Video

Your work is invaluable, thank you so very much!

You said in an earlier episode, that commands START execution in the order they have been registered onto the command buffer, but do not necessarily FINISH execution in that same order. I think that could just be because some commands are just inherently more compute intensive than others, but could another reason be, the introduction of these cross pipeline dependencies? For example cmd X is not compute intensive and is registered to the command buffer first. CMD Y (which has been registered after CMD x) is a very compute intensive command. Now we introduce a stage dependency for a stage that cmd X is concerned with (for example a copy command) to another stage (for example the ALL_COMMANDS of the Graphics Pipeline). This would mean, that while cmd x which would be really quick in theory and also would start execution first in the queue, would most likely finish after cmd y, which was registered after cmd x to the command buffer, even though it would be way more compute intensive, right? Also you said in this episode, that the GPU has a mechanism/system to decide which workloads to parallelize and which clusters to assign those workloads to. If we assume a cmd A (or rather a pipeline stage) has a dependency to cmd B (or rather its pipeline ALL_COMMANDS stage) to finish, and maybe the cmd B needs only 2 clusters to do some work. Now, as to be as efficient as possible the GPU decides to assign 2 clusters to workloads from another cmd C (which has started after A and B), since cmd A is dependant on cmd B. Are these assumptions correct?

amazing

Thank you sir, best in depth explanation available 👍

Does this theory apply for Unreal Engine's pathtracing ?

Uniforms have two huge disadvantages over Storages. Veeery small size and no write access. Everyone tells this. I am looking for advantages of Uniforms more than month and still fail to find. Even here.

This is kind of surprise to me, how the first video in this list is probably the worse video about Vulkan on YT, and the rest of the videos in the list are the best videos on the YT :) After watching the first video, I was so disappointed and stopped watching at that point, until later when I found a Reddit post that someone linked to the second video in the list and I'm still keep watching the rest of the videos

Excellent presentation! One detail about the way you start the traversal of BVH, no need to start outside. A simple additional test on point in bound (the origin of the ray) when testing the tree nodes from root is enough and faster.

Dear Mr. Johannes Unterguggenberger, I would like to extend my sincere appreciation for the outstanding masterpiece you have created in your CZcams video on Synchronization in Vulkan. Your concise yet highly illuminating explanations, coupled with fantastic animation examples, showcase your exceptional expertise in the field. Thank you immensely for generously sharing your outstanding knowledge of Vulkan with the community.

We have similar last names. Mine is Uberguggenburger.

Excellent video. You make a strong case for Vulkan for someone starting off on graphics programming in 2024. I'd also recommend reading Jim Blinn's classic - A Trip Down the Graphics Pipeline, and play with an easy API like three.js before jumping into the deep end with Vulkan. (From a long-time OpenGL programmer.)

25:22 you mutilated the cat wtf :(

LMAO :)

this channel is gold. Explained fundamentals very well 👌👌👌👌

Thank you this is so good great talk ....

I don't understand the point of light source importance sampling when bidirectional path tracing exists.

@ 21:34 when you say binding descriptors to one bind point, will not disturb the others. Does that mean if I bind Desc. A to set 0 in the Graphics bind point, and then bind Desc. B to set 0 in the Compute bind point, A will still be bound to set 0 in Graphics? Because later on @ 22:01 it seems like the sets are being made copies of, one for each pipeline bind point??

From what I have read, ray-tracing is still not done 100% in games, and they still use rasterization.

Great series, thank you. Could you please explain the sentence, about images used as input attachments, "the access to an input attachment is framebuffer local, which means that only one pixel can be accessed, but not the neighbouring ones" - I don't know why that second part follows?

I'm watching these videos after going thru a ton of other videos..These are the best...with so much clear animations and fundamentals..thanks a lot

This is impressive! Will this be open-sourced?

👍

PLEEEEAAAASE REVIVE THIS!!!

Wow this looks great! Still has a lot work to do to get to the production level quality but I love the direction that you're going. Developing such tools and technology will help animators a lot and since the demand for animation is getting higher and higher this kind of technology will sure be welcomed in time. Also, I'm really fund of this project and idea. Is there anyway to be able to contribute to it?

See also the video where we stress test the network: czcams.com/video/YO8Fb2cpSRQ/video.html

If you are not sure what this is about, see the explainer video first! czcams.com/video/8YeDvrdvA5A/video.html

what should be noted if you are using volk?

What's the purpose of the demo code ?

These lectures are some of the most valuable resources for learning Vulkan in existence. Thank you so much for producing these videos, they've helped improve my understanding and mental model of Vulkan tremendously! Please please please make more of these.

OMG you explained sooo clearly!

"New Generation API" - Also, written in C, and plagued with all sorts of C shenanigans

thank you, very clear explanation

Dude these lectures have been so helpful in filling in the gaps in my understanding. Big thanks!!

Someone had a lot of fun animating that underscore at 30:32 :)

Great talk, thanks a lot.

This is great, as someone who has played in little with OpenGL and Metal more recently. You've changed my views on OpenGL. The high/level low level debate is in interesting one and I strongly believe in education its important for student to form mental models all the way from low level to high level. You need to know when to go high and when to go low. Metal is low-level and was fun but pretty disappointing when I ran the same code on a different Mac and I got complete nonsense visuals :-(

Very good, thank you, Johannes!

Thanks, but this doesn't appear to be for beginner anymore. Suddenly from ep2 to 3 became so hard.

Thank you sir. I really need to very basic learning material. Thank you for making the world a better place.

28:58 got plenty of them in zoos :P

Great lecture!

great video. Just wondering, are there any upcoming C/C++ lec in the future?

Seems like Vulkan has added a cross-vendor extension for task / mesh pipelines in September 2022. It's called VK_EXT_mesh_shader

This helps me a lot. thanks!

"this is my first video , bear with me" *proceeds to give the clearest explanation to AD in youtube*

32:35 stride - you better use sizeof(MyVertexData). Don't forget about padding! E.g. : if structure has 1 float and 1 char, its stride is 8, not sizeof(float) + sizeof(char)

it's time to let go opengl.

youtube is recommendation system broken. I got such great content push this week.

thanks.