Code: <https://gist.github.com/HurricanKai/0125b844666b26b531bb204f099d513d>, would be cool if someo

Code: https://gist.github.com/HurricanKai/0125b844666b26b531bb204f099d513d, would be cool if someone can test this on some other devices (maybe even iGPU?) to see what can be achieved there

KaiOP•6/5/22, 10:49 AM

note that there are a bunch of hardcoded values in there and the program will just die when something goes wrong

KaiOP•6/5/22, 10:50 AM

Most important are the config values at line 92, that decides how much data is shoved down the lanes and in how large blocks

KaiOP•6/5/22, 10:51 AM

(by passing instead of as the last parameter on line 96 an ideal scenario can be created, where there is no wasted memory. This seems to have little to no influence on speed)

KaiOP•6/5/22, 10:52 AM

Also note that my device has 2 Transfer queues (I'm using both in round-robin here) & I'm not using Graphics or Compute queue here, so in theory there is zero interference with other parts, other then memory usage

KaiOP•6/5/22, 10:58 AM

I can also get up to 80% throughput on smaller (100mb) blocks by tuning the block count.

KaiOP•6/5/22, 10:59 AM

@techpizza iirc you were interested in this. If you want I'm also happy to explain how this works

KaiOP•6/5/22, 11:01 AM

quite happy with the memory footprint here. That big buffer on the right is the target

KaiOP•6/5/22, 11:17 AM

I'll measure the precise throughput using write timestamp later, but I don't expect there to be any surprises there

Perksey•6/5/22, 12:07 PM

i love how Qualcomm are basically the only ones keeping OpenGL alive now

Perksey•6/5/22, 12:07 PM

they just added variable rate shading to GLES

Supine•6/5/22, 12:17 PM

lol?

KKai Code: <https://gist.github.com/HurricanKai/0125b844666b26b531bb204f099d513d>, wo...

TechPizza•6/5/22, 1:06 PM

looks good to me. i'll check it out later on both my igpu and dgpu

TechPizza•6/5/22, 1:06 PM

too bad that i don't have an AMD laptop anymore

Perksey•6/5/22, 7:00 PM

god i was just scrolled really far back in #dev-feed and had a minor panic attack as i thought my github had been hacked

Aqua•6/6/22, 7:52 PM

GLFW controller support is a bit lacking...
No rumble support, and to add insult to injury I can't get GLFW to recognize any controller I throw at it, lol

Lyris the Kitori•6/6/22, 7:53 PM

controllers have been very iffy for me too

Lyris the Kitori•6/6/22, 7:53 PM

not sure why

Lyris the Kitori•6/6/22, 7:53 PM

tried to look into it but got nowhere

Aqua•6/6/22, 7:53 PM

hrm

Aqua•6/6/22, 7:54 PM

i mean the issue for rumble support is literally 9 years old https://github.com/glfw/glfw/issues/57

Aqua•6/6/22, 7:54 PM

buuut no point having any of that if it won't work with the darn things in the first place

Aqua•6/6/22, 8:02 PM

lmao it only sees it as a joystick
at index 3

Aqua•6/6/22, 8:39 PM

ok, I got it working, and did some hackery to make it appear at index 0
but weirdly I have to manually update the gamepad mappings every time, even though glfw is supposed to come with default ones? hm

Aqua•6/6/22, 8:39 PM

that's a bit annoying

queenvicky•6/8/22, 12:39 PM

String-Chain simulation.
totally not proving my dislike that
delta
delta
increments if the window is focused

Lyris the Kitori•6/10/22, 1:03 AM

i wonder how much id gain from rewriting my renderer's performance critical things in Zig then interoping with Zig

Lyris the Kitori•6/10/22, 1:03 AM

so itd be mixed Silk/zgl

Lyris the Kitori•6/10/22, 1:13 AM

cus at the very worst im making 66 gl calls per flush of data to the display (glactivetexture, glbindtexture taking up 64 gl calls with 32 textures used) then a setsubdata and drawelementsinstanced

Lyris the Kitori•6/10/22, 5:55 AM

unrelated to silk but i figured people here might be able to help, how would i go about generating c# bindings for a cpp library? im trying to use this fork of libInput https://github.com/netpipe/libInput/ to allow me to distinguish multiple keyboards

i tried CppSharp but it doesnt generate any actual bindings and i cant figure out why

Lyris the Kitori•6/10/22, 5:56 AM

heres the cppsharp log if some wizard knows what went wrong, but it only generates a blank Std.cs file

message.txt30.85KB

Perksey•6/10/22, 9:27 AM

I advise against making C++ bindings

Perksey•6/10/22, 9:27 AM

The ABI is pretty non-standard and no tools have proper support for it, best bet is making a C layer and binding to that

LLyris the Kitori cus at the *very* worst im making 66 gl calls per flush of data to the display (...

KaiOP•6/10/22, 9:36 AM

You'll gain little to no performance improvement in the GL calls. Maybe you get some improvement in CPU heavy workloads, but I kinda doubt it.

Supine•6/10/22, 1:02 PM

Don't care.

KaiOP•6/10/22, 1:03 PM

Supine•6/10/22, 1:06 PM

It's better to implement use of atlasing/texture array/array of textures than to optimize 64 bind texture calls

KaiOP•6/10/22, 1:07 PM

I mean it depends massively on what the performance characteristics are. Depending on what the textures are you just have to use normal binds.

KaiOP•6/10/22, 1:07 PM

There's rarely a silver bullet in (graphics) programming

Perksey•6/10/22, 1:15 PM

The max 4 bound textures in standard Vulkan 1.2 will really make you work hard to find them though!

KaiOP•6/10/22, 1:15 PM

I mean that's the minimum of the minimum

Perksey•6/10/22, 1:16 PM

yeah ofc, just the constraints you have to work with to have guaranteed portable code

Perksey•6/10/22, 1:16 PM

glad 1.3 has increased it but I hate the lag time for adoption :/

KaiOP•6/10/22, 1:20 PM

There's been the introduction of the compatability subset too, and that stuff is wild

KaiOP•6/10/22, 1:20 PM

I've tried building on that, but it's sooo many hoops to jump

PPerksey The ABI is pretty non-standard and no tools have proper support for it, best bet...

Lyris the Kitori•6/10/22, 6:42 PM

darn i can barely see with glasses at all let alone see incremented

Lyris the Kitori•6/10/22, 6:43 PM

have you tried seeing incremented

Lyris the Kitori•6/10/22, 6:43 PM

its hard

SSupine It's better to implement use of atlasing/texture array/array of textures than to...

Lyris the Kitori•6/10/22, 6:44 PM

oh shit is a thing, so i dont need to do any of that

Lyris the Kitori•6/10/22, 6:44 PM

and i can do it all in 1 gl call

Lyris the Kitori•6/10/22, 6:45 PM

(on gl4 at least, ill have to reimplement glBindTextures using glActiveTexture glBindTexture for gl2 and gles)