Move your GPU to the Cloud with powerful, scalable, and elastic virtualization technology
GPUBox documentation online


GPUBox is a technology useful for everyone who uses GPUs for something more than gaming. It is the first publicly available solution that allows many users to dynamically share dozens of GPU devices at the same time, for such purposes as 3D rendering, bioinformatics or molecular dynamics.

By implementing GPUBox to your infrastructure, you can achieve significant advantages:

Virtualization of the GPU devices
GPUBox separates application and operating systems from the GPU devices. Thanks to this, the GPUs can be also used on Virtual Machines.
Providing the GPU devices on demand
Users can add and remove the GPUs at any time without the need to restart the system.
Lowering total costs
The GPU utilization is improved by sharing the pool of the devices between many users. The total number of GPUs required for efficient work within a team can be lowered.
Reducing the power consumption
Sharing the GPU, and lowering the total number of devices, allows a decrease in power consumption.
Exceeding the physical limitations
The number of available PCIe slots on the motherboard is no longer a limitation when using the multiple GPU devices.
Outstanding performance
Using multiple GPUs over the GPUBox infrastructure significantly decreases the rendering times.

The current version of GPUBox comes with features such as:

  • Windows and Linux support
  • Convenient Web Console
  • Command line interface
  • Accounting and monitoring tools
  • Native InfiniBand support
  • Extensive documentation and detailed messages description


Compatible applications

Application Type Known issues
BarraCUDA Bioinformatics -
Blender Cycles 3D renderer -
CUDASW++ Bioinformatics -
HOOMDBlue Molecular Dynamics -
Gromacs Molecular Dynamics -
John The Ripper Password cracker -
LAMMPS Molecular Dynamics -
mCUDA-MEME Bioinformatics -
Octane Render 3D renderer

Point 4. of OTOY Terms and Conditions of Trade

V-Ray RT 3D renderer -
WideLM Analytics -

* Only 3D renderers are supported by official release. Other application needs additional libraries. For more information please contact support.

System requirements

Operating systems
GPUBox servers and clients can be run on most of the 64-bit Linux distributions. The following distributions are officially supported:

  • Ubuntu 14.4
  • Debian 7.6
  • Fedora 20
  • CentOS 6.5
GPUBox Client can also run on:
  • Windows 7 64-bit
  • Windows 8 64-bit

Network connection
For using the GPUBox infrastructure we recommend a minimum 10 Gb/s network connection.

CUDA-enabled GPU

Subscribe to our newsletter to receive latest
news and updates.