Fcalva/cyclone
1
0
Fork 0
mirror of https://github.com/justinethier/cyclone.git synced 2025-05-23 20:15:05 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

Handle remaining TODO's

Browse source
This commit is contained in:
Justin Ethier 2018-07-26 17:43:26 -04:00
parent 09d563bf53
commit 700e39e9cb
1 changed files with 3 additions and 5 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

8
docs/Garbage-Collector-Lazy-Sweeping.md
View file

@ -207,18 +207,16 @@ Average Speedup | N/A | 10.74%
Maximum Speedup | deriv | 36.90%
Minimum Speedup | wc | -2.07%
So we achieve on average a speedup of 10.74% with lazy sweeping. It is interesting that there is such a wide range of performance impacts across the whole benchmark suite. Most likely those benchmarks with the biggest speedups are those that are generating the most garbage. Alternatively, benchmarks such as `wc` are likely generating very little garbage. In particular, `wc` spends most of its time running in a tight loop that performs few allocations. Unfortunately there is some additional overhead that slightly increases the runtime for some programs. But the overall performance improvement more than compensates.
TODO: discuss why benchmarks turned out the way they did, does wc have no dead objects? what about deriv?
Overall we achieve an average speedup of 10.74% with lazy sweeping. It is interesting that there is such a wide range of performance impacts across the whole benchmark suite. Most likely those benchmarks with the biggest speedups are those that are generating the most garbage. Alternatively, benchmarks such as `wc` are likely generating very little garbage. Case in point, `wc` spends most of its time running in a tight loop that performs few allocations. It is unfortunate that lazy sweeping slightly increases the runtime for some programs but the overall performance improvement more than compensates.
# Conclusion
TODO: are results what we expect? observations, impressions, and next steps
By all accounts lazy sweeping is a great win for Cyclone and has exceeded performance expectations. Though it took some time to integrate lazy sweeping and then re-stabilize the GC code, this opens the possibility to experiment with other larger-scale GC optimizations.
# References
1. [The Garbage Collection Handbook: The Art of Automatic Memory Management](http://gchandbook.org/), by Antony Hosking, Eliot Moss, and Richard Jones
2. [Introducing Riptide: WebKit’s Retreating Wavefront Concurrent Garbage Collector](https://webkit.org/blog/7122/introducing-riptide-webkits-retreating-wavefront-concurrent-garbage-collector/), by Filip Pizlo
2. [Introducing Riptide: WebKit's Retreating Wavefront Concurrent Garbage Collector](https://webkit.org/blog/7122/introducing-riptide-webkits-retreating-wavefront-concurrent-garbage-collector/), by Filip Pizlo
3. [Scheme Benchmarks](https://ecraven.github.io/r7rs-benchmarks/), by [ecraven](https://github.com/ecraven)
4. [The Ramsey sweep](http://people.csail.mit.edu/gregs/ll1-discuss-archive-html/msg00761.html), by Olin Shivers
5. [The Cyclone Scheme Garbage Collector](Garbage-Collector.md), by Justin Ethier