Metal GPU Support: Unlocking Performance in Rendering

As designers and engineers, we often find ourselves pushing the limits of our hardware to achieve high-quality renders. At my office, we have a dedicated workstation for rendering large files, equipped with a 14-core Xeon processor. While investing in expensive video cards and enclosures has certainly improved our user experience, it's clear that most users ultimately rely on 4 CPU cores when rendering at high quality.

One of the key factors limiting GPU performance is the way Vectorworks utilizes processing power on these ever-more-expensive cards. Interestingly, a model that would utilize 90% of the GPU clock on my previous card (a 460) may only use around 65% on my newer 580, and less than 40% on my Radeon VII. This doesn't mean we're running out of VRAM; rather, Vectorworks is simply not utilizing the full potential of these high-end cards.

The difference in framerate stability and model pop-in is noticeable with each step up the ladder when working with large files. In our office, we've experimented with various GPUs, including a Titan X(p) on a Windows machine and Vega FE cards with 12GB and 16GB respectively. On my main workstation, I've upgraded from an AMD Pro 460 discrete internal GPU with 4GB to an RX580 with 8GB and now a Radeon VII with 16GB.

With the latter, I'm happy to say that I recently acquired a screaming fast video card for my primary workstation. But what if Vectorworks could unlock even more performance potential? Adding a 'compute on GPU' option to Renderworks modes or creating a new class of rendering modes that utilize GPU compute would be game-changers.

These features would allow designers and engineers to harness the massive processing power of modern GPUs, achieving faster render times while maintaining fidelity options. For now, we can only imagine what's possible if Vectorworks were to explore this area further.

Metal gpu support