Aligned, in partnership with AMD, delivers the groundbreaking power of the AMD Instinct™ MI355X accelerators, the new benchmark for AI and high-performance computing (HPC). Purpose-built to meet the explosive demands of modern AI and generative workloads, the MI355X delivers nearly 2x the FP64 and FP32 performance of competing GPUs like the NVIDIA B200.
Aligned’s optimized infrastructure brings this performance to scale with up to eight MI355X GPUs per module, giving enterprises the ability to process massive proprietary datasets, train large-scale models faster, and deploy advanced AI systems with unmatched efficiency.
Equipped with 288 GB of next-generation HBM3e memory and 8 TB/s of peak bandwidth, the MI355X eliminates data bottlenecks that slow traditional architectures.
This expanded memory pool 1.5x greater than NVIDIA’s B200, enables higher model fidelity, faster training, and smoother multi-GPU scaling across large clusters.
Supporting a full range of AI data formats, FP64, FP32, BF16, FP16, FP8, FP6, and FP4, the MI355X offers unprecedented flexibility across training and inference workloads.
With 5.0 PFLOPS of FP16/BF16 compute and advanced precision support down to FP4, it delivers the throughput required for next-generation generative models while maintaining energy efficiency and computational accuracy.
The MI355X’s advanced Matrix Cores deliver up to 3× the throughput for FP16 and BF16 operations, unlocking dramatic performance gains for both AI training and inference. This next-generation architecture is purpose-built to accelerate large-scale models, such as LLMs and generative AI systems, providing unparalleled speed, scalability, and efficiency.
