charmplusplus / projections

Performance Analysis Tool for Charm++
Apache License 2.0
5 stars 3 forks source link

Timeline space/performance optimizations #120

Closed rbuch closed 3 years ago

rbuch commented 3 years ago

Various optimizations mostly aimed toward saving memory and improving performance when using Timeline. For the test case of Frontera NAMD logs I've been using, VisualVM reports that this branch results in a ~40% smaller total heap size (from 1,124 MB to 666 MB for the fairly small test case, I expect a better percentage savings for bigger logs since there is a significant fixed size component).

There are several other similar optimizations that can be made to other parts of Projections, this PR focuses on just the Timeline hot path lest it become too big.

lgtm-com[bot] commented 3 years ago

This pull request introduces 1 alert and fixes 1 when merging 861f2065970c9ac9d2d5d242f2f027338b07e7ff into 99a7fc2635997378d02a6c83e64e4a1baa473dd5 - view on LGTM.com

new alerts:

fixed alerts:

lgtm-com[bot] commented 3 years ago

This pull request fixes 1 alert when merging 67ab578122c9a6349b0ec88944d8eddbd5cf3e7c into 99a7fc2635997378d02a6c83e64e4a1baa473dd5 - view on LGTM.com

fixed alerts:

lgtm-com[bot] commented 3 years ago

This pull request fixes 1 alert when merging 9295669c576f74f0f54333c2521ea932353ce3aa into 99a7fc2635997378d02a6c83e64e4a1baa473dd5 - view on LGTM.com

fixed alerts: