OEM
MI355X
MI
AMD Instinct The Best High-Density Server for AI Inference
Powered by the latest AMD Instinct™ MI325X, MI350X, and MI355X accelerators, this solution is ideal for advanced AI training and large-scale model inference. It ensures that GPUs meet the demanding data requirements of intensive AI workloads, delivering unmatched speed and efficiency for cutting-edge applications.
Specifications
AMD INSTICT MI325X/MI350X/MI355X | |
GPU | 8 OAM MI325X / MI350X / MI355X |
GPU Memory | 8 256GB / 288GB / 288GB |
Performance | FP8 MI325X: 2.614 (PFLOPS) / 5.229 (PFLOPS) w/sparsity MI350X: 4.614 (PFLOPS) / 9.227 (PFLOPS) w/sparsity MI355X: 5.033 (PFLOPS) / 10.066 (PFLOPS) w/sparsity |
GPU Compute Units | 304 / 256 / 256 |
Matrix cores | 1024 |
Stream Processors | 19,456 / 16,384 / 16,384 |
Memory Bandwidth | 6 TB/s / 8 TB/s / 8 TB/s |
AMD Infinity Fabric Link | 7x 128GB/s , 7x 153.6GB/s , 7x 153.6GB/s |
Power Usage | ~ 18 KW |
CPU (Intel/AMD) | Dual EPYC 9004/9005 |
System Memory | 24 DIMM slots Max Memory (1DPC): Up to 6TB 6400MT/s ECC DDR5 RDIMM (AMD EPYC™ 9005 Series Processor) Max Memory (1DPC): Up to 6TB 4800MT/s ECC DDR5 RDIMM (AMD EPYC™ 9004 Series Processor) |
Interface | 2 RJ45 1 GbE 1~4 USB 3.0 ports 1 VGA port 1 TPM header |
Network | Optional |
Storage | 8~10 x 2.5" Gen5 NVMe/SATA |
Power | 12 3000W (6+6) |
Rack Units (RU) | 7/8 RU |
Brand | SMCI, Asus, Gigabyte, Dell |