JXNET RS5428 G5 AI Server

Leading Architecture

Upgraded to PCIe Gen 5.0 architecture supporting eight dual-width PCIe accelerator cards.

Dual-platform architecture flexibly adapts to different customer technology stacks.

New CXL 1.1 interconnect supports storage-class memory expansion and shared memory space.

Supports new 400G NDR IB networking and 400G OCP 3.0.

Exceptional Performance

Two Intel® 4th/5th Generation Xeon® Scalable processors or two AMD 4th Generation EPYC™ processors.

Built-in 8 TB system memory and 300 TB local high-speed storage for faster data access.

Supports four GPU topologies to flexibly match different AI compute scenarios.

PCIe topology architecture delivers up to 4x bandwidth improvement for high-throughput data transfer.

Flexible configuration.

Modular design supports flexible component-level configuration and free combination.

Powerful heterogeneous compute platform compatible with latest AI accelerator cards from all major brands.

25% improved cooling performance, 5% lower noise, and non-disruptive maintenance.

Supports N+N redundant 3000 W platinum/titanium power supplies for maximum system stability.

Outstanding ecosystem.

Rich global AI partners including AMD and Intel.

Leading deep learning frameworks: TensorFlow, PyTorch, PaddlePaddle, and more.

Powerful in-house AIStation resource platform, algorithm platform, and application optimization services.

Mature ecosystem with industry-wide, full-scenario algorithm and solution services.