I apologize for being gone for so long, my university work has been keeping me busy as usual. I'm sorry to say that it's probably always going to be this way until I graduate two years from now, but I'll still continue to get faster with each new video and learn more ways to carve out time to work on the engine and these videos. Also I'm sorry that I never made any community posts or comments or anything showing my progress on this devlog, I promise to do that going forward. My determination has not been defeated, there are wonderful things in the future for this project B)
I tried to keep them short but couldn't find a good contraction unfortunately :'3 But I don't think it'll be an issue since in the future all assets will get compiled into a single binary blob file when you actually build a game with Omni for the sake of optimization and asset protection. Thank you for watching and thank you for commenting, my friend :]
outright saying that opengl is slow and vulkan is fast is very misleading, it all depends on how you implement stuff. you could write a renderer in opengl that is fast even by modern standards by using a mix of indirect rendering with compute shader culling and bindless textures. same applies for vulkan, you could easily write a slow renderer in it by just doing a drawcall / mesh. the biggest difference between the two is how much control each gives you over the hardware, vulkan is a lower level api, it expects the programmer to know how the gpu works and build everyything by hand. opengl on the other hand abstracts a lot of stuff away and doesnt "trust" you as much as vulkan does. and considering that the opengl drivers have been through a lot of updates and optimisations id argue that most people would end up with a faster renderer using opengl simply because writing something in vulkan is just harder and shooting yourself in the foot or doing something wrong that you just simply dont know about is much easier in it.
OpenGL is definitely easier to use and most people would probably have better results with it if they're just getting into graphics, but assuming that you're doing things in mostly the correct way Vulkan will be faster unless you're on some older hardware. Also like I said in the video, there are some really great guides on how to use both, so I think it's fair to compare them when they're both at their best. Of course, everything in programming depends on how you implement it, but because both OpenGL and Vulkan have excellent documentation and resources, I feel like as long as you can follow guides and documentation you will most likely have a faster renderer in Vulkan. I appreciate your comment and I hope that this helps your concern :]
1:52 is not 100% true. We have the option to make OpenGL almost as fast as Vulkan using a set of techniques called AZDO (Approaching Zero Driver Overhead). OpenGL with AZDO is fast, but in Vulkan, you still have more control over the rendering process, giving you more options to optimize bottlenecks.
I really should have included that to begin with lol :'3 But here are my specs: AMD Ryzen 7 4800H 2.9 GHz 8 GB RAM NVIDIA GeForce GTZ 1660 Ti Windows 10 Also both versions were compiled in release mode with Visual Studio 2022, I could've done even more with the old bgfx test but I was having a lot of trouble understanding it at the time
Thank you! Another question. During the benchmark, was the transform data updated every frame? Or was it the same? Trying to reason about why my renderer runs like crap lol.
@@oh-facts In the bgfx benchmark I was updating the transform data every frame but in the OpenGL benchmark I calculated all the transforms on the first frame and then cached them. If you're also trying to make a 2D renderer, have you looked into geometry instancing yet?
man, you need to upload more frequently. Also about your claim that you need only opengl and vulkan. the xbox i believe use direct12, Playstation use their own api, apple abandoned vulkan project and made their own api, metal. so to target xbox, ps4&5, macOS/iPhone/ipad you still need to add another set of APIs.
I apologize for being gone for so long, my university work has been keeping me busy as usual. I'm sorry to say that it's probably always going to be this way until I graduate two years from now, but I'll still continue to get faster with each new video and learn more ways to carve out time to work on the engine and these videos. Also I'm sorry that I never made any community posts or comments or anything showing my progress on this devlog, I promise to do that going forward. My determination has not been defeated, there are wonderful things in the future for this project B)
YEAHHHHHH OMNI UPDATE GRAHHHH 🔥💥
Damn that was 10 months.
But I know how long engine and or game development can take.
Glad to see you back
Glad to see you are still going strong with it :) worth the wait
Hope it releases someday, can't wait till I can try it out to make games
Amazing new channel I found greetings btw
Hell yeah loosejuice
Long time no see :)
Some of the file extensions are pretty long, I hope that isn't an issue on some systems.
I tried to keep them short but couldn't find a good contraction unfortunately :'3 But I don't think it'll be an issue since in the future all assets will get compiled into a single binary blob file when you actually build a game with Omni for the sake of optimization and asset protection. Thank you for watching and thank you for commenting, my friend :]
HOLY SHIT
IVE BEEN WAITING
I TURNED GAY OVER THIS TIME
NEW OMNI UPDATE RAHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH
how long, babe. I miss u, bro. As a computer graphics aprentice is quite lonely and sad, but your content tell me that is not true
instead of Vulkan, maybe you can use WebGPU? like the cpp version of it. Its a good API thats powerfull, but not too complex like Vulkan
outright saying that opengl is slow and vulkan is fast is very misleading, it all depends on how you implement stuff. you could write a renderer in opengl that is fast even by modern standards by using a mix of indirect rendering with compute shader culling and bindless textures. same applies for vulkan, you could easily write a slow renderer in it by just doing a drawcall / mesh. the biggest difference between the two is how much control each gives you over the hardware, vulkan is a lower level api, it expects the programmer to know how the gpu works and build everyything by hand. opengl on the other hand abstracts a lot of stuff away and doesnt "trust" you as much as vulkan does. and considering that the opengl drivers have been through a lot of updates and optimisations id argue that most people would end up with a faster renderer using opengl simply because writing something in vulkan is just harder and shooting yourself in the foot or doing something wrong that you just simply dont know about is much easier in it.
👆🤓
OpenGL is definitely easier to use and most people would probably have better results with it if they're just getting into graphics, but assuming that you're doing things in mostly the correct way Vulkan will be faster unless you're on some older hardware. Also like I said in the video, there are some really great guides on how to use both, so I think it's fair to compare them when they're both at their best. Of course, everything in programming depends on how you implement it, but because both OpenGL and Vulkan have excellent documentation and resources, I feel like as long as you can follow guides and documentation you will most likely have a faster renderer in Vulkan. I appreciate your comment and I hope that this helps your concern :]
wo
1:52 is not 100% true. We have the option to make OpenGL almost as fast as Vulkan using a set of techniques called AZDO (Approaching Zero Driver Overhead). OpenGL with AZDO is fast, but in Vulkan, you still have more control over the rendering process, giving you more options to optimize bottlenecks.
Hi, what are your computer specs? (So I can better understand the perf test at 2:25)
I really should have included that to begin with lol :'3
But here are my specs:
AMD Ryzen 7 4800H 2.9 GHz
8 GB RAM
NVIDIA GeForce GTZ 1660 Ti
Windows 10
Also both versions were compiled in release mode with Visual Studio 2022, I could've done even more with the old bgfx test but I was having a lot of trouble understanding it at the time
Thank you!
Another question. During the benchmark, was the transform data updated every frame? Or was it the same? Trying to reason about why my renderer runs like crap lol.
@@oh-facts In the bgfx benchmark I was updating the transform data every frame but in the OpenGL benchmark I calculated all the transforms on the first frame and then cached them. If you're also trying to make a 2D renderer, have you looked into geometry instancing yet?
man, you need to upload more frequently.
Also about your claim that you need only opengl and vulkan. the xbox i believe use direct12, Playstation use their own api, apple abandoned vulkan project and made their own api, metal. so to target xbox, ps4&5, macOS/iPhone/ipad you still need to add another set of APIs.
wtf was that ending?!
;]