OpenFrameworks in the amazon EC2 cloud

I have an openframeworks program that I want to run on the Amazon EC2 cloud.

It will render to a window for a long time, then save a frame, clear the window and start over.

I can connect vis SSH terminal, but I can’t run Xwindows apps. So I got VNC installed, but it fails trying to draw the window using VNC.

Anyone have any ideas?

Can I use an FBO and save the .JPG?

I don’t know much about how FBO’s work.


Seems like the GPU enabled instances run CentOS, trying to get codeblocks installed on that one… will report back.

I have codeblocks installed in CentOS, but I can’t get the dependencies installed to compile. It hangs on GSTutils.

I’m going to see if I can get a local setup of centos rolling and just make binaries and run them in the EC2 instance…

this is going to be so cool when I get it running

CentOS is basically RHEL. I’m figuring out all the dependencies, I’ll check in once I get this working.

I’m going to bankrupt myself with EC2 instances hahahah


just saw this post, indeed it will be pretty cool! : )

i have never used ec2, but the latest ubuntu versions have some kind of support for it:

also if centos is mainly rhel, the fedora scripts should do.

Have you been able to make it work yet, memphistechno? It sounds all pretty complicated to me but thats because Im just in the process of familiarizing myself with Amazon EC2 cloud. What kind of openframeworks program do you have?

I’m banging my head against this, I can’t seem to get a remote X session to spawn OpenGL windows.

Is there any way to get OF to run on the command line only? Ie. not spawn a GLUT window?


int alpha = ofMap(NUM_PTS,0,500,0,255);  

ofMap maps a value from one range to another, in this case from 0…500 to 0.255

I’m doing some more testing, this is going to be a major part of my future body of work ( supercomputing lends itself so well to art ). Don’t worry I haven’t abandoned the idea.

My next paycheck I have budgeted in far too much money to fuck around on the EC2 cloud.

The dual Tesla instance is intense the cuda strength is unimaginable.

Sorry to revive an old topic, but did anyone have success running OF applications on Amazon EC2?
I recently spent a day trying to get OpenGL running on a virtual Ubuntu instance, but nothing seemed to work for me.

Yes… But it wasn’t easy at all and only by using a virtual GPU

I got it running on a Windows instance by following this guide:

Thanks, we’ll need Ubuntu, but reading any success story can be helpful :slight_smile:

@evilliam if you managed to make it work with linux, would you share what worked for you? thanks a lot :slight_smile:

We downloaded O/F and compiled on the actual EC2 box rather than trying to move artefacts across. The hard part was all the Jenkins / Chef stuff that our ops guys here did. But if you don’t CI I would compile your source on the Linux box using make and the O/F Linux tutorial

Make sure you have Mesa installed and working :

And try and keep your virtual window as small as possible as the update / draw loop in the software stack is very cpu intensive in the software stack.

I’m using @bakercp’s ofxhttp as the server

We’re using all of this to make personalised kids books


To be clear, I’m using cheap boxes with no GPU here, not the full on GPU backed extravaganza outlined above.

Thanks for your answer: our problem seems to be that, even if we installed all the dependencies, we don’t have OpenGL support on display 0. So we can compile OF code, but cannot run anything using GLFW. Tricky sysadmin problem :slight_smile:

And by the way: I love the books!

awesome - we use virtual a GPU (Mesa) to have openGL support - it’s virtual.

This is incredibly slow for shader code - hundreds of times slower - but it’s ok at doing simple stuff.

Thanks,we just have to render an obj with a fixed pipeline light, so we’re in the simple stuff realm :slight_smile: