Lenovo ThinkSystem SR670 V3 with NVIDIA L40S 48GB PCIe GPU
Brand: Lenovo | Category: GPUs
SKU: 4X97A90246 | Part #: 4X97A90246 | MPN: 4X97A90246
Contact for Pricing — Request a Quote
About the Lenovo ThinkSystem SR670 V3 with NVIDIA L40S 48GB PCIe GPU
The Lenovo ThinkSystem SR670 V3 with NVIDIA L40S 48GB PCIe GPU is a 2U rack-mounted server built for organizations deploying GPU-accelerated inference, visualization, and AI workloads at enterprise scale. This configuration pairs Lenovo's robust SR670 V3 platform with the NVIDIA L40S GPU, which features the Ada Lovelace architecture, 48GB of GDDR6 ECC memory, and 18,176 CUDA cores optimized for demanding computational tasks. The NVIDIA L40S delivers 91.6 TFLOPS of FP32 performance, 733 TOPS of TF32 tensor performance with sparsity, and 2,936 TOPS of FP8 tensor performance with sparsity, all supported by 864 GB/s of memory bandwidth over PCIe Gen 4 x16. The 350W GPU thermal design power is efficiently managed within the 2U form factor, making this system ideal for dense deployments in space-constrained data centers.
The underlying Lenovo platform supports dual 4th Gen Intel Xeon Scalable processors (Sapphire Rapids), DDR5 system memory, and Lenovo XClarity Administrator management software for streamlined operations. The SR670 V3 can accommodate up to 8 single-width or 4 double-width GPUs per server, enabling flexible scaling for AI infrastructure teams and enterprises building high-performance computing clusters. Part number 4X97A90246 represents this preconfigured solution, delivering the performance and reliability expected from Lenovo's enterprise-grade ThinkSystem portfolio.
Typical Enterprise Deployment Scenarios
- GPU-accelerated machine learning inference and model serving at scale
- Professional visualization and rendering workloads for design and media studios
- High-performance computing research requiring dense GPU-to-CPU ratios
- Real-time analytics and data processing pipelines in financial services
- Large language model inference and generative AI inference operations
For detailed specifications, availability, and procurement options, contact the Omnixon Global team to request a quotation on part number 4X97A90246.
Technical Specifications
| Brand | Lenovo |
| Category | GPUs |
| SKU | 4X97A90246 |
| Part Number | 4X97A90246 |
| Condition | New |
| Manufacturer Part Number | 4X97A90246 |
| Product Line | ThinkSystem SR670 V3 |
| GPU Model | NVIDIA L40S |
| GPU Architecture | NVIDIA Ada Lovelace |
| GPU Memory | 48GB GDDR6 ECC |
| GPU Memory Bandwidth | 864 GB/s |
| GPU Interface | PCIe Gen 4 x16 |
| GPU TDP | 350W |
| CUDA Cores | 18176 |
| Tensor Cores | 568 (3rd Generation) |
| RT Cores | 142 (3rd Generation) |
| FP32 Performance | 91.6 TFLOPS |
| TF32 Tensor Performance (with sparsity) | 733 TOPS |
| FP8 Tensor Performance (with sparsity) | 2936 TOPS |
| Server Form Factor | 2U Rack |
| Processor Support | Dual 4th Gen Intel Xeon Scalable (Sapphire Rapids) |
| Maximum GPU Configuration | Up to 8 single-width or 4 double-width GPUs per server |
| System Memory Type | DDR5 |
| Management Software | Lenovo XClarity Administrator |
| Operating System Support | Windows Server, Red Hat Enterprise Linux, SUSE Linux Enterprise Server, VMware vSphere |
Frequently Asked Questions about Lenovo ThinkSystem SR670 V3 with NVIDIA L40S 48GB PCIe GPU
What server platforms accept the Lenovo ThinkSystem SR670 V3 with NVIDIA L40S 48GB PCIe GPU?
Reference servers include Dell PowerEdge XE9680 / XE9712, HPE Cray XD670, Lenovo ThinkSystem SR685a / SR675 V3, Supermicro AS-A21GE / SYS-821GE, Gigabyte G593 / G894, ASUS ESC. Share your target platform in the RFQ and we will confirm chassis-to-GPU compatibility and recommended NIC pairing.
How long is the lead time on AI GPUs?
Highly model-dependent. L40S / RTX-class: typically 3-6 weeks. H100/H200/B200 in SXM form factor: 12-16 weeks for whole-platform allocations. We quote genuine-channel ETAs only — no grey-market promises.
Do you supply matched networking (Quantum InfiniBand / Spectrum-X)?
Yes — Omnixon stocks the full NVIDIA networking lineup (Quantum-2 / Quantum-X InfiniBand, Spectrum-X Ethernet, ConnectX NICs, BlueField DPUs) so we can quote a complete training-cluster BOM, not just the GPUs.
Can you help with NVIDIA AI Enterprise licensing?
Yes. We hold genuine-channels for NVIDIA AI Enterprise software subscriptions. Add it to your RFQ and we quote node-aligned licensing along with the hardware.