Benchmarking memory traffic from the while loop inside async CE

Program.fs

open System
open BenchmarkDotNet.Attributes
open BenchmarkDotNet.Running
open BenchmarkDotNet.Jobs
open FSharp.Control.Tasks.V2

[<SimpleJob(RuntimeMoniker.NetCoreApp31); MemoryDiagnoser>]
type Test() =

    [<Params(100, 1000, 10000)>]
    member val Length = 0 with get, set

    [<Benchmark(Baseline=true)>]
    member this.WhileAsync() =
        async {
            let mutable i = 0
            while i < this.Length do
                i <- i + 1
            return i
        } |> Async.RunSynchronously
        
    [<Benchmark>]
    member this.WhileNoAsync() =
        async {
            let mutable i = 0
            let while' () =
                while i < this.Length do
                    i <- i + 1
            do while'()
            return i
        } |> Async.RunSynchronously

    [<Benchmark>]
    member this.WhileTask () =
        (task {
            let mutable i = 0
            while i < this.Length do
                i <- i + 1
            return i
        }).Result


    [<Benchmark>]
    member this.RecNoAsync () =
        async {
            let mutable i = 0
            let rec while' () =
                if i = this.Length 
                then i
                else
                  i <- i + 1
                  while' ()
            return while' ()
        } |> Async.RunSynchronously

    [<Benchmark>]
    member this.RecAsync () =
        async {
            let mutable i = 0
            let rec while' () = async{
                if i = this.Length 
                then i
                else
                  i <- i + 1
                  return! while' ()
                 }
            return! while' ()
        } |> Async.RunSynchronously   
[<EntryPoint>]
let main _ =
    BenchmarkRunner.Run<Test>() |> ignore
    0

Result.md

BenchmarkDotNet=v0.12.0, OS=Windows 10.0.18362
AMD Ryzen 7 1700, 2 CPU, 4 logical and 4 physical cores
.NET Core SDK=3.1.101
  [Host]     : .NET Core 3.1.1 (CoreCLR 4.700.19.60701, CoreFX 4.700.19.60801), X64 RyuJIT DEBUG
  Job-KMAWNL : .NET Core 3.1.1 (CoreCLR 4.700.19.60701, CoreFX 4.700.19.60801), X64 RyuJIT

Runtime=.NET Core 3.1  

Method	Length	Mean	Error	StdDev	Ratio	RatioSD	Gen 0	Gen 1	Gen 2	Allocated
WhileAsync	100	7,323.1 ns	116.16 ns	108.65 ns	1.00	0.00	0.9232	0.0076	-	7776 B
WhileNoAsync	100	758.2 ns	12.61 ns	11.79 ns	0.10	0.00	0.1059	-	-	888 B
WhileTask	100	2,189.8 ns	48.89 ns	58.20 ns	0.30	0.01	0.4120	-	-	3456 B
RecNoAsync	100	741.1 ns	10.29 ns	9.63 ns	0.10	0.00	0.1059	-	-	888 B
RecAsync	100	4,974.8 ns	96.64 ns	85.67 ns	0.68	0.01	1.0605	-	-	8920 B
WhileAsync	1000	66,316.9 ns	1,309.51 ns	1,224.91 ns	1.00	0.00	7.8125	0.1221	-	65880 B
WhileNoAsync	1000	3,143.7 ns	60.17 ns	59.10 ns	0.05	0.00	0.1030	-	-	888 B
WhileTask	1000	20,019.2 ns	389.63 ns	478.50 ns	0.30	0.01	3.8452	-	-	32256 B
RecNoAsync	1000	3,168.0 ns	52.43 ns	49.04 ns	0.05	0.00	0.1030	-	-	888 B
RecAsync	1000	45,091.1 ns	824.55 ns	771.28 ns	0.68	0.02	9.6436	-	-	81088 B
WhileAsync	10000	644,875.3 ns	9,344.67 ns	8,283.81 ns	1.00	0.00	77.1484	0.9766	-	646920 B
WhileNoAsync	10000	27,566.9 ns	418.43 ns	391.40 ns	0.04	0.00	0.0916	-	-	888 B
WhileTask	10000	197,274.1 ns	2,292.18 ns	2,031.96 ns	0.31	0.01	38.0859	-	-	320257 B
RecNoAsync	10000	27,269.9 ns	371.37 ns	347.38 ns	0.04	0.00	0.0916	-	-	888 B
RecAsync	10000	445,178.9 ns	5,984.27 ns	5,304.90 ns	0.69	0.01	95.7031	-	-	802768 B