foundationdb/flowbench
Aaron Molitor 7cadc5420e update target_link with new path from build image 2023-03-03 16:30:07 -06:00
..
include/flowbench Move sample benchmarks to Flowbench. 2022-09-12 01:31:09 -07:00
BenchBlobDelta.cpp Add micro-benchmark for blob delta serialization 2022-11-11 09:22:11 -08:00
BenchCallback.actor.cpp Update copyright header dates 2022-03-21 13:36:23 -07:00
BenchEncrypt.cpp Use faster KeepRunning loops in flowbench 2022-09-11 20:32:28 -07:00
BenchHash.cpp Use faster KeepRunning loops in flowbench 2022-09-11 20:32:28 -07:00
BenchIONet2.actor.cpp moved wellknownendpoints and fixed some includes 2022-06-23 17:03:53 -06:00
BenchIdempotencyIds.cpp Performance improvements and benchmark tweaks 2022-10-12 13:53:11 -07:00
BenchIterate.cpp Use faster KeepRunning loops in flowbench 2022-09-11 20:32:28 -07:00
BenchMem.cpp Use faster KeepRunning loops in flowbench 2022-09-11 20:32:28 -07:00
BenchMetadataCheck.cpp Convert literal string ref instances to use _sr suffix 2022-09-19 11:35:58 -07:00
BenchNet2.actor.cpp Bug fix in items processed count, changed delay/yield test to be a template. 2022-10-31 12:49:38 -07:00
BenchPopulate.cpp Use faster KeepRunning loops in flowbench 2022-09-11 20:32:28 -07:00
BenchPriorityMultiLock.actor.cpp Remove runners list from PriorityMultiLock and rely on reference counting in the release handler instead of canceling the release handlers. This improves the microbenchmark by 26%. 2022-11-11 00:34:03 -08:00
BenchRandom.cpp Use faster KeepRunning loops in flowbench 2022-09-11 20:32:28 -07:00
BenchRef.cpp Use faster KeepRunning loops in flowbench 2022-09-11 20:32:28 -07:00
BenchSamples.cpp Change Histogram::Unit::microseconds to milliseconds 2022-11-21 08:03:56 -08:00
BenchStream.actor.cpp Update copyright header dates 2022-03-21 13:36:23 -07:00
BenchTimer.cpp Revert "Use real clock source for trace events in real fdbserver, but now() in simulation. (#9270)" 2023-02-02 10:30:31 -08:00
BenchVersionVector.cpp Use faster KeepRunning loops in flowbench 2022-09-11 20:32:28 -07:00
BenchVersionVectorSerialization.cpp Use faster KeepRunning loops in flowbench 2022-09-11 20:32:28 -07:00
BenchZstd.cpp add benchmark for zstd 2022-10-11 08:23:55 -07:00
CMakeLists.txt update target_link with new path from build image 2023-03-03 16:30:07 -06:00
GlobalData.cpp Place generateRandomData() under {I|Deterministic}Random 2022-07-20 13:21:11 +02:00
README.md Exclude flowbench target from all 2020-10-10 15:03:53 -07:00
benchmark.cmake Add support for pre-built googlebenchmark 2023-02-27 15:38:58 -06:00
flowbench.actor.cpp Simplify function call when transaction is null 2022-06-22 14:50:17 -07:00

README.md

Summary

flowbench is an executable that can be used to microbenchmark parts of the FoundationDB code. The goal is to make it easy to test the performance of various sub-millisecond operations using flow and fdbrpc. Specifically, this tool can be used to:

  • Test the performance effects of changes to the actor compiler or to the flow and fdbrpc libraries
  • Test the performance of various uses of the flow and fdbrpc libraries
  • Find areas for improvement in the flow and fdbrpc libraries
  • Compare flow/fdbrpc primitives to alternatives provided by the standard library or other third-party libraries.

Usage

  • To build the flowbench executable, run ninja flowbench or make flowbench depending on which build system you're using.
  • Then you can run bin/flowbench --help to see possible uses of flowbench.
  • Running bin/flowbench directly will run all registered benchmarks, but you may want to limit your run to a subset of benchmarks. This can be done by running bin/flowbench --benchmark_filter=<regex>
  • All benchmark names can be listed with bin/flowbench --benchmark_list_tests
  • Example output:
$ bin/flowbench --benchmark_filter=bench_ref
2020-08-04 21:49:40
Running bin/flowbench
Run on (7 X 2904 MHz CPU s)
CPU Caches:
  L1 Data 32 KiB (x7)
  L1 Instruction 32 KiB (x7)
  L2 Unified 256 KiB (x7)
  L3 Unified 12288 KiB (x1)
Load Average: 0.15, 0.15, 0.72
---------------------------------------------------------------------------------------------------------------
Benchmark                                                     Time             CPU   Iterations UserCounters...
---------------------------------------------------------------------------------------------------------------
bench_ref_create_and_destroy<RefType::RawPointer>          4.90 ns         4.90 ns    116822124 items_per_second=203.88M/s
bench_ref_create_and_destroy<RefType::UniquePointer>       4.94 ns         4.94 ns    141101924 items_per_second=202.555M/s
bench_ref_create_and_destroy<RefType::SharedPointer>       42.5 ns         42.5 ns     13802909 items_per_second=23.531M/s
bench_ref_create_and_destroy<RefType::FlowReference>       5.05 ns         5.05 ns    100000000 items_per_second=197.955M/s
bench_ref_copy<RefType::RawPointer>                        1.15 ns         1.15 ns    612121585 items_per_second=871.218M/s
bench_ref_copy<RefType::SharedPointer>                     10.0 ns         10.0 ns     67553102 items_per_second=99.8113M/s
bench_ref_copy<RefType::FlowReference>                     2.33 ns         2.33 ns    292317474 items_per_second=428.507M/s

Existing Benchmarks

  • bench_populate measures the population of a vector of mutations
  • bench_ref compares the performance of the flow Reference type to other pointer types
  • bench_iterate measures iteration over a list of mutations
  • bench_stream measures the performance of writing to and reading from a PromiseStream
  • bench_random measures the performance of DeterministicRandom.
  • bench_timer measures the perforamnce of FoundationDB timers.

Future use cases

  • Benchmark the overhead of sending and receiving messages through FlowTransport
  • Benchmark the performance of serializing/deserializing various types