OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Infrastructure & Serving

Di Hpc

Di Hpc is a Python library of accelerated operators for reinforcement-learning components, including GAE, n-step TD and LSTM. The operators support forward and backward propagation and are intended for use in training, data collection and testing modules. Its documented setup targets Linux with CUDA and PyTorch.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Di Hpc — OpenViST project profile in AI Infrastructure & Serving.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Di Hpc when a Linux-based PyTorch reinforcement-learning workload needs accelerated operators for GAE, n-step TD or LSTM, and its documented CUDA, Python and PyTorch requirements fit the intended environment.

Who it is for

  • Reinforcement-learning developers integrating optimized algorithm operators into PyTorch workloads.
  • Practitioners who want to benchmark the library's GAE operator on Linux.

Documented uses

  • Accelerate GAE calculations in a reinforcement-learning workload.
  • Add n-step TD operators to an RL training pipeline.
  • Use accelerated LSTM operations in training or data collection.
  • Run the documented GAE benchmark script.
Before you choose

Requirements & limits

The public GitHub repository releases the library under the Apache 2.0 license. The README documents installation from a matching wheel or from source; it does not describe an account requirement.

  • The library is narrowly focused on reinforcement-learning operators rather than providing a complete training application.
  • The README lists older, specific CUDA, Python and PyTorch configurations; one setup also specifies GCC 5.4.0.
  • The project recommends sharing an environment with DI-Engine and states that PyTorch 1.1.0 to 1.10.0 should be suitable.
  • The documented benchmark command targets GAE and reports a benchmark result; no expected result values are given.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install a matching Linux wheel and run the GAE benchmark

Install Di Hpc from a provided wheel and invoke the documented GAE test script to obtain a benchmark result.

Before you start

  • Linux
  • A compatible documented environment: CUDA 9.2, recommended PyTorch 1.5 and Python 3.6, 3.7 or 3.8; or CUDA 9.0, GCC 5.4.0, PyTorch 1.1.0 and Python 3.6 or 3.7.
  • A wheel matching the Python version: cp36, cp37 or cp38.

Steps

  1. Obtain the wheel matching the Python version from the project README.

  2. Install it with pip using the wheel file path.

  3. Run the command below from the project checkout to obtain the documented benchmark result.

    Documented command
    python3 tests/test_gae.py

Expected outcome

The README says the test command produces a GAE benchmark result.

Guide limitations

  • The README does not provide expected benchmark values or performance guarantees.
  • The documented wheel and environment options are limited to the versions listed in the README.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
16 October 2026
Original website title
opendilab/DI-hpc · Official homepage title · retained in its source language
How we review sourcesSuggest a correction