|
|
Joined: Mar 2002
Posts: 119
Member
|
Member
Joined: Mar 2002
Posts: 119 |
So what happened to the 'endianness nightmare' that ZiNc was supposed to be ? Also, does the interpreter actually still really work after all this time it has been gathering dust all this time ? thats interesting 
|
|
|
|
Joined: Dec 1969
Posts: 921 Likes: 3
Senior Member
|
Senior Member
Joined: Dec 1969
Posts: 921 Likes: 3 |
Remembering some of the code Gil has worked on, I'd imagine he hasn't submitted it back yet because it's a royal mess and he's embarrassed by it! Anyway, I've just sent a mail to him to put you two in touch.
|
|
|
|
Joined: Mar 2001
Posts: 17,284 Likes: 268
Very Senior Member
|
Very Senior Member
Joined: Mar 2001
Posts: 17,284 Likes: 268 |
DC: the interpreter actually still works, yes. smf and I use it all the time for debugging stuff. The endianness nightmare hasn't been disproven yet since nothing runs without plugins, but it was alarmingly easy to make it compile and link all the same. I was expecting hell and high water with the plugin loader, but OSX has a working emulation layer for Linux's ld.so stuff so that stuff just works.
|
|
|
|
Joined: Oct 2004
Posts: 63
Junior Member
|
Junior Member
Joined: Oct 2004
Posts: 63 |
Hmmmm.... I wonder if there are still *surprises* hidden in the code that will show up in the world of ppc.
|
|
|
|
Joined: Oct 2004
Posts: 63
Junior Member
|
Junior Member
Joined: Oct 2004
Posts: 63 |
PeteB: Why you cannot filter at shader level?
arekkusu: I just read that 5200 as well as my 5900 are supposed to support unlimited shader depth (at driver level) in OpenGL 2.0. Still, dont know how big is native shader in 5200, but 5900 has 1024 instruction slots, and it is really a lot. It is easier to run out of vertex shader than fragment shader.
|
|
|
|
Joined: Mar 2001
Posts: 17,284 Likes: 268
Very Senior Member
|
Very Senior Member
Joined: Mar 2001
Posts: 17,284 Likes: 268 |
I think all GeForce 5xxx have the same shader depth, just some execute them faster :-)
BTW, Mac minis have the ATI Radeon 9200 - I believe it's faster than a GF5200, but it's also a lower-generation part (DirectX 8.1 vs. 9.0 on the PC side), so it's limited to 22 instructions with 6 texture inputs for pixel shaders and 128 instructions for vertex shaders.
|
|
|
|
Joined: Feb 2004
Posts: 314
Senior Member
|
Senior Member
Joined: Feb 2004
Posts: 314 |
The max fragment program size on 9600/9700/9800 is 96 instructions.
Now ask me which is easier: writing a different texture method for every GPU from 8500->X800, GF2MX->6800, or just doing everything on the CPU with a smart texture cache.
|
|
|
|
Joined: Mar 2002
Posts: 119
Member
|
Member
Joined: Mar 2002
Posts: 119 |
easy, easy .. its not about easy. Its all about fast emulation  If you want easy then dont even bother with texture cache, it'll work anyway. A smart texture cache still will have trouble with games that color cycle the palette during special effects. Too bad that the people who need the CPU speed are also the ones that have the cheap/old graphics cards.
|
|
|
|
Joined: Oct 2004
Posts: 63
Junior Member
|
Junior Member
Joined: Oct 2004
Posts: 63 |
Originally posted by DynaChicken: Too bad that the people who need the CPU speed are also the ones that have the cheap/old graphics cards. Cheap/old/ATI gfx cards have paletted textures 
|
|
|
|
Joined: Feb 2004
Posts: 314
Senior Member
|
Senior Member
Joined: Feb 2004
Posts: 314 |
No, they don\'t . Not even on Windows . ATI has never supported GL_EXT_paletted_texture.
|
|
|
2 members (R. Belmont, 1 invisible),
208
guests, and
2
robots. |
Key:
Admin,
Global Mod,
Mod
|
|
Forums9
Topics9,372
Posts122,600
Members5,085
|
Most Online1,529 Jun 7th, 2025
|
|
These forums are sponsored by Superior Solitaire, an ad-free card game collection for macOS and iOS. Download it today!
|
|
|
|
|