NVIDIA A16 - dense graphics VDI on a single board
The A16 is built around a solution unconventional for the data center: four independent Ampere GPUs on one board, 16 GB of GDDR6 with ECC each - 64 GB in total. This architecture allows up to 64 concurrent vPC users per card and mixing different vGPU profiles and workload types without allocating separate hardware for each scenario.
The card is designed for virtual desktop infrastructure: knowledge workers get native-PC-level responsiveness remotely, while entry-level virtual workstations for CAD/CAM run through RTX vWS with sufficient framebuffer per user. The full NVIDIA virtualization stack is supported - vPC, vApps, RTX vWS, AI Enterprise and vCS.
- 4 Ampere GPUs, 5120 CUDA cores (4x 1280), 160 3rd-gen Tensor cores, 40 2nd-gen RT cores
- 64 GB GDDR6 with ECC (4x 16 GB), 128-bit bus per GPU
- 4x NVENC + 8x NVDEC: H.265, VP9, AV1 decode for dense video transcoding
- PCI Express 4.0 x16, passive cooling, 8-pin EPS power, up to 250 W
- FHFL form factor, dual slot, no physical video outputs - output via vGPU
Thanks to four encoders and eight decoders, the A16 sustains a high density of video conferencing and streaming media, while idle VDI resources can be given over to AI inference through NVIDIA AI Enterprise. This is the retail version in PNY branded packaging with full warranty. ETE.UA is an official PNY/NVIDIA supplier in Ukraine; selection for your hypervisor and user profiles - phone +380 93,594 00 77.
