620bae3 Require a steady clock for bench with at least micro precision (Matt Corallo)
Pull request description:
Using a non-steady high_precision_clock by default is definitely not what we want, and in practice steady_clock has more than enough precision. Should double-check that travis passes on this one to make sure we actually have at least microsecond precision on all platforms.
Tree-SHA512: 54a4af3b6addca9897e8ab04694f9461343691b475ca3ed2368595c37520612e284969be94a8ee3d7c66d16532f7bb16b6ad80284cbc153653e8ef2d56696e9d
24a0bdd bench: prefer a steady clock if the resolution is no worse (Cory Fields)
c515d26 bench: switch to std::chrono for time measurements (Cory Fields)
Pull request description:
gettimeofday has portability issues, see for example #11558.
Regardless of large-scale clock refactors in the future, I think it's fine for bench to just use std::chrono itself.
Note that this may slightly improve bench accuracy and changes the display from tiny floats to nanosecond counts instead.
Tree-SHA512: 122355456d01ec6cfcf6867991715cf3a95eabbf5a4f2adc26a059b50382ffb318b7639cdd575197fc4ee5be8b967c0404f1f920d6f5bd4ddd0bd63b5e5c5632
1789e4675 Force explicit double -> int conversion for CFeeRate constructor (Matt Corallo)
53a6590f4 Make float <-> int casts explicit outside of test, qt, CFeeRate (Matt Corallo)
0b1b9148c Remove countMaskInv caching in bench framework (Matt Corallo)
Pull request description:
This fixes an issue where estimatesmartfee which matches at the min relay fee will return 999 sat/byte instead of 1000 sat/byte due to a float rounding issue. I went ahead and made all float <-> int conversion outside of test/qt explicit (test only had one or two more, Qt had quite a few, including many in the Qt headers themselves) and added overloads to CFeeRate to force callers to do an explicit round themselves. Easy to test with -Wfloat-conversion.
Tree-SHA512: 66087b08e5dfca67506da54ae057c2f9d86184415e8fa4fa0199e38839e06a3ce96c836fcb7593b7d960065f5240c594ff3a0cfa14333ac528421f5aeac835c9
fd05132e5 Restore default format state of cout after printing with std::fixed/setprecision (practicalswift)
Pull request description:
Restore default format state of `std::cout` after printing with `std::fixed`/`std::setprecision`.
Tree-SHA512: 445b5b42aff58e2350939e8febc9b4a6fff478616abfe831aec42bee906cefac7a153c93d506407fb213d04dae9c7afbb5bfd344be63ca0f40ae39b331a4144f
db07f91 Assert that what might look like a possible division by zero is actually unreachable (practicalswift)
Tree-SHA512: f1652eb37196a5b72f356503a1fbb44fb98aa8a94954ad1765f86d81ebf41a2337d4eb58c4f19937fda3752f5d2d642756e44afdbd438015b87ac20801246bff
Avoid calling gettimeofday every time through the benchmarking loop, by keeping
track of how long each loop takes and doubling the number of iterations done
between time checks when they take less than 1/16'th of the total elapsed time.
Benchmarking framework, loosely based on google's micro-benchmarking
library (https://github.com/google/benchmark)
Wny not use the Google Benchmark framework? Because adding Even More Dependencies
isn't worth it. If we get a dozen or three benchmarks and need nanosecond-accurate
timings of threaded code then switching to the full-blown Google Benchmark library
should be considered.
The benchmark framework is hard-coded to run each benchmark for one wall-clock second,
and then spits out .csv-format timing information to stdout. It is left as an
exercise for later (or maybe never) to add command-line arguments to specify which
benchmark(s) to run, how long to run them for, how to format results, etc etc etc.
Again, see the Google Benchmark framework for where that might end up.
See src/bench/MilliSleep.cpp for a sanity-test benchmark that just benchmarks
'sleep 100 milliseconds.'
To compile and run benchmarks:
cd src; make bench
Sample output:
Benchmark,count,min,max,average
Sleep100ms,10,0.101854,0.105059,0.103881
