Installing gevent

If you're a Python noob like me, installing gevent is not trivial. In my case, I'm using Ubuntu 10.04. These are the steps to follow:

1. Install libevent package (apt or synaptic)
2. sudo easy_install greenlet
3. sudo easy_install gevent
4. ldconfig
5. test that 'import gevent' and 'from gevent import socket' work.

Step 4 was necessary in my case since step 5 wasn't working, but you may try to see if it does in your case.

Now, I was interested specifically in the datagram server. At the moment of writing this, the release version of gevent was 0.13.6, which didn't include it, so I had to perform a check out of the latest version in the repository.

However, I didn't want to install it as I wanted to be able to switch between different versions or easily move a setup to another computer. Therefore, I wanted to manually build it into a given directory. For this, I had to do a few more undocumented steps:

1. sudo easy_install cython
2. (in gevent root folder) python setup.py build, which generates a build in build/lib.linux...
3. In my Eclipse project, link the build folder as an external library source.

Now, I wanted to be able to execute my project without eclipse... so I added a main file which set up the paths:

...

GEVENT_PATH = "../gevent/build/lib.linux-i686-2.6"

import os
import sys

# append paths to PYTHONPATH
sys.path.append(os.path.abspath(GEVENT_PATH))

# begin importing and using gevent!

...

And finally... that was it!

Installing Ack on Windows

Ack is a great command line grep alternative. In order to install it on Windows, follow the instructions here:

http://blog.thecybershadow.net/2010/09/12/color-grep-for-windows/

Vertex throughput on modern GPUs

I've read that the optimal number of triangles in order to maintain 60fps that the GPU can process is something between 1500 and 4000.

Found this interesting link:

http://nuclexframework.codeplex.com/wikipage?title=PrimitiveBatch&referringTitle=Nuclex.Graphics

But according to this presentation about nvidia http://origin-developer.nvidia.com/docs/IO/8230/BatchBatchBatch.pdf?q=docs/IO/8230/BatchBatchBatch.pdf, the question is not how many triangles per batch, but how many batches per frame!

Following that last link, the conclusions that can be drawn are:

- Number of vertexes per batch: whatever between 1 and 10.000.
- Number of batches per frame (CPU bound): 25k * GHz * Percentage_available_for_draw_calls/Framerate
- Bad performance at a small number of vertexes per batch could mean:
- GPU vertex engines are idle but the triangles itself could be expensive
- CPU bound --> add triangles for free!
- GPU idle: add complicate pixel shaders for free!