Alternate Frame Rendering AFR работает в основном верно
“Рассмотренные выше архитектуры sort-first, sort-middle и sort-last описывают классическое разбиение графического конвейера по этапам обработки примитивов и служат базовой моделью для анализа пространственного распределения вычислительной нагрузки между несколькими графическими процессорами. Однако на практике такие подходы оказываются ограниченно применимыми в потребительских графических системах из-за необходимости тесной координации между GPU, значительных накладных расходов на обмен данными и сложности интеграции на уровне приложений. В связи с этим в реальных графических драйверах и API получили распространение альтернативные методы параллелизации рендеринга, ориентированные на минимальное вмешательство в графический конвейер приложения. Их основная цель заключается в повышении производительности за счёт прозрачного для разработчика распределения вычислений между несколькими GPU, без необходимости явного разбиения сцены или модификации алгоритма рендеринга. Такой подход позволяет использовать многопроцессорные графические конфигурации в массовых потребительских системах, где критически важны совместимость, простота интеграции и отсутствие требований к специализированной архитектуре приложения. К таким методам относится, в частности, \textit{alternate frame rendering (AFR)}, реализуемый на уровне драйверов графических процессоров.”
Summary
В потребительских графических системах традиционные схемы распределения нагрузки — sort‑first, sort‑middle и sort‑last — действительно редко применяются из‑за высокой сложности координации и больших накладных расходов. Вместо них драйверы и API используют более простые методы, такие как alternate frame rendering (AFR), который реализуется на уровне драйвера и позволяет распределять рендеринг между несколькими GPU без изменения кода приложения. Этот подход широко применяется в современных потребительских конфигурациях.
Sources 59 searched
- Comparison Based Sorting for Systems with Multiple GPUs
To avoid these overheads in · GPU accelerated applications that use sorting, several sort- ing algorithms that execute in the GPU have already been · proposed, including [9, 10, 11, 12, 13]. Most of the existing · GPU sorting algorithms only support single-GPU systems.
- GPU-ArraySort: A parallel, in-place algorithm for sorting large ...
A digital showcase of the research, scholarly and creative output of the faculty, students and staff of Western Michigan University community. The ScholarWorks repository is administered by WMU Libraries and serves as a permanent digital archive for these materials.
- Poly
Note: The copyright for most of the following papers has been transferred to a publisher. The ACM and IEEE grant permission to make digital or hard copies of part or all of the works they publish for personal or classroom use. See any copyright notices on the individual papers for details
- Alternate frame rendering - Wikipedia
Alternate frame rendering (AFR) is a technique of graphics rendering in personal computers which combines the work output of two or more graphics processing units (GPU) for a single monitor, in order to improve image quality, or to accelerate the rendering performance.
- Discussion - A tribute to multi-GPU consumer gaming pc setups: Were they ever worth it? | Tom's Hardware Forum
The primary issue with this technology is that it requires extensive driver tuning to make sure that the frames are rendered with the same time intervals in between, because if the frame times are even slightly off, stutter is perceived by the player. As has been well documented by now, both of these methods for multi-gpu rendering required an extensive amount of work either from NVIDIA, or from game manufacturers to work properly.
- Might as well delete this post --- Split Frame Rendering vs Alternate Frame Rendering | Tom's Hardware Forum
AFR. there are effort to bring back SFR with Mantle but they still did not solve the main issue with SFR that make GPU maker end up choosing AFR over SFR in the first place: scaling. ... Thanks. I've always had this theory that in SFR, dual cards (SLI/CF) could effectively combine their VRAM because each card would be only having to render part of a frame. Hence, each whole frame would be using both sets of VRAM. I always get an argument with that one, lol. But it's a moot point if SFR isn't used any more. ... AMD claim it is possible to combine VRAM with low level API multi GPU (though they did not mention if it's based on AFR or SFR or something new entirely) but to this day they never show us how that's work in real world situation either despite the claim.
- Why Multi GPU dont scale in every game? | Tom's Hardware Forum
What about where the shadow of an object on the left hits the right etc. managing those edge cases is tricky, vs AFR where each draws the whole screen alternately there are no edge cases as there is no edge. Thats the Game Engine problem to deal with ... in case of 2 GPU ... Looking at the split screen , I have better idea , even and odd vertical lines .. ODD on one card and Even on another. ... Thats the Game Engine problem to deal with ... $$$$ in development time. ... Original dual GPU's had this as an option (a long time ago, split frame rendering), BUT the top half of the screen in many games is less busy than the bottom half.
- (PDF) Multi-GPU Parallel Pipeline Rendering with Splitting Frame
The results demonstrate a significant improvement in performance with a very little perceptual loss in quality of our hybrid algorithm, when compared to the full ray tracing solution.
- SLI is dead, but dual-GPUs aren't: 5 ways to offload Lossless Scaling and streaming to a second card
As you probably know, vendor-specific tech like Nvidia DLSS, AMD FSR, and Intel XeSS is great, but it's also limited. Not all games support it, and even those that do will only use a single GPU to handle all that frame gen and upscaling.
This fact check is free and donation-based. $1 powers ~30 fact-checks.
