Editor’s Pick: Next-generation Engineering GPU

PNY introduces the first graphics processing unit (GPU) using second-generation RTX technology, the NVIDIA RTX A5500. PNY says the second-generation gains typical of technology innovation make the RTX A5500 “previously aspirational but now obtainable.” 

The RTX A5500 is designed for professional graphics applications including mechanical design and engineering. The increases in raw processing speed, and in various components on the GPU, allows demanding professional users to run multiple tasks concurrently, tasks that previously had to run sequentially. 

Specific to manufacturing and product engineering, PNY says the NVIDIA RTX A5500 can now provide accelerated real-time design review with physically accurate materials, lighting and reflections. Such speed also allows real-time evaluation of the impact of design changes. 

The RTX A5500 features 10,240 CUDA Cores, the primary processors on NVIDIA GPUs. PNY says the NVIDIA Ampere GPU architecture for CUDA Cores runs up to 3x faster for 32-bit floating point calculations than the previous generation. 

RT Cores are specialized processors for ray tracing. The 80 cores in the RTX A5500 provide up to 2x the throughput of the previous generation, according to PNY. This speed increase allows the board to run concurrent compute tasks, such as ray tracing, simultaneously with shading or denoising. 

Tensor Cores accelerate matrix multiplication. They are important not only for raw graphics performance, but provide significant benefit to deep learning, artificial intelligence development and neural network training. New in the RTX A5500 is support for structural sparsity, which reduces network model sizes. PNY says this new feature offers up to 12x throughput compared to the previous generation. 

This board comes standard with 24GB of GDDR6 ECC RAM, which can be doubled to 48GB using NVLink. Input/Output with the computer motherboard uses PCI Express Gen 4, which doubled the bandwidth from Gen 3, from 15.75 GB/sec to 31.5 GB/sec. 

The RTX A5500 supports NVIDIA RTX Virtual Workstation software, which allows a personal workstation to be repurposed into multiple high-performance virtual workstation instances. PNY says this allows remote users to share resources to drive high-end design, AI, and compute workload. 

The RTX A5500 supports a wide variety of display options. There are 4 DisplayPort 1.4a connectors and support for multiple 8K monitors. NVIDIA Quadro Sync II support enhances NVIDIA Mosaic, along with NVIDIA Warp and Blend, both of which PNY says enable massive immersive environments for CAVE systems or other complex projection environments. 

The RTX A5500 is a two-slot device, allowing it to fit in most deskside workstation chassis. 

PNY says NVIDIA Omniverse use is increasing among engineering companies. The RTX A5500 supports this open and extensible platform for virtual collaboration.

The NVIDIA RTX A5500 is available now from PNY’s Authorized Resellers and System Integrators. A list of resellers is available here. A detailed spec sheet for the NVIDIA RTX A5500 is available on the PNY website. 

Sources: Press materials received from the company and additional information gleaned from the company’s website. 

More PNY Technologies Coverage

Building the GPU-Accelerated Data Center

Attend this webinar with Brian Corn, Vice President of Product Integration at Thinkmate, and learn more about GPU-accelerated technologies.