T-Chip Group
Intelligent hardware website
Innovate intelligent hardware and solutions.
Server Business Website
Server products and efficient solutions.
Consumer Business Website
StationPC NAS products and services.

RS-F1SG2042 RISC-V Server

Open Source Instruction Set

· RISC-V Processor SG2042 

· Deployment of Large AI Models 

· Multiple Deep Learning Frameworks 

· Features 12 Drive Bays for Large Storage Capacity 

· Secure ＆ High-Speed Network Communication 

· Supports privacy computing technologies 

· Standard 2U rack server size 

· Supports Multiple Operating Systems


Manual Wiki Download

Gemma

Qwen

Llama

Phi

RS-F1SG2042
RS-F1SG2042 RISC-V Server

Powered by a single high-performance RISC-V processor SG2042, equipped with 64 64-bit cores, delivering powerful performance and supporting mixed-precision computing. It supports private deployment of mainstream AI large models and a variety of deep learning frameworks. It features 12 drive bays for 3.5-inch / 2.5-inch SAS / SATA hard drives and 2 5G Ethernet ports. It comes with an intelligent management system and supports mainstream privacy computing technologies such as homomorphic encryption and differential privacy. The design adopts a standard 2U rackmount server chassis.

RISC-V Processor SG2042

Powered by a single high-performance RISC-V processor SG2042, equipped with 64 64-bit cores, delivering powerful performance. It supports mixed-precision computing with INT8, FP16, BF16, and FP32, enabling efficient adaptation for scenarios such as scientific computing and AI inference.

Private Deployment

Supports private deployment of mainstream large language models (such as Gemma, Llama, Qwen, etc.), vision large models (such as EfficientVIT, SAM, TAM, etc.), and image generation models (such as Flux, Stable Diffusion, etc.).

Deep Learning Frameworks

Supports traditional network architectures such as CNN, RNN, and LSTM, as well as a variety of deep learning frameworks including TensorFlow, PyTorch, PaddlePaddle, ONNX, etc. It also supports custom operator development and Docker containerization management technology.

Ultra-Large Storage Capacity

Featuring 12 drive bays for 3.5-inch / 2.5-inch SAS / SATA hard drives, supporting SATA3.0 HDD / SSD expansion, allowing the device to easily scale up to TB-level ultra-large capacity, and supports hot-swapping for quick replacement.

Secure Network Communication

Equipped with dual 5G Ethernet ports, providing high-speed and stable network communication to meet the needs of different application scenarios. An independent MGMT management network interface separates the management network from the business network, ensuring the security and reliability of network communication.

Privacy Computing Technologies

Supports mainstream privacy computing technologies such as homomorphic encryption, differential privacy, and secure multi-party computation. Throughout the lifecycle of data usage and analysis, it achieves "data usable but invisible," providing comprehensive protection for data security and user privacy.

Standard 2U Rack Server Size

Features a standard 2U rackmount server chassis design for flexible deployment, fully compatible with most types of data center racks. It is equipped with four high-efficiency cooling fans and two 550W hot-swappable power supplies for easy maintenance.

Multiple Operating Systems

Supports operating systems such as Fedora, Ubuntu, openKylin, openEuler, and deepin, meeting the needs of various scenarios.
Open Source Instruction Set

Based on the international open-source RISC-V instruction set, there is no authorization restriction, and it supports mainstream operating systems, creating a truly open computing platform and making innovation ubiquitous.

Application Scenarios

Widely applicable in industry fields such as private deployment of large models, deep learning, cloud storage, ecosystem development, telecommunications, and privacy computing.

Private Deployment of Large Models
Private Deployment of Large Models
Deep Learning
Deep Learning
Cloud Storage
Cloud Storage
Ecosystem Development
Ecosystem Development
Telecommunications
Telecommunications
Privacy Computing
Privacy Computing
Interfaces
Specifications

RS-F1SG2042

Technical Specifications

Server form factor

2U rackmount computing server

Architecture

RISC-V architecture

Processor

1 × SG2042 64-core RISC-V processor @2.0GHz

Computing precision

INT8/FP16/BF16/FP32

Memory

4 × DDR4 DIMM slots, supporting a maximum frequency of 3200MHz, maximum single module capacity of 64GB, and a maximum total memory capacity of 64×4=256GB

Expanded storage

12 × 3.5"/2.5" SATA3.0 SSD drive bays (Hot-swappable; BMC enables direct hard drive operation, and computing nodes can access hard drives indirectly via the network sharing provided by BMC)

PCIe expansion slots

4 × GEN4 X16 PCIe Slots, 4 × GEN4 X8 PCIe Slots

Power supply

2 × 550W Hot-swappable power supplies, supporting 1+1 redundancy (Input: 100V AC ~ 240V AC)

Fan module

4 × Hot-swappable fans, supporting N+1 redundancy

Physical Specifications

Size

481mm(L) × 677mm(W) × 88mm(H)

Weight

Net server weight: 15kg, Gross weight with packaging: 17.2kg

Environment

Operating Temperature: 0ºC ~ 35ºC, Storage Temperature: -40ºC ~ 60ºC, Operating Humidity: 5% ~ 80%RH (Non-condensing)

Software Specifications

BMC

Integrated BMC management system with web-based UI, supporting redfish, VNC, NTP, advanced monitoring and virtual media; the BMC management system supports secondary development

Large language models

Supports on-premises deployment of ultra-large parameter models based on Transformer architecture, such as Deepseek- R1 Series, Gemma Series, Llama Series, ChatGLM Series, Qwen Series and other LLMs.

Vision large models

Supports on-premises deployment of vision large models such as ViT, Grounding DINO, SAM.

AI Painting

Supports on-premises deployment of image generation models including Flux, Stable Diffusion, Stable Diffusion XL.

Deep learning

Supports traditional network architectures such as CNN, RNN, LSTM; compatible with multiple deep learning frameworks including TensorFlow, PyTorch, PaddlePaddle, ONNX, Caffe; supports custom operator development and Docker containerization management technology.

Interface Specifications

Ethernet

2 × 5Gbps Ethernet (RJ45), 1 × Gigabit Ethernet (RJ45, MGMT for BMC management network)

Display

2 × VGA (Maximum resolution 1080P, for BMC management display)

USB

4 × USB3.0, 1 × Type-C (BMC debug serial port, built-in UART chip, baud rate 115200)

Button

1 × Reset, 1 × UID, 1 × Power
Customization

Firefly team, with over 20 years of experience in product design, research and development, and production, provides you with services such as hardware, software, complete machine customization, and OEM server.

Hardware Solution

Software Solution

Turnkey Solution