inferenceserver (computer) Wave NF5468 M6 Wave Server GPUsServer AI Server Training Server AI Learning
Wave NF5468 M6 is a resilient AI server that Wave sells well at home and abroad.NF5468The NF5468M6 is the latest flagship product of the NF5468 series, which supports a variety of Al acceleration cards and domestic AI chips, and is widely used in Internet AI public cloud, enterprise AI cloud platform, intelligent security, video encoding and decoding and other scenarios.stockpileIt is the industry's only single machine that supports one-click switching of Balance/Common/Cascade topology software, which can flexibly respond to the performance tuning needs of different AI application scenarios.
Functional characteristics
Leadership Architecture
- Supports up to 20 PCIe form factor external cards in 4U of space
- Supports NVIDIA A40 graphics cards and A100 graphics cards with MIG technology
- Support for AMD Instinct MI100 with new Matrix Core technology
- Supports the new 200G HDR IBreticulationCard and 100Gb OCP 3.0
superior performance
- Two 3rd Generation Intelligent Intel® Xeon® Scalable Processors, 3 UPI Interconnects, TDP 270W
- 4-card pass-through design eliminates the need for a PCIe Switch for CPU-GPU communication
- 8 hot-swappable NVMe SSDs for superb data throughput performance
- Up to eight 350W GPU cards on PCIe Gen4 platforms with double the bandwidth
Flexible Configuration
- Flexibly adapts to the new generation of AI acceleration cards and smart network cards of various brands
- Supports one-click topology switching to meet differentAI applicationsdemand (economics)
- Three GPU configurations to flexibly match customers' different application scenarios
Ecological excellence
- Abundant global AI acceleration card partners, NVIDIA, AMD, Intel, Cambium, Fusion, etc.
- Leading deep learning frameworks, TensorFlow, PyTorch, PaddlePaddle, MXNet, etc.
- Powerful Wave AIStation Resource Platform, Algorithm Platform and Application Optimization Services
- Mature meta-brain ecosystem, providing industry-wide application solution support for all scenarios
technical specification
| Product Model | NF5468M6-P | NF5468M6-T | NF5468M6-V |
| high degree | 4U | ||
| CPU | Supports 2 new generation Intel® Xeon® Icelake Scalable Processors, TDP 270W | ||
| GPU card | Supports 8 full-height, full-length, double-width PCIe interface GPU cards such as A100, A30, A40, MI100, etc., while the rear supports up to 4 PCIe4.0 x16 slots | Supports 4 full-height, full-length and double-width PCIe interface GPU cards such as A100, A30, A40, MI100, etc., while the rear supports 2 PCIe3.0 x8 and 2 PCIe4.0 x8 slots | Supports 16 full-height, full-length, single-width GPU cards such as A10, while the rear supports up to 4 PCIe4.0 x16 slots |
| chipsets | Intel® C621A series chipset ((LBG-R)) | ||
| random access memory (RAM) | Supports 32 DDR4 3200MHz LRDIMM/RDIMMs | ||
| Built-in PCIE card | Supports up to 2 built-in standard RAID cards | ||
| Front I/O | 2 USB 3.0, 1 VGA port, 1 RJ45 serial port | ||
| Rear I/O | 1 serial port, 2 USB 3.0 ports, 1 VGA port, 1 RJ45 management port, 1 OCP3.0 network card (supports NCSI function) | ||
| local storage | 24 x 2.5 or 12 x 3.5" SAS/SATA drives (supports up to 8 NVME SSDs), 2 x M.2 SATA SSDs | 16 x 2.5 or 12 x 3.5" SAS/SATA drives (supports up to 2 NVME SSDs), 2 x M.2 SATA SSDs | 24 x 2.5 or 12 x 3.5" SAS/SATA drives (supports up to 8 NVME SSDs), 2 x M.2 SATA SSDs |
| RAID Support | Optional support for RAID0, 1, 10, 5, 50, 6, 60, etc., support for Cache supercapacitor protection, provide RAID state migration, RAID configuration memory | ||
| operating system | Microsoft Windows Sever, Red Hat Enterprise Linux, Ubuntu Linux, CentOS and other major operating systems | ||
| System cooling | N+1 redundant system cooling fans | ||
| power supply | 4 x 1600W/2000W/2200W/3000W 80Plus Platinum PSUs supporting 2+2 redundancy mode | ||
| case (computer) (lit. machine box) | Width 483mm, height 175.5mm, depth 830mm | ||
| operating temperature | 5 - 35°C / 41°F - 95°F | ||
| full complementary weight | ≤85kg |
