Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/prsyahmi/GpuRamDrive

RamDrive that is backed by GPU Memory
https://github.com/prsyahmi/GpuRamDrive

Last synced: 10 days ago
JSON representation

RamDrive that is backed by GPU Memory

Host: GitHub
URL: https://github.com/prsyahmi/GpuRamDrive
Owner: prsyahmi
License: mit
Created: 2016-11-25T12:34:22.000Z (almost 8 years ago)
Default Branch: master
Last Pushed: 2022-09-04T16:55:32.000Z (about 2 years ago)
Last Synced: 2024-08-02T11:14:45.647Z (3 months ago)
Language: C++
Size: 539 KB
Stars: 1,031
Watchers: 38
Forks: 85
Open Issues: 35
Metadata Files:
- Readme: README.md
- License: LICENSE

Awesome Lists containing this project

README

# GpuRamDrive
Create a virtual drive backed by GPU RAM.

![Screenshot](https://repository-images.githubusercontent.com/302840646/59966600-3285-11eb-94aa-9c2c917da458)

This application simply allocates a memory buffer inside GPU RAM and use it as virtual ram disk. The project is made possible by ImDisk and its proxy feature.

Using GPU RAM isn't as fast as host main memory, however it is still faster than a regular HDD. The result below is taken on my system with GTX 850M and i7-4710MQ. As IO operation is happening between the CPU and GPU, the GPU can become active from idle state and might causes system to draw more power. This merely just a PoC, user who search for this kind of solution is advised to upgrade the RAM or buy a faster storage.

![Benchmark](https://cloud.githubusercontent.com/assets/1040494/20632692/65470470-b37a-11e6-908d-e08687a757d3.png)

People who interested in this might also want to check BadMemory: https://github.com/prsyahmi/BadMemory

## Usage
1. Install ImDisk Virtual drive (https://www.ltr-data.se/opencode.html)
2. Download GpuRamDrive from https://github.com/prsyahmi/GpuRamDrive/releases
3. Run GpuRamDrive_x64.exe or GpuRamDrive-cuda_x64.exe according to your platform

## Compiling
To compile, open the solution, configure the target platform and build the project.

## License
This project is licensed under MIT. See LICENSE.