The Graphics Device Interface (GDI/GDI+) of Windows just can’t be killed. No matter if it’s older graphical applications up to CAD programs or just displaying the simple GUI of many programs – it’s still output via the GDI like there’s no tomorrow. But all Windows versions after XP don’t support hardware accelerated output for the graphics functions anymore, because with the introduction of the unified shaders and the elimination of the specialized 2D units on the graphics cards, all these things virtually run through the driver as a kind of wrapper.
The CPU has to do most of the computing, and in the end it’s also up to the driver how efficiently and with what overhead the output to the D3D is done. The driver models since Vista still support, at least in parts, hardware acceleration when blitting, i.e. copying graphics content within the memory. This is where the relatively weak CPU comes into play again in the final result, which slows down the graphics card a bit. But it’s all really fluid, no issue.
AMD and Nvidia take different approaches here, so it’s very easy to see which of the individual functions can be better implemented (or not) by the respective driver. According to my research back in 2009, the AMD driver team in Toronto in particular did a lot of catching up, as some of the ATI cards at the time had glaring disadvantages. Many features in AMD drivers still benefit from this to this day. A current inventory shows that they are now pretty close together.
AMD is still a bit weak in TextOut, while in some areas like lines, splines, and rectangles it’s effectively a tie, but things like polygons are where AMD really shines. NVIDIA can still improve a bit here.
Now let’s move on to the rendering. First, let’s take stock with CUDA vs. OpenCL first, which includes all Radeon cards. Purely in terms of computing power, even the notebook with the 150-watt GPU and CUDA is still really competitive, also a realization, and the frugal GPU even eats up a Quadro RTX 6000, albeit just barely.
You can also run and benchmark the AI-based detail enhancement in Adobe Lightroom Classic with multiple images in a catalog. This is exactly what I repeated with the new notebook. At 150 watts TDP, the mobile RTX 3080 is significantly faster than the older RTX 2080 Ti with almost twice the TDP!
This is where things get really interesting, as Arnold is Autodesk’s latest ray tracing renderer designed to meet the demands of feature length animation and visual effects. Arnold is used in over 300 studios worldwide, including ILM, Framestore, MPC, The Mill and Digic Pictures among others. Arnold has also been the primary renderer on dozens of films, from Monster House and Cloudy with a Chance of Meatballs to Pacific Rim and Gravity . He is available as a standalone renderer for Linux, Windows, and Mac OS X, and as a plug-in for Maya, 3ds Max , Houdini, Cinema 4D, and Katana.
The combination of fast, RTX-accelerated rendering with OptiX and GPU-accelerated AI denoising (denoising filters) also enables rapid iteration through changes in materials, lighting, camera position, and environment, and the effects are visible with lightning-fast updates. The OptiX denoiser is built into Arnold for use with IPR and Look Dev and can be applied to any AOV. The performance of the AI-Denoiser in the editing phase is strikingly high, but it is recommended to use the Arnold-Denoiser in the final rendering to avoid sporadic flickering. But I don’t think anyone will ever think of rendering the finished film on a notebook. Not yet.
You can see the advantage of the RTX cards over the older Pascal models very clearly, as the slowest Turing RTX is still almost twice as fast as the older Pascal flagship. That’s what it takes to make it.
However, I don’t bother with all the CAD applications, because the display is simply not enough for that. Even the applications shown here are borderline, because you don’t know what to do with all the menus and toolbars.
18 Antworten
Kommentar
Lade neue Kommentare
Urgestein
1
Neuling
1
Urgestein
Neuling
1
Neuling
Mitglied
Urgestein
Mitglied
Urgestein
Veteran
Neuling
1
Alle Kommentare lesen unter igor´sLAB Community →