GPUFractal-2

[Burton Radons]

Matti Niemenmaa Wrote:

> Burton Radons wrote:
> > Ohskays, update time! Pretty much everything is changed. Added language
> > parameter to schema, added Cg (Cg and GLSL use the same formulae, only their
> > base schema file is different), added global parameters, greatly improved the
> > code (although I still wouldn't bring it home to mama), greatly improved text
> > rendering (the X850 renders it weirdly, I don't know what's up with that),
> > added gradient parameters (format is the same as in Ultra Fractal), and you
> > can zoom and pan with the same button.
> 
> Nice! The Cg version works on my X800XL now, with some limitations.
> 
> Vendor: 'ATI Technologies Inc.'
> Renderer: 'RADEON X800 Series x86/MMX/3DNow!/SSE2'
> Version: '2.0.6847 WinXP Release'
> 
> Changing the "Inside" option gives this, I guess C6003 is a limitation of my card:
> 
> CG Compilation failed:
>         (114) : warning C7011: implicit cast from "float4" to "float2"
>         (114) : warning C7011: implicit cast from "float4" to "float2"
>         (0) : error C6003: Arithmetic instruction limit of 512 exceeded; 519
> arithmetic instructions needed to compile program

Yeah, I gave up trying to figure out how to keep under the limits for the R420. What you scrolled into was exponential smoothing, which attaches code to each loop. I had previously hacked down MAXITER to get under a limit (I think it was 18), but then it stopped working in my last tests and I decided to leave it as is rather than make a mess out of the code for a card I didn't actually support.

One thing in my plans is to make the code more forgiving of errors, so that it will just say that this is an invalid condition, here is why, enjoy this black screen until you fix the problem.

> Trying to set Gradient Offsets and Gradients to their defaults with middle click
> gives this:
> 
> ArrayBoundsError gpufractal.main(265)

Whoops!