v0.4.0

It's time for another release!

What's New

New Cord Benchmark

We're excited to introduce a new benchmark for absl::cord, expanding our performance testing suite. This benchmark supports complex cord operation lifecycles that represent the real world usage.

Updated Proto Benchmark

We've also taken the opportunity to enhance our existing Proto benchmark. This update supports several new features, including:

• Repeated messages and their usages
• Cord datatype proto variables
• More representative field type distributions
• Readability improvement

More Realistic Libc Benchmark

• The memmove benchmarks have been updated to include more realistic scenarios that account for overlap probability in real workloads. Additionally, we've managed to reduce overhead, resulting in a more efficient performance evaluation.

Acknowledgments

A big thanks to everyone who contributed to this release! Your dedication and hard work are invaluable to the success of this project.

Happy Holidays!

As we approach the holiday season, we want to extend our warmest wishes to the entire community. May this festive season bring joy, peace, and relaxation to you and your loved ones.

Thank you for your continued support, and here's to a fantastic holiday season! 🎉🌟

v0.3.3

Output an error message if cache size cannot be detected.

PiperOrigin-RevId: 585684097
Change-Id: Ieae63812addbfd1ff6ed6c268900dd1c7867cf8c

v0.3.2

Update contributing document.

PiperOrigin-RevId: 578301578
Change-Id: I596fcb8f3f8a8383d65a372f620068b09856065e

v0.3.1

Proto: remove std::shuffle() in Run().

PiperOrigin-RevId: 567707154
Change-Id: Ib24495d2c416cbc82d1721950650c8c1816c366e

v0.3.0

• @andreas-abel Greatly reduced the cold swissmap execution time, and removed the --benchmark_filter constraint in the previous release.
• @liyuying0000 helps to refactor libc benchmark by relaxing L1-resident constraints.
• Add coverages for two additional benchmarks to include compression and hashing algorithms.
• Update README.md.
• Provide a default benchmark minimal set. For details, please refer to README.md.

v0.2.3

Use GetFleetbenchRuntimePath to fetch compression_parameter

PiperOrigin-RevId: 561163538
Change-Id: Ib7115ea329af951e05345e1be7fe04cf303d84c2

v0.2.2

Load Python rules from rules_python

PiperOrigin-RevId: 552531154
Change-Id: Ibd0b20db91897f3f463a9197030bf6ff991aa2c0

v0.2.1

Update fleetbench readme to account for removing the default args on …

v0.2

This release expands the coverage by adding benchmarks for:

• TCMalloc: memory allocation: C's malloc() and C++'s operator new.
• Libc mem* : memory functions memcpy(), memcmp(), memmove(), etc..

It also supports building the benchmark with Clang / LLVM toolchain, and FDO.

v0.1.1

• Fix memory issue in Proto BM @liyuying0000.
• Fix parameters to be consistent in hot & cold swissmap BM @aysylu.

Full Changelog: v0.1...v0.1.1