![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
) Killer NIC relies on completely bypassing the Windows networking stack to leverage a gain - I don't think regular cards do this.IMHO I think that some places will elect to use cards in servers so that should an onboard NIC die you don't have to swap out the entire MB. Sure its another point of failure, but it beats missing your SLA because you bend some pins on the CPU/socket in your rush to put everything back together. I'm not sure if they provide a performance increase? Most server mainboards come with at least 2 NICs that are easily teamed.
In a desktop environment (which I think is what is being discussed here - for the most part
Is a $250 NIC worth it for most people? Playing MMORPGs, or even fragfests? Pings are faster? What's the benefit?
If worth it, may get one. I can afford it as well
I blame Dan
I don't own a physics card not because I think it's a waste of money but simply because I'm not sure if people will ever figure it out that this card could truely change things for the better, therefore not let me take advantage of a $150 peice of hardware.Exactly, people just don't get the potential that these physics cards truely provide..
imzjustplayin said:
..Ofcourse They will optimise Physics code and wont spill gflops on things not seen. Dev do optimise. That is a standard need, because instead the need of a PPU for ageia island map in Graw2 they would need a Cray suppercomputer if they waist more then 10x the gflops.
I did play Redfraction and it Destructable enviorment is crap. It's just a few wall's.
Sure more Physic give gamedevelopment and also the artist and map designers a higher work load. For instance in reality builder you must give objects it's Physics properties. And for destructable object you must create there destroyed counter parts. Special physics rich object like cloth ropes fluids need more atention and work to tweak it.
I only know Farcry in that listComing back to something like Deus Ex (Because non-on-rails games like that, or FarCry, or STALKER, or GTA are the ones we really want this stuff for), it's generally agreed that HongKong is one of the best environments, imagine if Ion Storm had had to build what was on the other side of all the walls, behind each locked apartment door, they'd still be working on it
Well for textures that used already procedural Textures Roboblitz to keep the size down for Xboxlive dowload size requierment to be able to make a bigger project then with brute force was possible. For geometry it's not used. But looks interesting aspecialy for small teams or indie developers.
U3 is long in development this means they can do PPU specific maps. Like Graw2 did. As the artist have also a large part of that time avaiable.That was longer than I thought it was going to be, the freaky thing is that I could keep going
The ultimate on-topic point is that U3 is not going to be a "PPU Game", because EPIC can't sell it if PPUs significantly alter the actual play experience or are required for more than a margin benefit, because nobody has a PPU. It's sort of a Catch22.
No it's exactly a catch22.
To be fair though, it was probably the Original Unreal, along with Quake2 that really broke 3d cards into the mainstream, even though neither game required it to work, so maybe , just maybe U3 can do it for PPU if the content is compelling enough. I don't think it will though. We'll see.
Ofcourse They will optimise Physics code and wont spill gflops on things not seen. Dev do optimise. That is a standard need, because instead the need of a PPU for ageia island map in Graw2 they would need a Cray suppercomputer if they waist more then 10x the gflops.
As for this fine detail. The GPU Physics accelerated Tech benches who show off just ONE physics feature like fluid ONLY. Show what impact fine grain Phsyics computing can show with 250 to 500Gflop dedicated to that. It looks more real.
A CPU dont have that horse power for a fluid flow in front of the player or a decent detaild Phyiscal tent in front of the player and will be replace with a splash decal or nondestructable stiff tent. Because it would have to do some priority average Gameplay Physics wich takes all reserved CPU cycles.
I did play Redfraction and it Destructable enviorment is crap. It's just a few wall's.
What more desturbing all kind off walls are undestructable expept one rare type, with a homegenious texture. While other weaker walls in reallife aren't destructable and only a few so at very low scale implemented. Those non destructable are more detaild and also the textures.
Beside that it don't support gameplay in a way, as it stand loose from it. So in single player it was just a another shooter. Clone nr 101?
I don't expect the PPU can do Full destrucable maps and structures with full realism and fine detail, but it would be a lot more to destroy in a map then Redfraction did. So on a larger scale or all but at a crude low res detail. Predefined debri.
Warmonger let you show at what level PPU can hanle destructable enviorment.
It would be at a level where it impact gameplay at large scale.
And then more power full PPU and gPU generation could pump up the Scale, fine grain detail and combination of Physics Feature up. Where it will look better and more emersive
Sure more Physic give gamedevelopment and also the artist and map designers a higher work load. For instance in reality builder you must give objects it's Physics properties. And for destructable object you must create there destroyed counter parts. Special physics rich object like cloth ropes fluids need more atention and work to tweak it.
Of course wenn PPU and GPU get so powerfull some things could be computed in real time. If destruction would be generated on runtime.
Object need material properties. They must have also a 3D texture.
As in Redfraction the Texture was a simple one and on break sides it was a darker simple texture. Also each material has material Phsyical behavior. Like brick wall consist of bricks that are stronger then the spacie between them in need more force to brake the brick it self. And such wall would brake differently. BEsides that there is the wooden wall.
Consist alot of wooden parts that destroyed seperatly and have a more streght in the longway direction of there wood material structure.
This methode need a technolgie that give 3d engine a 3D texture
This means a Artis just make the object give it there Physics properties and 3D texture
Doesn't have to do the fix destuctable instances as it's generated on the fly.
I only know Farcry in that list
Well for textures that used already procedural Textures Roboblitz to keep the size down for Xboxlive dowload size requierment to be able to make a bigger project then with brute force was possible. For geometry it's not used. But looks interesting aspecialy for small teams or indie developers.
I think this is where the game industry must be heading to takle the complexity grow of game development and produktion. Because game projcect and design get more complex with more work and at a time trowing more people at it doesn't work.
As more people means more expensive produktion and the yield of result get less with the overhead to steer them all and keep them busy.
But instead of that the work load shift to game engine and the need for Algoritme expert mathematic guru's And artist with a procedural feel.
The problem is in the brute force way it take a large artist team to make some game content. But if games get more complext and also it's content the can make less content as in campain length and game will be shorter or be longer in development. The solution at the moment for that is Episode game design.
Will wright has a solution let the player create there own content thru community. But this is part of a gameplay level and not suitable for current style games.
Where content is a community effort to enrich the full game so extend content or rebalance it. They started with textures. With Procedural content also geometry your game get very compact But Willwright result look like fisherprice content.
I think to make high detaild contend like as made by hand or mouse. It need a lot of complex factors and variables to make procedural but believable detailed content.
For Procedural Textures there is a licencable Library for it.
Procedural shift the game load form. Hardisk Optical stream content.
To CPU genreated content.
So less IO but more CPU stressing.
Such Procedural tree for the death star will start with a
sphere
the first branch split to trench and Main weapon.
and then a few dozen to hunderd levels of branches to give a fly through the trench enough fine detail.
U3 is long in development this means they can do PPU specific maps. Like Graw2 did. As the artist have also a large part of that time avaiable.
For single player PPU support is no isue because no online load problem.
Unreal could offer Bot maps or small player PPU Mplay maps.
Well I wait and see.
Graw2 is done in a short time Unreal3 they have some more time to do PPU stuff.
Most will be likely effect Physics. And possibly a bit gameplay Phsyics specific maps.
your post hurts my eyesMulti CPU cores do not scale up in performance in a linear fashion at all, not to mention that the performance from a CPU is absymal compared to a dedicated physics card.
At the moment there is very little reason to buy one except to help ageia stay afloat until some devs come out with some worthy games.
Yea there is a tech opt type game mod on it similar to css. it was hell of fun and in someways better than cssI'am more a
Graw 2
Armed Asauld
R6 las vagas
Unreal 3 is to unreal for me.
Maybe if they do a tactical ops mod for it.
Once I bought UT wenn it's was a old game and run it once after that a put Tactical ops on it I think. Using weapons based on reallife counter parts.
But if the Unreal editor 3 is great I might just buy it for the engine