I’m a heavy D5 user currently running an RTX 5080 with 16GB of VRAM. I’m consistently hitting the VRAM ceiling, resulting in frequent crashes and an inability to effectively multitask.
My work involves massive digital twins of manufacturing campuses, including both interior and exterior environments. My primary workflow is navigating these models in real time during live client presentations, with relatively little traditional rendering or animation output.
My company is considering upgrading to an RTX 5090 or potentially an RTX PRO 6000, as even an idle D5 scene consumes nearly all 16GB of my available VRAM.
I’m particularly curious about NVIDIA’s DGX Spark and its 128GB of unified memory. Given the substantial price difference, and the fact that D5 is featured on NVIDIA’s Spark website, I’m wondering whether anyone on the D5 team has experience testing this hardware.
I realize performance benchmarks may be limited, but I’d be interested in understanding how D5 is expected to perform on Spark, particularly with extremely large, VRAM-intensive models.
My main question is whether Spark’s larger unified memory could offer a meaningful advantage for real-time navigation of massive scenes, or whether its GPU performance and memory bandwidth would make a traditional high-end RTX card the better choice.
I’d appreciate any insights from the D5 team or others exploring similar workflows.