This is a rather bad intro to a great topic, the comments point out a trove of errors and false parallels in the video.
All the videos I鈥檝e encountered from this channel have a disconcerting combination of incredible production value, and highly questionable attention to detail with respect to the information that鈥檚 being disseminated
Vulkan is probably better than OpenGL. But don鈥檛 confuse that with Vulkan being good. It鈥檚 not. Vulkan is bloody terrible. Khronos design by committee is incapable of designing an API that doesn鈥檛 suck. It鈥檚 a shame we have such few choices.
Recent Blender versions have OpenGL version dependencies and VRAM requirements that aren't supported on a lot of older or lower-end PCs.
For fans of, say, older laptops that have "mechanical" keyboards, or people who only have access to less-powerful hardware, I wonder whether Vulkan will re-enable support for the latest Blender.
(Though the VRAM might still be a problem. As will local rendering of expensive animations.)
What鈥檚 really groundbreaking is the amount of ignorance displayed in this video. Also, I鈥檓 curious how long will it take for blender to reach performance parity with OpenGL. Houdini has been taking a few years now and VK is still 2x slower then opengl apparently.
I hate seeing how awesome Blender...free software is..in comparison to roughly $5k a seat program like Revit who's graphics rendering looks like it comes from a 90s shareware floppy disk.
Shipping Vulkan in production on Linux is a challenge. Chrome was dealing with it for a while. Recently, with Zed ported to Vulkan, we saw the variety of half-broken platforms and user configurations.
I'd recommend Blended to not close the door on OpenGL and instead keeping it as a compatibility fallback.
I just tested a cold start of blender 4.4.3 on my system (10th gen i7, 32gb ram, 4070 ti S 16gb, Fedora Linux) it was about 8 seconds, and that was through steam... Warm start was about 3 seconds.
Abridgement - made by this poster, not automated - over the transcript:
> This year, Blender is transitioning to Vulkan, which marks one of the biggest performance boosts in its history: we始re talking about a leap from a cold start time of a minute 52 to opening Blender in 6 seconds. // Vulkan ... simply put, it始s the industry standard API. [...] Blender始s current API is OpenGL, which has some major flaws. OpenGL ... really struggles using multi-core CPUs. // OpenGL ... automatically sets up most low-level features like memory management, synchronization, and hardware interactions. ... But high-level APIs come with a huge trade-off: they始re slow. // Writing code that performs better in Vulkan is a difficult task ... OpenGL is like infinitely easier than Vulkan. Vulkan ... is a low-level API. [...] Vulkan can perform up to 18 times better on certain tasks. [...] For tasks that heavily rely on the UI, we will see a much more performant system. [...] Essentially, interactions between the user and the program will be much faster, but the underlying automatic system will not see a direct speed increase from Vulkan. ... While I would love for my final render to be faster, I would much prefer to have a performant viewport that始s more stable and faster, which is what we始ll have with Vulkan. But this is not the only thing it brings to the table. Many corporate sponsors have granted money to Blender primarily for a Vulkan integration. AMD is donating $120,000 per year to help fund Blender始s transition to Vulkan.