Quick question; what browser?

/R

On Mar 30, 1:29 am, Dave <[email protected]> wrote:
> I wanted to share some very basic performance numbers and get the team's
> take on these.
>
> First, some basic info:
>
>    - I have wmode=direct on (let's get that out of the way)
>    - We have a 3952 poly character model.  We're using 8 different
>    textures/materials.  7 are 512x512, and 1 is 256x256.
>    - We have 2 lights in the scene.
>    - I've added a basic "clone" keyboard event handler that clones the
>    character mesh and adds the clone to the scene.
>    - We're trying to simulate "unique" characters in the test.  Different
>    textures, different geometry, etc.  We are trying not to take advantage of
>    instancing, as our application eventually will have many unique characters
>    in the scene.
>    - At the moment, all geometry in the scene is within the camera view.
>     Culling code is of course running, but everything is testing "in frustum".
>    - Hardware: Core i7 Q740 1.73GHz with 8GB RAM.  NVidia Quadro FX 1800M
>    GPU.
>
> The numbers:
>
> *# Chars / # Polys / FPS Range*
>    1  3952  52-62
>    2  7904  41-47
>    3  11856  38-41
>    4  15508  32-34
>    5  19760  27-29
>    6  23712  21-27
>    7  27664  21-25
>    8  31616  18-22
>    9  35568  17-19
>    10  39520  15-18
>
> The questions:
>
>    1. The Mesh.clone() method - does this meet the "unique characters"
>    requirement of my test above? Is there instancing going on underneath?  It
>    looks like clones share geometry and textures, but I may not have dug deep
>    enough - the renderer itself may still create unique vertex buffers/etc for
>    this data when rendered.
>    2. These numbers seem abysmal in comparison to what I've seen in some
>    blog-posted molehill performance tests.  I have not looked far under the
>    hood of these tests yet.  (a)
>    http://iflash3d.com/performance/unity3d-vs-molehill/ (b)
>    http://www.nulldesign.de/2011/03/02/molehill-demo/.
>    3. Based on what you know about your own internal architecture, and
>    limitations of molehill, is there anything obvious in what i'm doing above
>    that would be causing the abysmal performance? Big textures?
>
> Any additional thoughts would be most welcome.
>
> -Dave

Reply via email to