October 24, 2018
When NVIDIA revealed its new Turing architecture last summer, it described the micro-architecture as bringing dedicated processors tuned for ray-tracing and artificial intelligence (AI) to desktop engineering workstations. It also previewed its first Turing-based products, the Quadro RTX 8000, Quadro RTX 6000 and Quadro RTX 5000 GPUs (graphics processing units). The company said at the time that it expected Quadro RTX GPUs to be available in the fourth quarter of 2018.
PNY Technologies, an authorized channel partner for NVIDIA Quadro GPUs, has announced that it anticipates having the NVIDIA Quadro RTX 6000 and RTX 5000 in stock shortly and has enabled its reseller, distribution and e-tail partners to begin taking pre-orders for them now. A PNY spokesperson tells DE that initial deliveries are planned to begin in November. Shipments of the top-of-the-line Quadro RTX 8000 should commence in early 2019.
For ray-tracing, the NVIDIA RTX Quadro GPUs' Turing architecture deploys RT Cores. These dedicated processors are said to compute how light and sound travel in 3D environments faster than previous technologies allowed. NVIDIA reports that the RT Cores can achieve light and sound computations at rates of up to 10 giga rays per second (10 billion rays cast per second). By way of comparison, NVIDIA reports that this acceleration speeds up real-time ray-tracing computations by 25x over its previous Pascal architecture generation.
The NVIDIA RTX Quadro GPUs' Turing architecture also deploys processors dedicated to AI computations. Called Tensor Cores, these processors provide up to 500 trillion operations per second, which can accelerate deep learning training and inferencing. NVIDIA reports that this level of performance can “dramatically” accelerate AI-enhanced features such as denoising, resolution scaling and video retiming.
Additionally, the Turing architecture has an enhanced graphics pipeline and new programmable shading technologies that are said to improve raster performance over the earlier Pascal generation GPUs. According to NVIDIA, these technologies include variable-rate shading, texture-space shading and multiview rendering, all of which combine to provide for more fluid interactivity with large models and scenes as well as improved VR (virtual reality) experiences.
All NVIDIA Quadro RTX GPUs come in a dual-slot form factor. Features shared across the family include GDDDR6 (graphics double data rate type six) SGRAM (synchronous graphics random-access memory), a PCI Express 3.0 x16 graphics bus, four DisplayPort 1.4 digital display interfaces and one VirtualLink VR connector. They support various digital audio modes, including HDMI digital audio. Two NVIDIA RTX GPUs can be paired with an NVIDIA NVLink high-speed GPU interconnect to scale up memory and performance.
The NVIDIA Quadro RTX6000 available for pre-order from PNY Technologies and its partners casts rays at a rate of 10 giga rays per second. It comes with 24GB of GDDDR6 memory, 576 Tensor Cores, 72 RT Cores and 4,608 NVIDIA CUDA parallel-processing cores. With an NVIDIA NVLink, coupled RTX 6000s can realize 48GB of memory.
The 16GB NVIDIA Quadro RTX5000 provides 8 giga rays per second computations, 3,072 CUDA cores, 384 Tensor Cores and 72 RT Cores. With NVLink, memory scales to 32GB.
On October 25, PNY Technologies will broadcast a live webinar exploring how NVIDIA's Turing architecture and the NVIDIA Quadro RTX 6000 GPU change computer graphics as well as how they accelerate a range of professional design and visualization workflows. The webinar will be available on demand as soon as practical at a later date, according to a PNY spokesperson.
See why DE's Editors selected the NVIDIA Quadro RTX GPUs from PNY Technologies as their Pick of the Week.
Sources: Press materials received from the company and additional information gleaned from the company's website.