NVIDIA officially announces PhysX 5.0

Demo is kind of dumb but the FEM stuff is really cool and with both AMD and Nvidia working on it sounds like it's going to be pushed into various games quickly.

 

Just wait till it gets implemented into all of those erotic weeb games. The possibilities are endless!

Spoiler

 

I'm glad to see Nvidia finally doing something new with this technology!

 

Another Dead or Alive beach ball spin-off perhaps or a sequel for Mount Your Friends the co-op puzzle game heh.

Interesting to see PhysX making somewhat of a return again though I suppose it never really left or anything just quieted down a bit after the various earlier showcases they had for it's effects.


EDIT: Well physics and proper interaction of physics in gaming would be a big step for the next-generation beyond 4k this or HDR that much as shiny immediate visuals do tend to be a big thing, might be more feasible too than decent AI so there's a chance at least.

Unreal and other engines are certainly stepping up things in this regard too so maybe we'll see more of this in a few years plus updates on middleware solutions like PhysX here driving it.

 

Personally I hope for better physics in general. Games are progressing visually very well (ray tracing can be a good example, however, ray tracing isn't the best technology for this either. There are other approaches that can further improve visual aspects), however, technically we see either very slow innovation or none at all.

I would rather have games with various unique damage models or new possibilities (imagine a Total War game, where you have a physics based damage model. Catapult lobs a massive rock at a castle and you can actually see damage and destruction on exactly where it has landed.) There can be countless examples, but for me the innovation should be in this direction. Not VR, nor graphics, but physical models.

My initial post in this thread is sort of sarcastic as most likely where we will see initial implementation of this tech will be some boobs jiggle in yet another weeb game.

 

I mean if you got to render physics for single pair of boobs, it's not a big deal. If you got physicals calculation for massive battles, it will destroy CPU and frames.

Look at Battlefield4 for example. It got decent physics happening, destroying covers and such things. Does Novidia bring anything new to the table, that cannot be done in games?

 

This is all lackluster as best. We need new engines that fully support this at the base, instead of being an oversight on-the-side feature.

At the current rate in about 2-3 years we will have CPUs with 20 or even 30 physical cores in mainstream segment, nobody will convince me that a well optimized engine can't use 5-10 of those cores solely for physics calculations. Slap on top of it what ever tech we will have in GPU segment, especially, considering that now both NVIDIA and AMD (+ Intel at some point) have full access to it.

 

Making believable physics is a lot of work and therefore it costs a lot of money. But it's kinda difficult to market it the same way the companies can market nice visuals. For one you can't show anything on screenshots and you need some detailed video showing and explaining it. Now don't get me wrong, I absolutely agree with you. We should have better physics and especially AI, but I think this is the reason why companies do not focus on it.

 

https://gpuopen.com/gaming-product/femfx/

"FEMFX is a multithreaded CPU library for deformable material physics, using the Finite Element Method (FEM)."

In Unreal Engine 4:

 

There are plenty of ways to show some aspect of physics even in a screenshot. Example: still image of a tank shooting at a wall which throws back troops being with + bricks flying all over the place (and what ever else: arms, legs, metal e.t.c.)

That is the future, but only if everyone can use such tech.

 

You talking about next decade, 2030 and beyond.

Nobody will create such game that utilizes different cores, unless we talk about tech demo.
4 cores mainstream will remain for a while, since people still own such hardware and publishers want their games to be played by majority, not geek minority.

https://store.steampowered.com/hwsurvey/cpus/ (2 cores are 26%, 4 cores are more than 50% and growing)
It's a slow shift.

 

I hope more games leverage this I loved games in the past with physx

 

Who shot that 1st video? It looks like the mirror room I got from Usaar33, when I tried to fiddle around in unrealtournament, waay back, doing convertions of h-l, q1, 2 models, "here, if the gun dont look right, you have to move the triangle, here you can see yourself, and try every gun in the game"

 

Sorry, 4C/4T are already struggling. Anthem now utilizes around 45~55% of my 8C/16T. His foreseeing games in 2~3 years using many more cores is correct. Maybe not whole 8C/16T, but thinking that even 4C/8T will do good in those games is misguided.

And if those studios want to differentiate themselves by physics based features, they'll simply put 4C/8T as minimum for 30~60fps experience. And recommend 8C/16T.

 

Probably from this one:



Though it's a 6 months old video.

 

That doesn't outright showcase physics and the image of a game with better visuals is still going to sell better.

I don't think that's true. 4 cores are already insufficient and with release of the next gen consoles they will likely become obsolete.

 

I fully disagree on this. If Intel was at the helm then yes, not with AMD and their chiplet approach. In next 2-3 years situation will change drastically.

That was an example. How you perceive it is up to you, and can be severely limited by your imagination. Point is - physics are the future, graphics alone won't cut it. Marketing has a plethora of ways on how to present that data to a customer.

 

The problem with many physics effects is not doing the effect it's keeping the results around afterwards. So if you realistically blow a hole in a wall and that can happen 1000 different ways depending on how the wall is hit and what hits it then you have to record all the changes to the wall, and if it's multiplayer broadcast that to all the other players. Now expand that to hit the side of a castle with an artillery strike and it's an awful lot of data.