GenAI Studio
  • Edge AI SDK/GenAI Studio
  • Getting Started
    • About GenAI Studio
    • Quickstart
      • Prerequisite
      • Installation
      • Utilities
    • Feature Overview
      • Inference Chat
      • Fine-tuning
      • Model Management
      • Application
    • Version History
      • Version 1.1
      • Version 1.0
  • Inference
    • Chat Inference
    • AI Agents
  • Finetune
    • Text-to-Text
      • Overview
      • Full Parameter
      • LoRA
    • Text-to-Image (Coming Soon)
    • Dataset Management
    • Schedule
  • Model
    • Model Management
  • Validation
  • Convert
  • Administration
    • Resource Monitoring
  • System Configuration
    • AI Providers
      • LLM Setup
      • Embedder Setup
      • Vector DB
      • Transcription Setup
    • System Administration
      • Users
      • Workspace Chats
      • Invites
      • GPU Resource
      • Register an App
    • Appearance Customization
    • Tools
      • Embedded Chat Widgets
      • Event Logs
      • Security & Access
  • Application
    • Text to Image
    • Background Removal
    • OCR
  • FAQ
    • Technical
Powered by GitBook
On this page
  • Hardware Requirement
  • Advantech AIR-520
  • NVIDIA GPU Supported
  • SQ Flash AI SSD (M.2)
  • System Configuration Suggestions Based on Different Model Sizes (Full Parameters)
  • Hard Drive
  • System Requirement
  1. Getting Started
  2. Quickstart

Prerequisite

Hardware Requirement

Advantech AIR-520

4U Edge AI Server powered by AMD EPYC 7003 series processor, provides up-to 64 cores, 768GB DRAM, four PCle x16 expansion slots

  • Server grade performance powered by AMD EPYC 7003 processor

  • Up-to 768GB DDR4 3200 DRAM

  • Four PCIe x16 Expansion slots

  • Integrated 1200W PSU for total 700W graphic card TDP

NVIDIA GPU Supported

Product Name

Bus

Memory

H100

PCIe 4.0 x16

80GB, HBM2e, 5120bit

RTX A6000

PCIe 4.0 x16

48 GB, GDDR6, 384 bit

RTX A5000

PCIe 4.0 x16

24 GB, GDDR6, 384 bit

GeForce RTX 4090 (With size restrictions)

PCIe 4.0 x16

24 GB, GDDR6X, 384 bit

L40

PCIe 4.0 x16

48 GB, GDDR6, 384 bit

L40S

PCIe 4.0 x16

48 GB, GDDR6, 384 bit

RTX 6000 Ada Generation

PCIe 4.0 x16

48 GB, GDDR6, 384 bit

GeForce RTX 4090 D (With size restrictions)

PCIe 4.0 x16

24 GB, GDDR6X, 384 bit

RTX 4000 Ada Generation

PCIe 4.0 x16

20 GB, GDDR6, 160 bit

RTX 4000 SFF Ada Generation

PCIe 4.0 x16

20 GB, GDDR6, 160 bit

RTX 5000 Ada Generation

PCIe 4.0 x16

32 GB, GDDR6, 256 bit

SQ Flash AI SSD (M.2)

  • SQ Flash AI SSD (M.2), 1T

  • SQ Flash AI SSD (M.2), 2T

System Configuration Suggestions Based on Different Model Sizes (Full Parameters)

Model Size
AI100E (NVMe Gen4x4)
GPU (Reference)
System Memory

Up to 34B

1TB * 2

RTX 4000 ada *2

RTX 5000 ada *2

RTX 6000 ada *2

128GB

Up to 70B

2TB * 2

RTX 4000 ada *4

RTX 5000 ada *4

RTX 6000 ada *4

128GB

Hard Drive

  • 4TB or 8TB SSD for Installation and Model Training

System Requirement

  • Ubuntu 22.04:

    • Ubuntu versions 22.04.3 and 22.04.4 were tested successfully. Generally, any 22.04.X version, where X is 3 or higher, should work without issues. The Desktop edition is recommended

  • GPU Driver:

    • NVIDIA driver version 535 installation.

It is crucial to maintain a fixed GPU driver version. Avoid installing a newer version, as it may cause fine-tuning tasks to fail

  • Installer Size (.run):

    • 30G ~ 35G

  • Application Footprint:

    • 60G ~ 70G

The GenAI Studio installation file is approximately 30GB. To ensure smooth system installation, we recommend having at least 100GB of free disk space.

PreviousQuickstartNextInstallation

Last updated 6 days ago