Get Started
Requirements
Before starting, please make sure that you have already met the following prerequisites.
Get ComIDP License Key
ComIDP provides two types of license key: 30-day free trial license, and formal license.
How to Get Free Trial License
Contact our sales team and we'll send you a 30-day free trial license for ComIDP.
How to Get Formal License
ComIDP is a commercial product that requires a license for application release. Any documents, sample code, or source code distribution from the released package of ComIDP to any third party is prohibited.
Install ComIDP Package
Contact us to obtain the ComIDP package.
System Requirements
Supported Operating Systems
ComIDP runs on a variety of platforms. The following operating systems are supported:
- Ubuntu, Fedora, Debian, or CentOS. Ubuntu and Debian derivatives, such as Kubuntu or Xubuntu, are also supported. Currently only 64-bit Intel (x86_64) processors are supported.
Regardless of which operating system you use, you will need at least 85GB of memory.
Recommended Configuration for Functional Combination
| System Combination | CPU Recommendation | GPU Recommendation | Memory Recommendation | Storage Recommendation |
|---|---|---|---|---|
| Knowledge Base (KB) | ≥ 4 cores (Intel i5 / AMD Ryzen 5) | ≥ 12GB video memory (such as RTX 3060 / A2000) can be adjusted according to the question-answering model | ≥ 16 GB | ≥ 100 GB SSD (SATA or NVMe) |
| Document Extraction (Extract) | ≥ 8 cores (Intel i7 / AMD Ryzen 7) | RTX 3090 / 4080 (≥16GB video memory) | ≥ 32 GB | ≥ 500 GB NVMe SSD |
| Document Parsing (Parse) | ≥ 16 cores (Xeon Gold 63xx / EPYC 7402) | No independent GPU required | ≥ 64 GB | ≥ 1 TB NVMe SSD (read speed > 2GB/s) |
| KB + Extract | ≥ 8 cores (Intel i7 / Ryzen 7) | RTX 3090 / 4080 (≥16GB VRAM) | ≥ 48 GB | ≥ 500 GB NVMe SSD |
| KB + Parse | ≥ 8 cores (Intel i7 / Ryzen 7) | ≥ 12GB VRAM (such as RTX 3060 / A2000) can be adjusted according to the question-answering model | ≥ 96 GB | ≥ 500 GB NVMe SSD |
| Extract + Parse | ≥ 16 cores (EPYC 7402 / Xeon Gold 63xx) | RTX 4090 / A100 (≥24GB VRAM) | ≥ 96 GB | ≥ 1 TB NVMe SSD (read speed >2GB/s) |
| KB + Extract + Parse | ≥ 16 cores (EPYC / Xeon Gold multi-core concurrent support) | RTX 4090 / A100 / H100 (≥24GB VRAM) | ≥ 128 GB | ≥ 1 TB high-speed NVMe SSD (read speed >2GB/s) |
Install Docker
ComIDP is distributed as a Docker container. To run it on your computer, you need to install the Docker runtime environment suitable for your operating system.
Please follow the instructions on the Docker official website to install and start Docker Engine.
After installing Docker, you can install Docker Compose by following the installation instructions.
Install NVIDIA Container Toolkit
Install NVIDIA Drivers
Visit the official NVIDIA Drivers page to download and install the correct driver.
Verify that your GPU is running and accessible.
The following example is Ubuntu 22.04 - NVIDIA T4
# Make sure the server has gcc and make environments installed. You can check the gcc version with the following command
gcc -v
# You can check the make version with the following command
make -v
# Download the installation package
wget https://us.download.nvidia.com/tesla/550.54.15/NVIDIA-Linux-x86_64-550.54.15.run
# Execute the installation package
sudo sh NVIDIA-Linux-x86_64-550.54.15.runUse nvidia-smi to check if the installation is successful. If the following screenshot appears, the driver has been successfully installed
Install NVIDIA Container Toolkit
Please follow the official installation instructions of nvidia-container-toolkit to install it.
The following example is based on Ubuntu 22.04 - NVIDIA T4
- Configure the production repository:
curl -fsSL https://nvidia.github.io/libnvidia-container/gpgkey | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit-keyring.gpg \
&& curl -s -L https://nvidia.github.io/libnvidia-container/stable/deb/nvidia-container-toolkit.list | \
sed 's#deb https://#deb [signed-by=/usr/share/keyrings/nvidia-container-toolkit-keyring.gpg] https://#g' | \
sudo tee /etc/apt/sources.list.d/nvidia-container-toolkit.list- Update the package list from the repository:
sudo apt-get update- Install the NVIDIA Container Toolkit package:
sudo apt-get install -y nvidia-container-toolkitRun nvidia-container-toolkit -version to verify that the installation was successful
- Configure Docker
sudo nvidia-ctk run-configure --runtime=docker# Restart the Docker daemon
sudo systemctl restart docker