NTS Elite APEX GIGABYTE G242-P36 Arm-Based 2U AI/HPC Server
The NTS Elite APEX GIGABYTE G242-P36 Arm-Based 2U AI/HPC Server is designed for cloud infrastructure, edge computing, accelerated computing, artificial intelligence (AI), and high-performance computing (HPC) workloads. Powered by a single Ampere® Altra® Max or Ampere® Altra® processor, this platform combines high core counts, PCIe Gen4 connectivity, and power-efficient Arm architecture in a compact 2U rackmount design.
The G242-P36 supports up to two NVIDIA® H100 PCIe Gen4 GPUs for accelerated AI training and compute-intensive workloads, along with support for up to two NVIDIA® BlueField®-2 DPUs for advanced data processing and infrastructure acceleration. Its eight-channel DDR4 memory architecture, flexible NVMe storage, PCIe expansion, and redundant power supplies make it suitable for cloud service providers, enterprise infrastructure, and research environments.
Built around Ampere's cloud-native processor architecture, the platform is designed to deliver predictable performance, scalability, and power efficiency. Hardware-level Root of Trust, remote server management, automatic fan speed control, and power-management technologies help support secure and reliable infrastructure operations.
New Tech Solutions (RackMountNTS) offers the GIGABYTE G242-P36 as an engineer-to-order server platform. Use Configure to select supported processors, memory, accelerators, storage, networking, and other system options. NTS reviews the proposed configuration and validates power, cooling, firmware, and drivers in Fremont, California before shipment.
Key Features and Benefits
Ampere Altra Max Arm Processor Architecture
Supports a single Ampere® Altra® Max or Ampere® Altra® processor.
Designed for cloud-native infrastructure, cloud service providers, and edge computing.
Ampere Altra Max processor family offers configurations with up to 128 cores.
High core density supports multiple concurrent workloads and scalable computing environments.
Designed to deliver predictable performance and performance-per-watt efficiency.
Supports PCIe Gen4 connectivity for high-throughput GPU, storage, and networking configurations.
NVIDIA GPU Acceleration
Supports up to two NVIDIA® H100 PCIe Gen4 GPUs.
Two full-height, full-length PCIe Gen4 x16 slots are provided for GPU installation.
Designed for GPU-accelerated AI training, inference, and HPC applications.
Supports accelerated computing environments that require parallel processing and high-throughput data access.
The GIGABYTE qualified component information also lists an A100 PCIe GPU card; confirm the exact supported accelerator and configuration with NTS before ordering.
NVIDIA BlueField-2 DPU Support
Supports up to two NVIDIA® BlueField®-2 DPUs.
Two rear low-profile PCIe Gen4 x16 slots are provided for DPU installation.
Designed to support infrastructure acceleration and data-processing workloads.
Helps provide additional flexibility for networking and data-center infrastructure configurations.
High-Bandwidth DDR4 Memory
16 DDR4 RDIMM/LRDIMM memory slots.
Eight-channel memory architecture per processor.
Supports memory speeds up to 3200 MHz.
Designed for memory-intensive cloud, edge, and enterprise applications.
Provides memory bandwidth for high-core-count Arm processor workloads.
Flexible NVMe Storage
Four front-accessible 2.5-inch PCIe Gen4 NVMe hot-swap drive bays.
Two internal M.2 slots with PCIe Gen4 x4 interfaces.
Tool-less drive bay design for convenient drive installation and replacement.
Supports enterprise-level SATA, SAS, SSD, and NVMe storage components through qualified configurations and optional controllers where applicable.
Optional PCIe RAID and HBA cards can be considered for supported storage configurations.
PCIe Gen4 Expansion and Networking
Two full-height, full-length PCIe Gen4 x16 slots for GPU accelerators.
Two rear low-profile PCIe Gen4 x16 slots for DPUs.
Two rear low-profile PCIe Gen4 x8 expansion slots.
Two 1Gb/s LAN ports powered by the Intel® I350-AM2 controller.
Optional Ethernet and InfiniBand networking cards are listed in the qualified component information.
Expansion flexibility for storage controllers, networking adapters, and infrastructure accelerators.
Hardware-Level Security
The GIGABYTE G242-P36 includes hardware-level Root of Trust (RoT) support to help protect the system during the boot process.
Hardware-level Root of Trust integrated into the platform.
Helps verify firmware integrity during system startup.
Designed to isolate and verify cryptographic operations during the boot process.
Helps protect against firmware compromise and unauthorized modifications.
Optional TPM 2.0 module support for hardware-based authentication.
TPM technology can securely store passwords, encryption keys, and digital certificates.
Power Efficiency and Intelligent Cooling
The platform incorporates GIGABYTE power-management and cooling technologies to help improve energy efficiency and system operation.
Automatic Fan Speed Control
Automatically adjusts individual fan speeds based on temperature sensor readings.
Helps balance system cooling requirements and power consumption.
Supports efficient thermal management under varying workloads.
Cold Redundancy
Designed for systems with N+1 redundant power supplies.
When total system load falls below 40%, one power supply can enter standby mode.
GIGABYTE states this feature can improve power efficiency by approximately 10% under the specified operating conditions.
Redundant Power Supply
High Availability and System Protection
Smart Ride Through (SmaRT)
Helps protect against unexpected AC power interruptions.
Temporarily adjusts system power consumption to maintain availability.
Power-supply capacitors provide approximately 10–20 ms of support during a transition to a backup power source.
Smart Crises Management and Protection (SCMP)
GIGABYTE technology designed to protect supported server configurations during PSU faults or overheating events.
Can force the CPU into an ultra-low-power operating mode to reduce system power demand.
Helps reduce the risk of unexpected shutdowns, component damage, and data loss.
Remote Server Management
The integrated Baseboard Management Controller (BMC) supports remote monitoring and administration through GIGABYTE server management applications.
GIGABYTE Management Console
Browser-based single-server management interface.
Server health monitoring.
Remote troubleshooting.
Resource utilization monitoring.
Automatic recording functions.
SAS and RAID controller monitoring.
Integration with third-party applications.
GIGABYTE Server Management (GSM)
Multiple-server management platform.
Supports IPMI and Redfish compatibility.
Helps administrators monitor and manage server infrastructure across deployments.
Supported GPU and Accelerator Configurations
The GIGABYTE G242-P36 supports GPU and DPU acceleration through dedicated PCIe Gen4 expansion slots.
NVIDIA GPU configuration
Up to 2 × NVIDIA® H100 PCIe Gen4 GPUs.
GIGABYTE's qualified component information also lists an NVIDIA A100 PCIe GPU card. Confirm the supported configuration and availability before selecting this option.
NVIDIA DPU configuration
Expansion architecture
2 × full-height, full-length PCIe Gen4 x16 GPU slots.
2 × rear low-profile PCIe Gen4 x16 DPU slots.
2 × rear low-profile PCIe Gen4 x8 slots.
The H100 and BlueField-2 configurations provide different types of acceleration: GPUs support parallel computing and AI workloads, while DPUs are designed for infrastructure and data-processing acceleration. Final accelerator combinations should be confirmed against platform compatibility and power requirements.
Key Applications and Use Cases
AI Training and Deep Learning
Supports GPU-accelerated AI training environments using up to two NVIDIA H100 PCIe GPUs. Suitable for deep learning development, neural network workloads, and computationally intensive AI applications.
AI Inference
Provides a platform for deploying trained AI models and running inference workloads that benefit from GPU acceleration and high-core-count Arm processors.
High-Performance Computing (HPC)
Designed for compute-intensive applications, parallel processing, research workloads, and scientific computing environments that require scalable processor and accelerator resources.
Cloud Computing and Cloud-Native Infrastructure
The Ampere Altra processor architecture is designed for cloud-native environments, offering high core counts, predictable performance, and power efficiency for cloud service providers and enterprise infrastructure.
Edge Computing
The compact 2U platform and power-efficient Arm architecture make it suitable for edge infrastructure deployments requiring compute density, accelerated processing, and flexible connectivity.
Data Center Infrastructure Acceleration
With optional NVIDIA BlueField-2 DPU support, the platform can be configured for infrastructure acceleration and data-processing workloads within supported data-center environments.
Virtualized and Containerized Workloads
The high-core-count Ampere processor architecture can support cloud-native, containerized, and consolidated infrastructure environments, subject to application and software compatibility.
Research and Scientific Computing
Supports research teams and technical environments running parallel computing, simulations, data analysis, and other compute-intensive workloads.
Software Ecosystem and Platform Compatibility
The GIGABYTE G242-P36 qualified software ecosystem information lists the following operating system versions:
SUSE: SLES 15 SP2, SLES 15 SP3.
Ubuntu: Ubuntu 18.04.5 LTS, Ubuntu 20.04.1 LTS.
Red Hat: RHEL 8.3, RHEL 8.5.
Fedora: Fedora Server 33-1.2, Fedora Server 35-1.2.
Debian: Debian 10.9 (Buster).
Oracle Linux: Oracle Linux R8 U2, Oracle Linux R8 U5.
CentOS: CentOS 8.3-2011, CentOS 8.4.2105, CentOS 8.5.2111.
Operating system and software compatibility should be confirmed for the intended processor, accelerator, driver, and application configuration before deployment.
NTS Engineer-to-Order Configuration and Validation
The NTS Elite APEX GIGABYTE G242-P36 Arm-Based 2U AI/HPC Server is quoted engineer-to-order by New Tech Solutions (RackMountNTS). Use Configure to select supported Ampere Altra or Altra Max processor options, memory, GPU accelerators, DPUs, storage, and networking. NTS reviews the proposed bill of materials (BOM) and validates power, cooling, firmware, and drivers in Fremont, California before systems ship.
Federal, defense, and SLED programs can align the same BOM to SEWP V, ITES-4H, GSA MAS, or applicable cooperative vehicles when the statement of work (SOW) allows. Commercial buyers receive the same configuration review, staging discipline, serial-tracked acceptance, and lifecycle support path after handoff.
Share workload requirements, target GPU or DPU counts, processor and memory requirements, Ethernet or InfiniBand fabric preferences, rack power budget, and delivery window through
contact NTS or Request a Quote. NTS can provide a reviewed configuration with applicable lead times, dual-source notes when required, and a package prepared for contracting review.
From individual systems to larger infrastructure deployments, NTS helps maintain configuration consistency across processors, accelerators, storage, and networking. This engineering approach supports deployment planning and expansion without restarting the configuration process for each system.
After you select available options in Configure, pricing updates in real time. Request a Quote to receive a configuration package with applicable lead times, dual-source notes when required, and acceptance documentation aligned with your receiving process. Configurable systems remain engineer-to-order, with the same Fremont staging and validation process.
Questions about GPU compatibility, BlueField-2 DPU configurations, Ethernet or InfiniBand networking, storage expansion, rack power requirements, or contract-vehicle packaging can be discussed with an NTS specialist using your draft BOM.
NTS supports commercial, federal, defense, and SLED buyers through the same engineering process: design review, component validation, assembly, firmware and driver preparation, and ship-ready documentation from the Fremont facility.