*** Vendetta 1.8.78
- Improved Shader 2.0 graphics and fixed other shaders.
- New compressed normal maps for Shader 2.0 hardware to improve on video memory usage. They are loaded as uncompressed maps for non-Shader 2.0 hardware, taking up the same amount of video memory as before.
- Fixed bug where items in local non-active ships were not sellable unless there is an active ship.
- Fixed major OpenGL vertex buffer crashing bug.
- Improved texture load time slightly.
- Initial loading screen now has a percent complete counter. It is a counter of how many objects/textures are loaded, not based on time, so some percents are faster than other percents.

You guys suck at posting update threads.

it's a game of chicken and you lost. better luck next time.

so where's the fucking chicken?

I like chicken.

om nom nom, oh, wait, heres the chicken. Its a bit crunchy, yannow, with all dem bones and crap.

You mentioned that you might be improving some of the old graphics assets with this one... any pointers to anything that was sharpened/prettified/etc?

the initial game load screen is noticeably faster for me now. still only using 1 cpu core though... :)

rogue: that'll mostly come on Fri. The EC-89 normal map is a little better in the current release, and a few of the others were snazzed slightly (subtle), but you'll be able to tell with the next batch.

I didn't have the ability to release texture-compressed diffuse maps in this version, yet, so I had to hold off on increasing resolution of various stuff. Still working on that now, while Ray makes all the fallbacks.

I've done a small comparison between the old client and this one. Luckily it still worked.
In brackets is the fps with shaders disabled.

close to station
old 43(78)
new 44(77)

dantia l6
old 30(78)
new 30(80-81)

space with sun
old 81(139)
new 81(139)

Soo ..the update didn't really change much for me and some of the difference could be rounding errors.

That's with the 64bit client on arch linux with the onboard graphics of a asrock a780gxh/128m at pretty highish settings

Well, we didn't really change anything that would directly cause improved framerates. What we're doing is reducing memory usage, which can sometimes improve framerates (and load times) in that it prevents texture thrashing, plus it can also reduce GPU memory bus contention. But the point of this was not so much to "speed things up" as it is to "use the memory much more efficiently, so we can do More Stuff with it". Same is true of the memory bandwidth, it's kind of a "works great" or "totally sucks" switch.. and the more textures we're using at once, the more bandwidth we need. If we want to add shadows and other new features, it's just as well that we make efficient use of what's there.

We are looking at doing some things that could directly improve framerates, like adding a partial-precision shader path specifically tuned for the GeForce FX series of chips. Plus some other optimizations to normalizing maps and such. But the former hasn't been released yet, and the latter has even been written yet. The partial precision (FP16 2.0a) shader path will probably only improve things for certain cards though.. really recent hardware will probably not benefit.

Yeah.. I was just curious if the better memory usage would show some improvements, because this onboard chip is somehwat bandwidth limited compared to dedicated cards.

I can't really complain, the game is perfectly playable as is.

The AMD 780G is basically an integrated Radeon HD 2400, which should be fairly decent. The 128MB of dedicated vram on that board should actually have reasonable bandwidth, compared to shared-only systems. Unfortunately, I don't think the new partial-precision path will help you at all, it will always use 128bit FP regardless of shader hinting.

Other possible changes, like optimizing how we normalize and stuff, could be of use.