Giga Computing, a subsidiary of GIGABYTE and an industry leader in generative AI servers and advanced cooling technologies, today announced its first wave of GIGABYTE servers for Intel Xeon 6 Processors with P-cores. This new Intel Xeon platform is engineered to optimize per-core-performance for compute-intensive and AI intensive workloads, as well as general purpose applications. GIGABYTE servers for these workloads are built to achieve the best possible performance by fine tuning the server design to the chip design and to specific workloads. ⁠

All new GIGABYTE servers support Intel Xeon 6900-series processors with P-cores that have up to 128 cores and up to 96 PCIe Gen 5 lanes. Additionally, for greater performance in memory intensive workloads, the 6900-series expands to 12 channel memory, and makes available up to 64 lanes CXL 2.0. Overall, this modular SOC architecture has great potential with the ability to leverage a shared platform for running both performance and efficiency optimized architecture.⁠General Purpose (Single-Socket) Servers

• R164 Series: Compact 1U design that supports a single Intel Xeon 6 processor and a dual-slot GPU. Server variation within is mostly focused on the storage solution with choices of either 12x 2.5" bays or 4x 3.5"/2.5" bays for NVMe/SATA/SAS drives.
• R264 Series: Expanding to a 2U chassis, this series also supports a single processor, but with the ability to support up to 4x dual-slot Gen 5 GPUs.

General Purpose (Dual-Socket) Servers

• R184 Series: Similar storage choices are possible like the R164 series; however, with the R184 comes greater compute density spread across two CPU sockets, and to accommodate for this there is less of a focus on the expansion slots and removed dual-slot GPU support.
• R284 Series: Moving to a 2U chassis, the greater server height affords the ability to have both dual CPU sockets and support for dual-slot GPUs. The models within the R284 series include options of up to 24x 2.5" Gen 5 drives or 12x 3.5"/2.5" mixed storage.

Edge Servers

• E264 Series: For those in need of short chassis depth, this series retains the ability to have a 2U form factor in support of a single processor and up to four dual-slot GPUs. The server depth has decreased by nearly 200 mm compared to general-purpose servers, yet the overall compute performance can be maintained with a tradeoff in less storage bays.
• E284 Series: With an emphasis on CPU compute, this series is a dual CPU socket configuration with 6x FHHL expansion slots and 2x OCP NIC 3.0 slots for a wealth of I/O options. ⁠

NVIDIA MGX Modularized Servers

• XV24 Series: An NVIDIA OVX optimized server platform, this new dual socket series supports up to four NVIDIA L40S GPUs and has additional FHFL slots for NICs and DPUs. ⁠

High-density, Multi-node Servers

• H374 Series: For the ultimate in dense CPU compute performance, this series supports up to eight CPU sockets spread across four nodes. Each node also has four low-profile expansions slots and either two or six 2.5" bays. ⁠

Server-grade Motherboards

• MA34 Series: High-quality server board packed with I/O options, the MA34-CP0 is a board rich in both MCIO 8i and SlimSAS connectors, yet it has four Gen 5 expansion slots and an OCP NIC 3.0 slot. ⁠

GIGABYTE continues to push the boundaries of server innovation, delivering solutions that meet the evolving demands of modern computing. The new servers built on Intel Xeon 6900-series processors are designed to empower businesses across various sectors, providing unmatched performance, flexibility, and efficiency. As industries increasingly rely on AI, cloud computing, and edge technologies, GIGABYTE's commitment to aligning its server designs with the latest advancements in processor technology ensures that our customers are well-equipped to handle the challenges of tomorrow. As more is unpacked with the possibilities in the Intel Xeon 6 platform, GIGABYTE will ready systems to match all workloads as well as vertical markets. Additionally, in the coming months, GIGABYTE will announce a new GIGABYTE server that integrates Intel Gaudi 3 GPUs. This scalable and cost-effective solution will target AI workloads, including generative AI inference workloads. And customers can expect to see it at SC24 in Atlanta.⁠