Attempt to do a comparison of forced UseVirtualTag

[manofstick]

Just creating PR to get AppVeyor to do some work...

This is just a shotgun approach to UseVirtualTag, and still tweaks are required, but should be a reasonable baseline...

[forki]

Compiler: 5c931e3149ebdd5d1e5f9721a3c6dbfb63e9d1c6
   Project: Paket 11872.0ms
   Project: FSharpx.Collections 2524.0ms
   Project: FSharp.Data 4001.0ms
   Project: SQLProvider 7960.0ms
   Average Run: 26358.0ms
Compiler: 9d5e36a7aa66f029640aa05f96bd7b54907aef81
   Project: Paket 11486.0ms
   Project: FSharpx.Collections 2404.0ms
   Project: FSharp.Data 3874.0ms
   Project: SQLProvider 7598.0ms
   Average Run: 25363.0ms

[forki]

this looks awesome. Only question is how trustworthy my perf framework is...

[manofstick]

Well a bit more what I was expecting:

Compiler: 32c530ebe42f34a863a9e55b682f90685e658125 Project: Paket 11217.0ms Project: FSharpx.Collections 2424.0ms Project: FSharp.Data 3801.0ms Project: SQLProvider 7483.0ms Average Run: 24926.0ms Compiler: 9d5e36a7aa66f029640aa05f96bd7b54907aef81 Project: Paket 11266.0ms Project: FSharpx.Collections 2417.0ms Project: FSharp.Data 3831.0ms Project: SQLProvider 7383.0ms Average Run: 24898.0ms

But times obviously jump around a bit; because the unmodified changeset went from:

Compiler: 9d5e36a7aa66f029640aa05f96bd7b54907aef81 Average Run: 25363.0ms (run from other day) to Average Run: 24898.0ms (today's run)

Which is an almost 2% change; so probably hard to detect any minor changes, but atleast should detect any major problems...