SoftWire? commercial, searching, anything, message

andy

, I'm redirected to a commercial web site, and searching for it doesn't turn up anything useful.

, I'm redirected to a commercial web site, and searching for it doesn't turn up anything useful.

It's owned by that company now, as the redirection message says.  So you'll have to obtain it through that company most likely.

SoftWire and swShader are now owned by TransGaming Technologies Inc. You will automatically be redirected to in 3 seconds. If your browser doesn't automatically load, click HERE.


But it was (L)GPL code? Doesn't that mean anyone has a right to redistribute it?

).

Don't hesitate to contact me and show what you've created with SoftWire!

So that probably means that I can send the sourcecode to you?

It's all there:
projects/softwire
softwire

It seems like the source has been closed. But since an LGPL cannot be reversed for released material, we can always create a fork, if we have to add stuff to it.

There are no problems in it for AviSynth, except there will be no further development from the original author. I would love to add AMD64 support, but I simply haven't got the time to commit to such a project. I've looked through the code, and it looks really well-organized. SSE3 support should not be that big a deal to do, though.

Thank you all for the information (and especially to tsp for the offer!). I have obtained it without any trouble.

also, some info (and Shodan's post) here :
forum/forum....orum_id=184692
and a link to gna (last post of the above link), where a version is hosted :
projects/softwire/

Just noticed, that SSE3 opcodes are actually present in Softwire.

What is Softwire?

It's a dynamic assembler, i.e. it lets you create assembly (well, machine code) at runtime and then execute it. I think sh0dan has used it in places in AVISynth, and Manao in MaskTools 2.0.

Its advantage is that it lets you tailor-build assembly for the case you're working with. For example, at runtime, when you know the properties of the clip your working with, you can construct a routine that is specifically designed to work on e.g. 704-pixel-wide frames, which is potentially faster than a corresponding general routine which can handle arbitrary frame widths.