed up util plots
- set version requirement for RCT stack
- sync with RU
- fix for setuptools upgrade
- add profile parser
- avoid broken stats
- report unknown events
- disable stats on utilization plots
- add standalone profile parser
- maintenance
- pypi fix
- maintenance
- added check for
re.Pattern
- create experiments from session sources and instances
- fixed concurrency offset
- improve filter semantics
- fix RTD
- Add long description for pypi
- consistency across RCT modules
- avoid hardcoding session IDs
- enable compilation of notebooks
- fix extension renaming
- Iterate and fix all the tutorials
- simplify session ingest
- use same ncores/gpus of frontier
- fix MANIFEST.in
- re-enable notebooks
- fix doc deps
- fix rtd build
- remove invalid assert
- remove outdated test
- sync with ru
- better sorting for state transition plots
- documentation for plot_util_2, fix transition events
- better etype guessing
- raptor tasks use normal task events now (
rank_*
) - separate raptor contributions to utilization plots
- sync ci workflow and linter settings
- use https instead of git for doc requirements
- get cores_per_node from rm_info
- handle session w/o bootstrap events
- add badges and links
- adding requirements
- documentation fixes (layout and content, notbeook based examples)
- support concurrency analysis
- convert all the protected char in latex
- remove debug prints
- remove outdated code
- stringent check for non-negative times
- timestamp analysis
- rename
unit
totask
- utils/plot.py: help functions for plotting with Matplotlib
- style/radical_mpl.txt: default Matplotlib style for publication-grade plots.
- Separate type of resources (e.g., 'cpu', 'gpu) in resource utilization plot
- New utilization plot. aggregated utilization to improve performance and scale
- Documentation: Resource utilization, renaming compute unit -> task
- Tools: bin/rp_inspect/*.py plot 4 metrics for RP sessions
- Various bug fixes
- Move to Python 3
- Fix multiple bugs
- Align with release number of RCT stack
- Linting and RCT best code practices
- New documentation
- fix #102
- introduce analytics.Experiment class and move utilization method to it
- sessions are now cached
- Expand upon modeling TTX
- TTC diagrams
- add jsrun version of metrics
- add quick plotter script
- documentation
- entity.ranges() always returns a list
- experiment level utilization plots
- fix off-by-one error
- increase timer precision
- re-align utilization plot with pilot runtime
- recursive profiles for RA sessions
- support bz2
- support global session time shift (session.tzero)
- plot improvements
- fix tarball handling
- fix deployment dependency (thanks Matteo!)
- fail on and mark missing mandatory events
- add a native radical.analytics session (#63)
- add events to capture pre-exec durations (#67)
- add consistency check
- add allocation plot script
- add figure size option
- add plotting class
- add concurrency plot
- add statistics
- add utilization plot
- example fixes
- fix issue (#74)
- fix logger creation
- fix prints in wrangler
- apply coding guidelines
- make pypi happy
- fix license inconsistency
- add license file
- fixes to EnTK integration
- support sessions which only have profiles, but a limited or no event or state model
- Adding plotting class
- add concurrency plot
- PR #41
- If ranges is empty do nothing, else update dict. Value of dict is a list of lists in case there are more than one ranges
- Single session, multiple pilots utilization
- TTC diagrams
- Utilization method. For the moment single RP session, single pilot
- add several examples
- add time filter tests
- add valid range testing
- adding test case for the magic function -- currently test fails for dataset with execution barrier
- behave on empty event queries
- better testing for time filters
- entk integration
- entity.ranges() always returns a list
- event profiling cleanup
- expose session id
- expose stats, new event type
- ignore gitignore
- import matplotlib only if needed, to avoid hard dependency
- keep up with RU changes
- make sure we get session ID from an experiment dir
- mocking tests file structure, better .gitignore
- owner and consumer are mandatory inputs in the API call
- remove unwanted files
- support tarballs tarballs: those named "sid.tbz" etc. are now transparently unpacked and used.
- testing ranges
- tests require pytest
- use empty range set and "None" to indicate condition mismatch
- work on RA memory consumption
- clean out repository
- sync branches
- sync version across radical stack
- (HEAD -> master, origin/devel, origin/HEAD, devel) devel version bump