Tim Cassell

Even if an architecture doesn't support these operations, couldn't they be implemented as CompareExchange loops? ```cs public static short CompareExchange(ref short location, short value, short comparand) { ref int locationAsInt...

> @timcassell `newValueAsInt = initialValue;` should be perhaps `newValueAsInt = initialValueAsInt;`? Ah yes, good catch. > And maybe `if (initialValue == comparand)` instead of `if (initialValue != comparand)`. No, that...

> In your code, if the byte/short/etc. is just before memory that's not accessible, such that extending by a few more bytes steps over that boundary, this could AV /...

> Could `byte` and `char` overloads be also considered? They'd be trivial to implement with this. Perhaps you meant `bool`? `byte` is already part of the proposal.

.NET Core 3.1 and 5.0 measurement issue: https://github.com/dotnet/runtime/issues/45446

Looks like measurement errors are fixed in .Net 7 as of https://github.com/dotnet/runtime/pull/67160.

I was thinking about how to achieve this. What @stephentoub said makes sense, so I was thinking of a way to do it without having to refactor the entire engine,...

With the interest in supporting Unity #1860 and potentially its async operations which usually require execution over multiple frames, would it make sense to refactor the entire engine to be...

@adamsitnik The way I was imagining an async engine working with sync benchmarks is, still using an `IMeasurer` interface, `IMeasurer.Measure(IClock)` returns a `ValueTask` which is already resolved with the measured...

I guess the next question is, does the engine care what thread it's running on if it's async? Because if the engine is async and does `double nanoseconds = await...