TES Skyrim 0.182

[--JawZ--]

Boris

I tested the binary file you posted and I could not see any performance gain or loss, only noticed that the detailed shadows was better placed on some objects or rather that they spread out on objects a bit better, in my opnion that is. I did not check to see if there was some settings or effects that were broken though.


Oyama:

No but will try and see what happens, *Comes back with a fried gfx card, and says it's working all right right just like "Nvidia" standard*

[Tansarville]

Boris

I tested new binary and and i didn't see noticeable FPS boost or drop. Also i didn't notice any bug so far.

[KuroTenshi]

Oyama :

Yes, still broken in it. Still broken too in 325.71 WHQL (shipped with Windows 8.1 (leaked)).
But 325.71 are a lot better than 320.xx.

Made a new post on Nvidia's forum.

Is it me or 0.182/0.183 is more performant ?

Edit : Just got a feedback from Nvidia. They found the problem. Will be fixed in next branch (in the next 325.xx driver in fact ).

[turgor128]

Удивительно то что в 2013 году графику такого уровня которую дарует мод нельзя увидеть в играх (бенчмарк крузис 3 нельзя брать в расчёт т.к это скорее всего маркетинговый ход) неужели им так не нравится их работа и они тупо кладут болт на графику, неужели это не интересно прочитать в журнале про то как постарались программисты и авторитетный журнал бы поставил 10/10 за графику??Отверстием в нижнем конце пищеварительного тракта чувствую не ладное.
PS Сори за флуд

[Oyama]

Edit : Just got a feedback from Nvidia. They found the problem. Will be fixed in next branch (in the next 325.xx driver in fact ).

Thank you ! Good news these are... !

Boris
No differences with the new binary so far. Nothing different in performance matter, and shadows are same apparently... I need to do some more testing with more interiors.

--JawZ--

[ENBSeries]

Well, this means that only mine old videocard have better performance then. Another proovement that i can't optimize any more without modern videocard.

[Oyama]

Well, this means that only mine old videocard have better performance then. Another proovement that i can't optimize any more without modern videocard.

Hmm... I remember the explanations you gave to me or to midhras about updating your GPU... lots to update around, OS, PSU... ?....
This would be tough... Thus said, midhras and I still have some GPUs to spare if you need it.

Maybe wait for Nvidia to correct the issues with next branch and ask us for testing different optimizations.
I mean, ALL latest optimizations you did were based on your actual GPU, and even if it's old, optimizations were visible and working for ALL modern cards users...

[ENBSeries]

That's not similar optimizations to those i have in plans and trying to make. For example, if i have code:

Code: Select all

while (i<samplecount)
{
    uv=computescreencoordinates();
    mix=computeblendingfactors();
    result+=tex2Dlod(texture, uv) * mix;
}

It can be optimized for example to this:

Code: Select all

while (i<samplecount)
{
    mix=computeblendingfactors();
    if (mix>0.0)
    {
        uv=computescreencoordinates();
        result+=tex2Dlod(texture, uv) * mix;
    }
}

Or to this:

Code: Select all

while (i<samplecount)
{
    mix=computeblendingfactors();
    if (mix>0.0)
    {
        uv=computescreencoordinates();
    }
    result+=tex2Dlod(texture, uv) * mix;
}

And on my videocard not all of them working as they must in theory, slower in some cases. When i tested on radeon 6450, performance increased as it should. Another optimization is when in single cycle placing several computations together, so many math and some texture reading operations share the same data and not recomputed. But the fact is different for me, if remember values, then on radeon two samples in parallel were slower than one sample by 10% (so each sample is almost twice faster) and on my current card performance decreased 2.5 times, so no gain at all. Of course many users still have old videocards like mine, but they can use old versions of the mod then.

[Rem_NL]

I am under the impression that performance has increased but I have to do some proper test first to confidently state this. Will get back on the subject later

[Oyama]

Boris
OK, these parts of code I'm able to "understand".
At least, written as they are.
Strange indeed that your GPU isn't reacting as it should with these. Especially the 1st optimization.
Can this be something related to a DirectX / Windows version problem, and the way older drivers handle them ?