Graphics Cards
Enterprise GPU Accelerators for Dell PowerEdge and HPE ProLiant
Enterasource stocks enterprise GPU accelerators and graphics cards for server workloads: NVIDIA Tesla/A-series data center GPUs, AMD Instinct accelerators, and workstation-class GPUs for visualization deployments. Server GPU accelerators are purpose-built for data center use — passive cooling, no display outputs, full-height PCIe form factor, and ECC memory — designed for continuous operation in rack environments.
GPUs in servers serve three distinct roles: AI/ML inference and training (NVIDIA A100, A30, A10), parallel computing and simulation (NVIDIA Tesla T4, V100), and remote workstation visualization (NVIDIA RTX series with GRID vGPU licensing). The correct GPU depends on your specific workload type and the PCIe power and slot availability in your server chassis.
Contact us for current GPU inventory — availability changes frequently based on sourcing.
Filters
-
NVIDIA Tesla K40 12GB GDDR5 PCI-E GPU Accelerator Card | 699-22081-0202-200Availability: Available by QuoteManufacturer: NVIDIA Condition: R660 1 Year WarrantyRequest PricingOut of stock - NVIDIA Tesla K80 24GB PCI-E GPU Accelerator Graphics Card | 699-22080-0200-500Availability: Available by QuoteManufacturer: NVIDIA Condition: R660 1 Year WarrantyRequest PricingOut of stock
- NVIDIA Quadro FX 4800 1.5GB GDDR3 PCI-E Graphics Card | 900-50607-2700-000Availability: Available by QuoteManufacturer: NVIDIA Condition: R660 1 Year WarrantyRequest PricingOut of stock
- NVIDIA Quadro K420 1GB GDDR3 Graphics Card | 699-52012-0505-400Availability: Available by QuoteManufacturer: NVIDIA Condition: R660 1 Year WarrantyRequest PricingOut of stock
- NVIDIA Quadro K620 2GB GDDR3 Graphics Card | 699-52012-0504-710Availability: Available by QuoteManufacturer: NVIDIA Condition: R660 1 Year WarrantyRequest PricingOut of stock
- Radian 800001-001 RMS-200/8G 8GB PCI-e NVRAM Accelerator Edge CardAvailability: In StockManufacturer: Radian Condition: R660 1 Year WarrantyBuy Now$40.99
- NVIDIA 699-2G402-0060-310 Tesla M60 16GB GDDR5 PCI-E GPU AcceleratorAvailability: Available by QuoteManufacturer: NVIDIA Condition: R660 1 Year WarrantyRequest PricingOut of stock
- NVIDIA 699-52401-0502-221 GRID K1 16GB PCI-E GPU Graphics CardAvailability: Available by QuoteManufacturer: NVIDIA Condition: R660 1 Year WarrantyRequest PricingOut of stock
- NVIDIA 699-2G402-0060-300 Tesla M60 16GB GDDR5 PCI-E GPU AcceleratorAvailability: Available by QuoteManufacturer: NVIDIA Condition: R660 1 Year WarrantyRequest PricingOut of stock
- NVIDIA 699-2G610-0200-100 Tesla P40 24GB GDDR5 PCI-E GPU AcceleratorAvailability: Available by QuoteManufacturer: NVIDIA Condition: R660 1 Year WarrantyRequest PricingOut of stock
- NVIDIA Tesla 699-2G402-0060-321 M60 16GB GDDR5 PCI-E GPU AcceleratorAvailability: Available by QuoteManufacturer: NVIDIA Condition: R660 1 Year WarrantyRequest PricingOut of stock
- Nvidia 699-52012-0504-700 Quadro K620 2GB PCI-E GPU Graphics CardAvailability: Available by QuoteManufacturer: NVIDIA Condition: R660 1 Year WarrantyRequest PricingOut of stock
Server GPU Selection Guide
| GPU | Use Case | VRAM | Form Factor |
|---|---|---|---|
| NVIDIA A100 80GB | AI/ML training, HPC | 80GB HBM2e | SXM4 or PCIe (double-wide) |
| NVIDIA A30 | AI inference, data analytics | 24GB HBM2 | PCIe (double-wide) |
| NVIDIA A10 | AI inference, vGPU, visualization | 24GB GDDR6 | PCIe (single-slot) |
| NVIDIA T4 | Inference, VDI, video | 16GB GDDR6 | PCIe (single-slot, low-profile) |
| NVIDIA V100 32GB | HPC, deep learning training | 32GB HBM2 | PCIe or SXM2 |
Server compatibility varies by chassis. The Dell R750xa and R760xa are purpose-built for up to 4 GPU accelerators in 2U. The standard R740, R750, and R760 support 1-2 double-wide GPUs depending on the PCIe riser configuration. Confirm your server’s PCIe riser and power budget before ordering.
- Do server GPU accelerators require display outputs?
- No. Data center GPU accelerators (NVIDIA A100, A30, A10, T4, V100) have no display outputs — they are compute-only devices accessed via CUDA or ROCm APIs. They are not suitable for desktop graphics. NVIDIA RTX-series cards with GRID licensing provide virtual display outputs for VDI deployments.
