Brand: NVIDIA | Category: GPUs
SKU: NVI-900-5G190-2270-000 | Part #: 900-5G190-2270-000 | MPN: 900-5G190-2270-000
Contact for Pricing — Request a Quote
Professional visualization and AI inference workloads benefit significantly from the Ada Lovelace architecture, and the NVIDIA RTX 4000 Ada Generation (part number 900-5G190-2270-000) is engineered to deliver enterprise-grade performance in compact, power-efficient form. This single-slot, low-profile GPU combines 6,144 CUDA cores with 192 fourth-generation Tensor cores and 48 third-generation RT cores, all backed by 20 GB of ECC-protected GDDR6 memory on a 160-bit bus. The card draws only 70 W thermal design power with no auxiliary power connector required, making it ideal for dense server deployments. NVIDIA equipped this adapter with dual NVENC and NVDEC engines supporting both AV1 and H.265 codecs, enabling high-throughput transcoding and real-time video processing alongside compute tasks.
Enterprise IT infrastructure teams, rendering studios, and AI inference platforms standardize on the NVIDIA RTX 4000 Ada because it bridges visualization-grade reliability with production-ready compute capability. The PCIe 4.0 x16 interface, four Mini DisplayPort 1.4a outputs, and support for up to 5K resolution at 5120 × 2880 make it suitable for multi-monitor CAD, simulation, and content creation workflows. Full support for DirectX 12 Ultimate, Vulkan 1.3, OpenGL 4.6, and OpenCL 3.0, combined with NVIDIA DLSS 3, NVIDIA RTX, and CUDA 12.x, ensures broad software ecosystem compatibility for both legacy and modern applications requiring deterministic, certified performance.
Organizations evaluating the NVIDIA 900-5G190-2270-000 for integration into virtualized, hybrid-cloud, or high-performance workstations should engage Omnixon Global to discuss volume pricing, lead times, and system integration support. Contact our team to submit a request for quotation and confirm availability for your enterprise deployment.
| Brand | NVIDIA |
| Category | GPUs |
| SKU | NVI-900-5G190-2270-000 |
| Part Number | 900-5G190-2270-000 |
| Condition | New |
| Capacity | 20GB |
| Interface | DDR6 |
| Manufacturer Part Number | 900-5G190-2270-000 |
| GPU Architecture | Ada Lovelace |
| CUDA Cores | 6,144 |
| Tensor Cores | 192 (4th Generation) |
| RT Cores | 48 (3rd Generation) |
| GPU Memory | 20 GB GDDR6 |
| Memory Bus Width | 160-bit |
| ECC Memory | Yes |
| TDP (Thermal Design Power) | 70 W |
| Auxiliary Power Connector | None required |
| Form Factor | Single-slot, low-profile compatible (4.4-inch height) |
| PCIe Interface | PCIe 4.0 x16 |
| Display Outputs | 4× Mini DisplayPort 1.4a |
| Maximum Simultaneous Displays | 4 |
| Maximum Display Resolution | 5K (5120 × 2880) |
| Video Encode / Decode Engines | Dual NVENC (AV1 + H.265), Dual NVDEC (AV1 + H.265) |
| NVLink / NVLink Bridge Support | No |
| Supported APIs | DirectX 12 Ultimate, Vulkan 1.3, OpenGL 4.6, OpenCL 3.0 |
| NVIDIA Software Features | DLSS 3, NVIDIA RTX, CUDA 12.x |
| ISV Certifications | Broad ISV certification across CAD, DCC, and medical imaging applications |
The NVIDIA 900-5G190-2270-000 RTX 4000 Ada Generation accelerates AI/ML training, inference, scientific HPC, and virtualization (vGPU) workloads. Typical deployments include LLM training clusters, computer-vision pipelines, financial risk modeling, and rendering farms.
Key specifications for the NVIDIA 900-5G190-2270-000 RTX 4000 Ada Generation: capacity 20GB; interface DDR6; new condition; model RTX 4000 Ada Generation; headline specification 20GB GDDR6; platform detail Single-slot PCIe workstation GPU; use case AI, graphics, simulation, analytics, edge, or network acceleration. Manufacturer part number 900-5G190-2270-000. For the full datasheet with electrical, environmental, and compliance details, contact our pre-sales engineering team.
The NVIDIA 900-5G190-2270-000 RTX 4000 Ada Generation requires a PCIe Gen4 or Gen5 x16 slot, server power adequate for the card's TDP, and CUDA/ROCm driver support in your hypervisor or bare-metal OS. Sales engineering will confirm chassis fit (1U/2U/4U), PCIe lane count, and PSU headroom before quoting.