the other benefit of this, is that it also better meshes with GPU compute doing integers on the GPU

the other benefit of this, is that it also better meshes with GPU compute
doing integers on the GPU is generally a bad time, and not always well supported

Perksey•10/21/23, 10:23 PM

it's worth noting there is some precedent to multiple types in silk maths e.g. see Box2DBox2D vs RectangleRectangle

Ttannergooding the other benefit of this, is that it also better meshes with GPU compute doing ...

tannergoodingOP•10/21/23, 10:25 PM

(not well supported meaning, GPUs don't often have many integer compute units; so you're often getting 25% of the throughput by using integers)

Kai•10/21/23, 10:27 PM

That's the other thing, and I'd need to do have a look around there. How important is it to have Vector3<T>Vector3<T> including integers to avoid extra overloads in bindings?
@Perksey do you have a clue? I'm fairly confident that there's a couple of cases in Vk that take Vector3<T> and you get to put whatever type

khitiara•10/21/23, 10:27 PM

oh hmm

khitiara•10/21/23, 10:27 PM

im not remembering anything

Kai•10/21/23, 10:28 PM

Cause Silk isn't just about extending the BCL maths to vectors here, we also (and maybe primarily) want this to work with bindings so CPUGPU is fairly painless

tannergoodingOP•10/21/23, 10:28 PM

extending the BCL maths to vectors here

(this is happening in .NET 9, it's already been partially started in an OOB in 8)

Kai•10/21/23, 10:30 PM

given your preferred design I'm assuming float only

tannergoodingOP•10/21/23, 10:30 PM

for the .NET 8 OOB feature it's float
but for .NET 9 it will be all floating-point types

KKai That's the other thing, and I'd need to do have a look around there. How importa...

Perksey•10/21/23, 10:31 PM

we haven't discussed this in terms of SilkX yet. for OpenGL i know there are a lot of manual overloads today, i'll check BT to see if there's any automated ones

tannergoodingOP•10/21/23, 10:32 PM

the OOB feature is TensorPrimitivesTensorPrimitives, which is math APIs over ROSpan<T>ROSpan<T> (so TensorPrimitives.Log(ROSpan<float> x, Span<float> destination)TensorPrimitives.Log(ROSpan<float> x, Span<float> destination))
in .NET 9, that will expand to where T : IFloatingPointIeee754<T>where T : IFloatingPointIeee754<T>
and also to Vector128.Log(Vector128<float> x)Vector128.Log(Vector128<float> x) and friends (+ double overloads)

Kai•10/21/23, 10:32 PM

floating-point types

yeah that's sort of what I ment

Ttannergooding the OOB feature is `TensorPrimitives`, which is math APIs over `ROSpan<T>` (so `...

Kai•10/21/23, 10:32 PM

ooooh that is really cool

tannergoodingOP•10/21/23, 10:32 PM

ah, well then yes
because Cos(int)Cos(int) doesn't make sense

Perksey•10/21/23, 10:32 PM

doesn't appear to be any automated ones

Perksey•10/21/23, 10:33 PM

so there is technically no precedent here

PPerksey doesn't appear to be any automated ones

Kai•10/21/23, 10:33 PM

not because we will generate 10 overloads for each possible T? Or because the APIs don't have any

Perksey•10/21/23, 10:33 PM

BuildTools does not generate any Silk.NET.Maths or System.Numerics overloads today

tannergoodingOP•10/21/23, 10:33 PM

we'll likely have some APIs where it can minimally make sense for integers though
like Log2(int)Log2(int) (and already have these for scalars, of course)

Kai•10/21/23, 10:38 PM

hrm I'll think this over but it's annoyingly a great argument for making more types. Hard to imagine what the rest of the types look like if we were to have Point + Vector. There is some interesting properties if we separate there also

Kai•10/21/23, 10:40 PM

Also thanks as always Tanner for discussing all this and sharing your wisdom :)

tannergoodingOP•10/21/23, 10:52 PM

Sure thing! And I’m definitely not always right nor is the BCL and it’s entire possible there’s alternatives that no one has considered yet.

I’m sure silk will end up with something that satisfies most of the user base regardless of what’s decided

nora•10/21/23, 11:32 PM

are there any plans to make these apis distinct between namespaces?

nora•10/21/23, 11:32 PM

or is this a case of me cramming everything into one file

Perksey•10/21/23, 11:33 PM

are there any plans to make these apis distinct between namespaces?

I think you've answered your own question

Perksey•10/21/23, 11:34 PM

namespaces are the language's built-in feature for disambiguating APIs

nora•10/21/23, 11:34 PM

ah i suppose i had a bit of a brain lapse

nora•10/21/23, 11:34 PM

my bad

Perksey•10/21/23, 11:35 PM

wow, that worked?

Perksey•10/21/23, 11:35 PM

haha no worries, not the first time we have got that complaint

nora•10/21/23, 11:36 PM

i was just a bit concerned because this doesnt look the best but its not the end of the world

nora•10/21/23, 11:36 PM

migrating some of my code to the silk apis

Perksey•10/21/23, 11:37 PM

yeah, there are other things you can do as well like using SpvCompiler = Silk.NET.SPIRV.Cross.Compilerusing SpvCompiler = Silk.NET.SPIRV.Cross.Compiler

nora•10/21/23, 11:37 PM

true

Perksey•10/21/23, 11:37 PM

but generally namespaces exist for a reason and i usually have a really hard time conveying that to people

Perksey•10/21/23, 11:37 PM

primarily because unlike something like C++, people don't really think about them in C#

PPerksey but generally namespaces exist for a reason and i usually have a really hard tim...

tannergoodingOP•10/22/23, 12:42 AM

Noting, however, that if two namespaces are expected to be commonly used together, it’s generally best practice to avoid conflicts and avoid explicitly disambiguate the real name instead

tannergoodingOP•10/22/23, 12:42 AM

No clue if that’s the case here or not, but if it is then API review would ask us to reconsider the names

Perksey•10/22/23, 12:43 AM

it’s an odd case

Perksey•10/22/23, 12:43 AM

especially considering we automatically trim prefixes

Perksey•10/22/23, 12:43 AM

perhaps if we’re concerned about namespaces we shouldn’t do that

Perksey•10/22/23, 12:43 AM

just feels wasteful

Perksey•10/22/23, 12:44 AM

perhaps we could add global usings with the prefix in to help for these cases

tannergoodingOP•10/22/23, 12:45 AM

It might be a one off case where it’s worth it. Compiler is itself fairly generic as a term, so SpirvCompiler makes it “clearer” and reduces conflict chance

I’d personally expect something just named Compiler was probably abstract

tannergoodingOP•10/22/23, 12:46 AM

I know ClangSharp has the capability to do one off remapping and aliasing. Might be worth it for the silk tool as well?

tannergoodingOP•10/22/23, 12:47 AM

Would probably also make it easier for enums and other cases where it’s PREFIX_3NAME and such

Perksey•10/22/23, 12:48 AM

yeah we have one off remapping as well, we just don’t really consider names beyond the individual project rn.

tannergoodingOP•10/22/23, 12:48 AM

Gotcha

Ttannergooding Would probably also make it easier for enums and other cases where it’s PREFIX_3...

Perksey•10/22/23, 12:49 AM

iirc we currently do implicitly add the prefix when the name isn’t legal in C#

Perksey•10/22/23, 12:49 AM

e.g. the numbers

the other benefit of this, is that it also better meshes with GPU compute doing integers on the GPU

Similar Threads