startautomating/benchpress

Easy Benchmarking with PowerShell

benchpress's Issues

Benchpress should build docs in CI/CD

Benchpress should use the OutputHelp action to publish docs in CI/CD

Include Clock Speed in Results Overview

Benchmarking information is improved by having some sense of the overall capabilities of the machine. It would be nice to include the clock speed in Benchpress results.

Benchpress should run GitPub on demand

Benchpress Action should be able to check in files

In order to facilitate easy consumption of benchmark metrics, the benchpress action should be able to check in files to the repository it is being used on.

Two files should be generated for each markdown file:

1. A .clixml file
2. A .md file

Measure-Benchmark output improvements

Measure-Benchmark's output could be improved.

• Adding the .RepeatCount to output would be nice

Benchmark using

Determine if it is faster to use using in a script than to reference the full name.

Checking If A file Exists

I'd like to know the fastest way to see if a file exists.

Benchpress Action should prefer local bits

When the Benchpress action encounters a local copy of Benchpress, it should prefer that.

Redo the built-in test at higher RepeatCount

I feel like the default value of 100 is low for some scenarios. Like for example Benchmark Should I Include The System Namespace.

With the default 100 I get that one is the better.

Technique               Time            RelativeSpeed Throughput
---------               ----            ------------- ----------
NoSytemNamespace        00:00:00.007584 1x            13184.44/s
SystemNamespaceIncluded 00:00:00.008561 1.13x         11680.06/s

Even your results shows this:
https://github.com/StartAutomating/Benchpress/blob/master/docs/Should_I_Include_The_System_Namespace.benchmark.benchmarkOutput.md

Now if I redo it manually with RepeatCount 1000 then they appear to be the same:

Technique               Time            RelativeSpeed Throughput
---------               ----            ------------- ----------
SystemNamespaceIncluded 00:00:00.007344 1x            136150.75/s
NoSytemNamespace        00:00:00.007373 1x            135620.8/s

So I think redoing the tests at least in some cases would be benefitial, where each Technique runs for a good few seconds at least.

Benchpress should build formatting in CI/CD

Benchpress should use the EZOut action to build formatting in CI/CD.

Benchpress should consolidate workflow

Incorrect $PSVersionTable.Platform check in How_Much_Faster_Is_Piping_To_A_ScriptBlock.benchmark

How_Much_Faster_Is_Piping_To_A_ScriptBlock.benchmark checks $PSVersionTable.Platform if it's windows, but on Powershell 7 at least that value is Win32NT, not Windows.

Benchpress/PowerShellPerformance/Ways_To_Hash_A_File.benchmark.ps1

Line 16 in b93fecd

if ($PSVersionTable.Platform -and $PSVersionTable.Platform -ne 'Windows') {

I would rather just check if certutil is there:

Get-Command -Name 'certutil.exe'

Benchmark output should not include empty groups.

Measure-Benchmark does not work in PowerShell 7

When running Measure-Benchmark in PowerShell 7, the following appears on screen:

cmdlet Get-CimInstance at command pipeline position 1
Supply values for the following parameters:
ClassName:

The problem is on line

Benchpress/Measure-Benchmark.ps1

Line 105 in 8e3b289

& $getCimInstance |

where is missing -ClassName parameter.

Benchmark Command Lookup

Benchpress should contain a benchmark to compare various command lookup techniques (e.g. Get-Command vs $executionContext)

Benchpress should test its action on a branch

Benchpress Should Have Markdown Formatters

In order to more effectively render output, benchpress should have Markdown formatters.

Benchpress should be able to Show-Benchmark

Show-Benchmark should default to a markdown file, however, if someone were to write an additional view for Benchmark.OutputFile, they should be able to use it with -View.

Support Sponsorship

Benchmark Output should include RepeatCount

Benchmark output accumulation

Sometimes you'll want to accumulate results from a pipeline. This can be done with arrays, arraylists, queues, and Tee-Object. It would be nice to settle which is the fastest.

What Is the Quickest Way to Compare Types

Benchpress formatting does not apply when deserialized

When a Benchmark.clixml file is imported, it is not formatted, because the typename changes to include Deserialized.

Benchmark Hashtables vs Ordered

What's the speed difference between using a [Hashtable] and an [Ordered] dictionary?

What is the fastest "random"?

It would be good to have a benchmark that can help determine the fastest way to make random numbers and strings.

Show-Benchmark should include a YAML Header

Show-Benchmark should include Title in metadata

Show-Benchmark mistakes output for benchmark files

This is due to the fact that benchmark files still contain .benchmark

Benchpress action should not rebase

Benchpress should use GitPub for microblogging

Readme.md - Reference PS Gallery?

More cosmetic than anything, but perhaps the readme.md could be updated to include something like:

Getting Started

Installation

Install-Module -Name Benchpress

Perhaps? (my immediate reaction on discovering this amazing tool was 'Oh, I wonder if it's on the gallery? or do I have to clone this from here at the moment?')

Happy for it to be dismissed though as it's a very minor thing :)

Path not resolved correctly when calling Checkpoint-Benchmark -ModuleName Benchpress

This looks like very interesting stuff!

My goal is just to run all of the benchmarks in the PowershellPerformance folder. I was following the readme, imported the psd1, and then called: Checkpoint-Benchmark -ModuleName Benchpress

What I get is some results, but a ton of
Resolve-Path : Cannot find path 'C:\PerformanceExperiments\WhatIsTheFastestWayToHashAFile.ps1' because it does not
exist.
At
C:\WINDOWS\system32\WindowsPowerShell\v1.0\Modules\Microsoft.PowerShell.Utility\Microsoft.PowerShell.Utility.psm1:110
char:36
when my repo is locally cloned to C:\ldx\Benchpress

I'm also noting that some of the tests don't like to run in a development environment (due to importing the module via name instead of relative path to the test file) so without the module installed via gallery I can't run some of the tests without error. (although I do still get results for those).

What Is The Fastest Way to Read a File?

Get-Content,

Get-Content -Raw?

IO.File.ReadAllText?

This question should be answered in a benchmark.