NVIDIA now works on Turing architecture to build its Quadro GPU technology, the company’s CEO Jensen Huang made an announcement at SIGGRAPH conference. Turing machine entirely works on artificial intelligence computations and ray tracing cores.
The combination of these two engines will surely lead to advanced improvements of a new level of hybrid rendering. The GPU’s introduced in the machine includes Quadro RTX 8000, Quadro RTX 5000 and Quadro RTX 5000. The entry-level will come at the flat price of $2300, the middle level at $6,300, while the high-end level at $10,000.
The company is primarily focusing on visual effects that offer wonderful new effects driven by neural networks, cinematic-quality interactive and fluid interactivity on highly complex models. Huang calls this technology “the Holy Grail of our industry”.
Following are some of the interesting specs of Quadro GPU Technology:
- The base of GPU, Quadro RTX 5000 features 16GB of GDDR6 memory, 32GB of memory with NVLink along with 6 Giga rays/second ray-tracing speed, 384 tensor cores and 3,072 CUDA cores.
- Secondary level Quadro RTX 6000 characterizes 24GB of GDDR6 memory, 24GB of memory with NVLink along with 10 Giga rays/second ray-tracing speed, 576 tensor cores and 4,608 CUDA cores.
- Top level Quadra RTX 8000 characterizes 48GB of GDDR6 memory, 96GB of memory with NVLink along with 10 Giga rays/second ray-tracing speed, 576 tensor cores and 4,608 CUDA cores.
- It supports 8K DisplayPort.
It contains computer hardware support from various tech giants including Lenovo, Dell, and HP. Nvidia’s Turing GPU architecture is the eighth-generation machine. This is the result of 10,000 engineering years of effort, the company says. The Turing die of the architecture measures 754mm^2 with around 18.6 billion of transistors.
It’s more powerful and efficient than Pascal’s 471mm^2 die and 11.8 billion transistors. Huang recently announced to exhibit “Material Definition Language Software development kit” that features ‘new software stack for computer graphics merging ray tracing, artificial intelligence, rastering, and computing.’
Apart from GPU, NVIDIA has also disclosed new RTX server consisting of eight Turing GPU’s that reduces the performance time from hours to minutes. The tech giant has not revealed any information regarding the availability of the device and how much it would cost. It might hit the market before the year ends.
The new RTX will support various graphics application including Adobe Dimensions CC, 3D imaging app, and Epic Games’ Unreal Engine. Till then, NVIDIA might make other changes in the Turing architecture and those will surely be more advanced and improved.