Doom uses OpenGL for rendering.
Doom uses OpenGL for rendering.
Vulkan is the upcoming API designed to eventually supersede open GL. If you're using an AMD graphics card, it's recommended to run Vulkan for better performance gains. On NVIDIA platforms, both technologies currently offer similar results because the driver manages asynchronous computation at the system level rather than the hardware level, unlike AMD cards.
Using a GPU that supports Vulkan will improve performance, while older cards without this support are fine.
We refer to "asynchronous compute" as running several hardware queues at once within the chip, meaning there isn't a separate "driver-based" version. Its character is determined by how the hardware is utilized. Both Maxwell and Pascal offer multiple queues, which provides the required infrastructure. For Maxwell cards, the driver blocks this capability, forcing all instructions through one queue. Pascal supports scaling for asynchronous compute, while AMD's GCN implementation appears more effective in practice.
I evaluated two setups for comparison: my second rig with an Intel i7 and high-end graphics versus a third rig using AMD FX8350. I tested both OGL and Vulkan, noting significant differences in performance—Vulkan delivered 60-90fps to 100+ with the latter, while OGL reached 70+ with over 100fps. The Kepler model lacked async compute, which affected Vulkan’s speed. On my HD7950, I was surprised by higher frame rates using Vulkan compared to my Titans’ OGL setup, suggesting a potential drop if I switched back to OGL for those GPUs.