In Situ Animation of Accelerated Computations 🔬
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
Modules
client
documentation
example
example_renderings
helper_scripts
lib
server
.gitmodules
INSTALL.md
LICENSE
README.md
TUNNEL.md
isaac.png
isaac.svg

README.md

ISAAC

In Situ Animation of Accelerated Computations

Wakefield visualization from PIConGPU

About ISAAC

Many computations like physics or biologists simulations these days run on accelerated hardware like CUDA GPUs or Intel Xeon Phi, which are itself distributed in a big compute cluster communicating over MPI. The goal of ISAAC is to visualize this data without the need to download it to the host while using the high computation speed of the accelerator.

ISAAC insists of two parts: The server and the insitu library. Furthermore there needs to be a client, which is able to show the transfered stream and meta data. A reference HTML5 client is provided, but needs to be adapted to specific simulations and is not part of ISAAC itself (but still in the repository).

Simulation code has just to add some calls and settings to the insitu template library. After that the server will notice when a simulation is running and give the user some options to observe the computations on the fly. It is also possible to send meta data back to the simulation, e.g. to restart it with improved settings.

Installing requirements, building and using in own application

Please see in INSTALL.md for installing, building and using ISAAC. If you need to install ISAAC on a server not accessible from the outside you need to tunnel the connections of the clients. A more detailed documentation about using ISAAC can be found here.

Client

Inside the client directory lay three files:

  • interface.htm
  • interface_vlc.htm
  • interface_presentation.htm

The very first uses direct JSON injection, but no real streaming protocol like RTP. The second can use RTP, but needs the (free) vlc browser plugin installed to work. Furthermore the server most be able to directly send UDP packages to you, which are blocked by most firewalls. Last but not least this adds some latency for the h264 encoding. However the stream itself will use way less bandwidth. The last client version is for presentation purposes, the table with meta data values is not shown, but some are shown directly in the stream box. E.g. if you visualize PIConGPU with ISAAC enabled this will show you the count of particles and cells used.

Known issues

  • If streaming over twitch or another rtmp compatible service is used, but the rtmp port (1935) ist blocked or a wrong url passed, the server will crash because of the underlying gStreamer rtmp implementation.

Licensing

ISAAC is licensed under the LGPLv3.