The HPC-R2280-U4-G8 incorporates 8 of the next generation NVIDIA Tesla V100 SXM format compute cards enabling NVLINK across all 8 cards. Due to the incredible memory capacity and speeds achieved by combining each cards memory resources into a single pool via NVLINK. Deep learning and scientific simulations can now be scaled up to far greater complexity than ever before allowing those working with research and development projects to achieve more than previously thought possible.
End of Life
End of Life
| Item Depth (CM) |
88 |
|---|---|
| Item Height (CM) |
17.6 |
| Item Weight (Kg) |
58.1 |
| Item Width (CM) |
44.8 |
| Rack Units |
4U |
| # Front 1Gbps RJ45 Ports |
3 |
|---|---|
| # Front USB 3.2 Gen 1 Type-A |
2 |
| TPM Support |
1 |
|---|
| # Memory Slots |
24 |
|---|---|
| DDR Rev. |
DDR4 |
| Max. Memory Capacity (GB) |
3072 |
| Max. Memory Channels |
6 |
| Memory Type |
ECC LRDIMM, ECC RDIMM |
| PCIe Configuration |
SMX2: |
|---|
| # PSU |
4 |
|---|---|
| Efficiency Rating |
Platinum |
| Input Voltage (v) |
100-127V~/ 14A, 47-63Hz |
| Wattage |
2200 |
| # CPUs |
2 |
|---|---|
| Processor Chipset |
Intel C621 |
| Processor Generation |
Intel Xeon Scalable 2nd Gen. |
| Socket |
3647 / Socket P |
| Rear Remote Management LAN |
1 |
|---|
| Supported OS |
Citrix Hypervisor, Citrix Xenserver, Red Hat, SUSE, Ubuntu, VMware ESXi, Windows Server 2012 R2, Windows Server 2016, Windows Server 2019 |
|---|
| # Front 2.5 SATA/SAS Hotswap |
6 |
|---|---|
| # Front 2.5 U.2 NVMe Hotswap |
4 |
| Warranty Duration |
36 Months |
|---|---|
| Warranty Type |
Return To Base |
The HPC-R2280-U4-G8 incorporates 8 of the next generation NVIDIA Tesla V100 SXM format compute cards enabling NVLINK across all 8 cards. Due to the incredible memory capacity and speeds achieved by combining each cards memory resources into a single pool via NVLINK. Deep learning and scientific simulations can now be scaled up to far greater complexity than ever before allowing those working with research and development projects to achieve more than previously thought possible.
The HPC-R2280-U4-G8-SXM2 is the first server solution in the Workstation Specialist product range specifically designed from the ground up to optimise Deep Learning and Artificial Intelligence workflows. With this solution users will have the capabilities to perform training and inference operations of neural networks faster than ever before due to the 8x NVIDIA Tesla P100 / V100 GPUs delivering performance levels equal to 250 CPU only servers.
Due to the SXM2 format of the 8x NVIDIA Tesla P100/V100 cards this solution takes advantage of, it is able to use the NVIDIA NVLINK technology across all GPUs unlike standard PCIe based NVIDIA Tesla cards. NVLINK allows for a high speed interconnect of 300GB/s per GPU which is around a 10x speed up over conventional PCIe interconnections. Combined with new tensor core technology, purposely design to accelerate Deep Learning processes, users can expect to gain up to 4x speed up on neural network training.
Accelerate performance with NVIDIA’s latest Volta architecture technology. The NVIDIA Tesla V100 in this solution has 32GB of the fastest HBM2 graphical memory, 5120 CUDA cores and 640 dedicated Tensor Deep Learning cores making these the fastest platform available today for Deep Learning, Artificial Intelligence and Machine Learning operations. Furthermore, with a HPC-R2280-U4-G8-SXM2 server providing up to 1000 teraFlops of Deep Learning performance this system can provide as much performance as vast racks of CPU based systems, dramatically reducing costs, complexity of setup and maintenance / administration.
If this is still not enough performance to run your complex operations in the required times then clustering multiple of these systems together would further increase performance by an order of magnitude. Depending on the efficiency of your software and speed of your internal network, clustering of HPC-R2280-U4-G8-SXM2 systems can offer roughly linear performance gains.
The HPC-R2280-U4-G8-SXM2 thanks to it’s unique SXM2 foundations and industry leading NVIDIA Tesla and NVLINK technology is a real breakthrough in driving Neural Network based technologies to new levels. This cost-effective solution promises to speed up lengthy training and inference tasks helping to make the world of tomorrow a reality.
Reviews
There are no reviews yet.