[Xiaopang]

well, i don't agree with you that the particle count was set too high. as said already, i ran the game on an x1600 and also had no prob on my gtx 260 despite those settings. you just have to balance the workload for the card by lowering other settings. i have no doubt that the programmers knew exactly which values they needed to make effects like using the water gun and splattering water look good. i also don't agree that the game will max out cards for 5 years to come, considering that the 3d-market is only 13 years old and how far cards have come along in the last 5 years, the increase in processing power we will experience will be massive. if rumors are true, then with the release of the gtx 300 series this game will just be one among many again. check out crysis. it can be run smoothly on today's hardware, e.g. gtx 260, 280 etc. but these are already a year old, so crysis' hunger for massive 3D-power could be satisfied after only a year. also, let's not forget that you can go up to quad-sli, which would make any game playable at 5 Megapixel resolutions, but why do we have the technology if there is no game that would profit from it, because all can be maxed out with single cards already?

i do agree that modding the physx value holds much potential for future benchmarking, but i doubt it will improve the visuals much. at certain levels, the particles would overlay so much, so that you wouldn't see a difference to a value were overlay is minimized. hmmm, you kind of gave me an idea right now. let's see what happens if i crank those values up to a million lol

i also think that it's great if developers do not limit their games to the cards that are available at release. this gives some headroom for the future, to experience even better visuals in a year or two. if i have the option between playing a toned down game, or a game that i can tone down myself with the chance of cranking it up for future hardware, then i'll always route for the latter. one example i'm thinking of right now is silent hill 3. even though the game came out in 2003, it supports texture rendering up to 4096x4096, something that no computer back then could have handled, but thanks to that ability, people could pump out better visuals over the years. compare that to highly anticipated titles like half-life 2. it only used texture sizes of 256x256 because cards back then couldn't handle more. so even if you play the game at high resolutions it doesn't look much better.

i don't really know much about 3D vision, but if it is comparable to the old shutter technology then there shouldn't be a problem.

if all your 4 cores are maxed out then this is a good thing, because you can get the maximum out of your cards. i don't know what version of the game you have, but as i already said, the european one is multicore capable right out of the box and the american is even newer, so it must be too. if there's a version that's not multi-threaded, then it might be the russian one, but i doubt it. the fact that they released the first patch so quickly with SO many bugfixes makes me doubt that they also implemented multithreading, because that would have required a rewrite of most parts of the code. no way that they pulled that off within only a couple of months.

as for windows balancing the load, that only applies to threads. if an application is not multi-core compatible as it is described in lay terms, it only uses a single thread and these can never be split among cores. windows will assign apps to different cores according to their workload and it will limit them to that core if they only support a single thread, but it can't split that thread among several cores. you can't drive on 2 streets with the same car at the same time.

the threading option in the cfg seems to be a remnant of the early days of the game. to me it looks like they needed a flag to force the game into a single thread mode for single-core cpus that didn't support hyperthreading. playing with that option did nothing for me, so i think it's inactive

i might have been not accurate enough with my usage of supersampling. what i tried to say was, that the result is similar to it, while a different technique is applied, that of course is not comparable to supersampling. thanks for the correction. i'll straighten it out.

btw, are you running the game on xp or vista? x86 or x64? could you do me a favor and upload a 1080p screenshot? i'd like to examine the jaggies. some games, like stalker, don't show any improvements at certain resolutions any more. i'd like to know if this is the case with cryostasis too.

i have no experience with sli, but can't you define a card as physx-exclusive so that it will do all the physics rendering, while the other one just handles the graphics? if that's the case, did you try it with and without that setting? i'm interested in knowing which would be more efficient.