bench

4 nov. 2021

Library "bench"
A simple banchmark library to analyse script performance and bottlenecks.

Very useful if you are developing an overly complex application in Pine Script, or trying to optimise a library / function / algorithm...

• Supports artificial looping benchmarks (of fast functions)
  
• Supports integrated linear benchmarks (of expensive scripts)
  

One important thing to note is that the Pine Script compiler will completely ignore any calculations that do not eventually produce chart output. Therefore, if you are performing an artificial benchmark you will need to use the bench.reference(value) function to ensure the calculations are executed.

Please check the examples towards the bottom of the script.

Quick Reference
(Be warned this uses non-standard space characters to get the line indentation to work in the description!)

```
// Looping benchmark style
benchmark = bench.new(samples = 500, loops = 5000)
data = array.new_int()

if bench.start(benchmark)
  while bench.loop(benchmark)
    array.unshift(data, timenow)

  bench.mark(benchmark)
  while bench.loop(benchmark)
    array.unshift(data, timenow)

  bench.mark(benchmark)
  while bench.loop(benchmark)
    array.unshift(data, timenow)

  bench.stop(benchmark)
  bench.reference(array.get(data, 0))
bench.report(benchmark, '1x array.unshift()')

// Linear benchmark style
benchmark = bench.new()
data = array.new_int()

bench.start(benchmark)
for i = 0 to 1000
  array.unshift(data, timenow)

bench.mark(benchmark)
for i = 0 to 1000
  array.unshift(data, timenow)

bench.stop(benchmark)
bench.reference(array.get(data, 0))
bench.report(benchmark,'1000x array.unshift()')
```

Detailed Interface

new(samples, loops) Initialises a new benchmark array
  Parameters:
    samples: int, the number of bars in which to collect samples
    loops: int, the number of loops to execute within each sample
  Returns: int[], the benchmark array

active(benchmark) Determing if the benchmarks state is active
  Parameters:
    benchmark: int[], the benchmark array
  Returns: bool, true only if the state is active

start(benchmark) Start recording a benchmark from this point
  Parameters:
    benchmark: int[], the benchmark array
  Returns: bool, true only if the benchmark is unfinished

loop(benchmark) Returns true until call count exceeds bench.new(loop) variable
  Parameters:
    benchmark: int[], the benchmark array
  Returns: bool, true while looping

reference(number, string) Add a compiler reference to the chart so the calculations don't get optimised away
  Parameters:
    number: float, a numeric value to reference
    string: string, a string value to reference

mark(benchmark, number, string) Marks the end of one recorded interval and the start of the next
  Parameters:
    benchmark: int[], the benchmark array
    number: float, a numeric value to reference
    string: string, a string value to reference

stop(benchmark, number, string) Stop the benchmark, ending the final interval
  Parameters:
    benchmark: int[], the benchmark array
    number: float, a numeric value to reference
    string: string, a string value to reference

report(Prints, benchmark, title, text_size, position)
  Parameters:
    Prints: the benchmarks results to the screen
    benchmark: int[], the benchmark array
    title: string, add a custom title to the report
    text_size: string, the text size of the log console (global size vars)
    position: string, the position of the log console (global position vars)

unittest_bench(case) Cache module unit tests, for inclusion in parent script test suite. Usage: bench.unittest_bench(__ASSERTS)
  Parameters:
    case: string[], the current test case and array of previous unit tests (__ASSERTS)

unittest(verbose) Run the bench module unit tests as a stand alone. Usage: bench.unittest()
  Parameters:
    verbose: bool, optionally disable the full report to only display failures

7 févr. 2022

Notes de version

v2 - Fixed "Cannot read property getText of undefined" error with default variables for reference function