Giga Computing launches two AmpereOne M-based servers, bringing ARM architecture for servers to a broader audience.
Giga Computing expands its server portfolio with the R1A3-T40 and R2A3-T40. These are two new models based on AmpereOne M processors. The servers are designed for cloud-native applications and AI inference workloads.
With these servers, Giga Computing expands the range of powerful ARM-based hardware. Ampere chips can be found in Microsoft’s ARM offerings, among others. The chips belong to the same category as Graviton at AWS. Giga Computing now brings these processors to colocation and on-premises environments as well.
1u and 2u
The R1A3-T40 is a 1U server with one socket. It offers four hot-swappable NVMe drives (3.5″/2.5″) via Gen5 connections. The system has two FHHL PCIe Gen5 x16 slots and is powered by two redundant 1,600-watt power supplies.
The R2A3-T40 is a 2U server, also with one socket, but with more storage capacity. This model features twelve 2.5″ Gen5 NVMe hot-swap bays, one FHHL PCIe Gen5 x16 slot, and also two redundant 1,600-watt power supplies.
Both servers are designed for efficient heat dissipation and system stability, with a focus on reducing the number of nodes required per data center environment. This should help lower total cost of ownership.
AmpereOne M
The AmpereOne M processors in these servers feature up to 192 ARM-based cores. Each core is single-threaded and equipped with dual 128-bit vector units. Furthermore, the processor configuration supports up to twelve DDR5 memory slots with one DIMM per channel, accommodating up to 1.5 TB of system memory. Large L2 caches ensure smooth processing of diverse workloads.
According to Giga Computing, the servers are suitable for applications with high memory requirements, such as AI inference and cloud-native services. By utilizing Ampere’s architecture, the company can offer an energy-efficient solution with stable performance.