qalgora-Q Docs Hub量子文档 ✦ Ask AI✦ 问问文档
/

Data Center Deployment (Planned)

○ Planned · Not yet implemented
The capabilities described on this page are planned and not yet implemented or released. They explain future design directions and should not be interpreted as delivered features.
Planned distribution — reference implementation not yet built
The GPU/multi-node build and the Docker image described here are planned; the reference implementation does not exist yet. The current reference build is a CPU-only Python package installed from source. This page documents the intended deployment only.

Build and deploy qalgora-Q from source on multi-GPU, multi-node infrastructure.

Prerequisites

  • A supported GPU with a matching driver and GPU compute runtime.
  • A C++20 toolchain and CMake.
  • An MPI implementation for multi-node runs.

Building from source

git clone <qalgora-repository-url>  # repository not yet public, illustrative only
# cd qalgora
# bash scripts/build_qalgora.sh        # planned — builds Python + C++ support

Installing on the host

# export QALGORA_INSTALL_PREFIX=/opt/qalgora
# bash scripts/install.sh                 # planned — not yet available
# wire up GPU runtime libraries and MPI on each node
Multi-node
Ensure the same build, driver, and MPI versions across all nodes. Validate the fabric with a small mgpu job before scaling to production workloads.

数据中心部署（规划）

○ 规划中
本页所述能力属于规划功能，当前尚未发布或尚未实现。相关内容仅用于说明未来设计方向，不应理解为已交付能力。
规划中的发行版 参考实现尚未实现
本页所述的 GPU 多节点构建与 Docker 镜像均属规划中，参考实现尚未实现，尚未发布。当前参考实现是 CPU-only Python 包，从源码安装。本页仅用于说明预期部署方式。

在多 GPU、多节点的基础设施上从源码构建并部署 qalgora-Q。

前置条件

  • 受支持的 GPU，以及配套的驱动和 GPU 计算运行时。
  • C++20 工具链和 CMake。
  • 用于多节点运行的 MPI 实现。

从源码构建

git clone <qalgora-repository-url>  # 仓库尚未公开，仅示意
# cd qalgora
# bash scripts/build_qalgora.sh        # 规划中，构建 Python + C++ 支持

安装至主机

# export QALGORA_INSTALL_PREFIX=/opt/qalgora
# bash scripts/install.sh                 # 规划中，暂不可用
# 在每个节点上配置 GPU 运行时库和 MPI
多节点
确保所有节点的构建版本、驱动和 MPI 版本一致。扩展到生产负载之前，先用小规模 mgpu 任务验证一下网络互联。