- Robust 4U 19” rack enclosure for GPU servers with maximum computing power
- Dual-socket Intel® Xeon® scalable processor
- 10x PCIe x16 Gen3 slots for high-performance GPUs and AI accelerators
- InfiniBand support for fast GPU cluster scaling
- Hot-swappable fan modules and redundant power supplies
FEATURES
Description
SHARK A.I. 10-slot GPU server – Maximum computing power for AI, deep learning and HPC
The SHARK A.I. 10-slot GPU server is a high-performance carrier-grade platform designed specifically for artificial intelligence (AI), deep learning and high-performance computing (HPC). With up to ten NVIDIA® GPUs, the system offers enormous computing power for data-intensive applications. By combining a powerful server node with PCIe expansion and PCIe switch technology, the server enables optimal communication between GPUs and CPUs. Thanks to InfiniBand support, multiple GPU clusters can be easily interconnected and scaled to perform even the most complex calculations efficiently.
Maximum GPU performance with flexible architecture
With ten PCIe slots for dual-slot GPUs, the SHARK A.I. delivers exceptional performance for neural networks, AI training and HPC applications. Support for single-root and dual-root complexes enables optimal adaptation to different workloads. Single-root configurations bundle all GPU resources for performance-intensive deep learning processes, while dual-root setups optimize CPU load and make parallel computations more efficient. The SHARK A.I. 10-slot GPU server uses dynamic PCIe lane allocation to optimally distribute computing power. GPUs can be assigned to specific I/O interfaces or CPU cores, which significantly improves the efficiency of virtualized applications and parallel computing. This architecture reduces the need for multiple physical systems and optimizes energy consumption, space requirements and cooling.
As a powerful Edge High Performance Computing (EHPC) solution, the SHARK A.I. significantly increases the flexibility of cloud and data center infrastructures. The platform allows the free allocation of GPUs per virtual machine (VM) and the implementation of various head nodes to efficiently manage large workloads. This makes the server ideal for AI-supported simulations, remote virtualization and data-intensive calculations.
Robust carrier-grade hardware for maximum reliability
The high-quality server architecture of the SHARK A.I. was developed for 24/7 operation in demanding environments. Hot-swappable fan modules and redundant power supplies ensure maximum availability and prevent downtime. GPU cards can be quickly installed or replaced by simply removing the top cover, enabling a maintenance-friendly and uninterrupted system configuration.
Specifications
| Product: SHARK A.I. 10-Slot GPU server | |
|---|---|
| SYSTEM | |
| CPU | Dual-socket Intel® Xeon® scalable processor |
| Chipset | Intel® C621 |
| GPU | Supports Single & Dual Root Complex Design 10x PCIe x16 Gen3 GPU slots |
| RAM | Up to 3TB DDR4 2666MHz (24x DIMM slots) |
| Storage | 14x 2.5″ SATA 6Gbit/s (hot-swap) |
| Expansion | 1x PCIe x16 Gen3 slot (low profile) 1x PCIe x8 Gen3 storage (mezzanine slot) |
| TPM | Optional |
| INTERFACE | |
| Ethernet | 2x 10-Gigabit Ethernet ports 1x Gigabit Ethernet ports (for IPMI) Intel® X550-AT2 controller |
| USB | 2x USB 3.0 (front) |
| COM | 1x COM (2×5-pin header) |
| Video | 1x VGA port |
| ENVIRONMENTAL | |
| Chassis | 4U 19″ rack enclosure |
| Cooling | 6x 12cm fans (hot-swap) |
| Power Supply | 4800W AC/DC Platinum power adapter; 3+1 redundancy Input: 4000W: 100~127V AC / 12A 4800W: 220 ~ 240V AC/ 9.48A |
| Operating Temperature | 10°C ~ 35°C |
| Storage Temperature | -40° ~ 70°C |
| Humidity | 90% @ 35° C non-condensing |
| Dimensions (W x H x D) | 438 x 176 x 770mm |
| Certifications | FCC (DoC), CE (DoC), CB/LVD, RCM, VCCI RoHS 6/6 conform |
