Let’s break down what’s new with the RTX 6000. Firstly, and just like its GeForce RTX 40 Series counterparts, the GPU is based on the new Lovelace architecture. To that end, it is flanked by the GPU maker’s next-generation CUDA Cores – 18176 units of it, to be precise – 3rd generation Ray-tracing (RT) Cores that churns out double the throughput over the previous generation RT Cores; and new 4th generation Tensor Cores, featuring fast AI Compute performance and, of course, double the performance over the previous generation. Oh, and just as before, the whole thing is supported over the PCIe 4.0 interface, which also provides double the bandwidth over the previous PCIe 3.0 interface. As for how much graphics memory the RTX 6000, the GPU gets a whopping 48GB of GDDR6 memory. While that may seem like a lot of memory being allocated to a GPU, it isn’t uncommon for workstation graphics cards to ship out with a higher-than-average amount of graphics memory; unlike consumer-ready GPUs, data scientists and creative professionals benefit immensely from the extra memory, especially when the tasks itself are, for example, render and simulation-intensive. Lastly, the RTX 6000 is also Virtualisation-Ready, meaning that the card ships out with support for NVIDIA RTX Virtual Workstation (vWS) software, which is just another way of saying that outfits that purchase the card can and will have the ability to repurpose a workstation into multiple virtual machines, thereby allowing multiple users to share the prowess of the GPU, either on-site or remotely. As for how much the RTX 6000 is going to set you back, that actually remains the million-dollar question. At the time of writing, NVIDIA still hasn’t quoted a price tag for it, but given that its predecessor, the RTX A6000 still goes for nearly US$5000 (~RM22762) a pop, it wouldn’t surprise us if this ends up costing you a kidney, or two. (Source: NVIDIA)
