• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Noobmatron - AMD/PhysX?

When you fail to understand that connecting the two is the part that's conjecture .... I really can't be bothered with this crap. Read the link and decide for yourself what you want. And move on.
 
Well I know for a fact tho I can't back it up publicly due to NDAs that nVidia were prepared to open up PhysX to anyone prepared to run CUDA on their hardware - not just AMD - and while I can see why AMD wouldn't want to support CUDA (tho IMO that was a bad decision as CUDA has increasingly become adopted in industry - tho if that had resulted in businesses having the choice of AMD or nVidia hardware for compute purposes I can quite see nVidia pulling something dirty to stop it) - this would have been fully licensed source code as well so nVidia couldn't have screwed them over - other than releasing a "2.0" version and not including it under the previous license.

On a related subject bullet itself has come on a long way, tho I still feel its more orientated towards "movie" like physics and gets in the way a bit for gaming, but playing around with a test program I was able to implement all the effects like in Batman AA - soft bodies like cobwebs, volumetric effects like steam and even deformable objects that changed shape when damaged, etc. and still run over 60fps using the CPU for physics processing :D unfortunatly there seems to be no real effort from AMD to push it :|
 
Forgive my noobity(!), but this seems kinda awful to me though - I mean, in moving to a platform such as the 7000 series AMD will have a substantial performance gap on Nvidia, but it's somewhat disconcerting to know that Nvidia users will have all the pretty effects that make us want to game on PC over consoles anyway whilst the AMD peeps won't.

Or am I missing something? I understand PhysX is only implemented in specific games, so is something else used as the engine for destructibles etc in games like BF3? Is it integrated into the engine like Frostbite 2?
 
This is the reason PhysX is unfortunatly unlikely to ever going to come to much - developers won't use it wholesale to the exclusion of half their audience.

CPU based physics has come on a long way, but widespread useage of fluid dynamics, soft bodies and destructable objects will still bring even the most advanced CPU system to its knees - even on CPU PhysX is much more suited to game physics than pretty much any other physics library but as its usually tacked on as an afterthought even when purely used on the CPU it rarely makes a good impression.
 
Last edited:
BF3 doesn't use PhysX in any shape or form so the physics will be there on whatever GPU you use for rendering.
 
Much appreciated. An interesting perspective on the CUDA stuff as well - I work for a financial services company and we're potentially looking at architectures that can do a lot of SMP for backhaul stuff. Increasingly it appears that lots of companies are doing this kind of thing, and I'd have expected AMD to at least have some kind of foothold in this area.

Thanks for the info though - appreciated. :)
 
DamoTourettes said:
Much appreciated. An interesting perspective on the CUDA stuff as well - I work for a financial services company and we're potentially looking at architectures that can do a lot of SMP for backhaul stuff. Increasingly it appears that lots of companies are doing this kind of thing, and I'd have expected AMD to at least have some kind of foothold in this area.

Thanks for the info though - appreciated. :)
Click to expand...

Amd's gnc 7 series i believe is a step in this direction.
 
DamoTourettes said:
Gotcha... so as a question the destructible pretties I see on BF3 with my GTX580 aren't going to be there when I go 7990/7970 Crossfire when they come out next year?
Click to expand...

Most games don't use physx, and most games that do use physx don't use hardware accelerated physx, and I'd say most of the destructible environment games around at the moment don't use physx.

People keep getting caught up in physx like it was something new and could bring new levels of something or other to gaming. It can't.

It CAN, but often doesn't, bring new levels of accuracy in physics in a game. Problem is, computers that search for oil or predict the weather need ultra accuracy, ALL gaming does NOT need any such level of accuracy. With physx you're paying the performance penalty of accuracy, without the slightest need for it.

In real time you CAN NOT tell if that shattered piece of glass would have fallen a bit to the left, or a bit to the right, or bounced half an inch or 0.4498 inches off the ground. You can't tell the difference between falling with 10g acceleration or 9.81g acceleration.

Games DO NOT need high levels of accuracy and that is ALL physx brings. I think its what, 6-7 years ago now that Ageia were showing off video's of a single screen NOT in a game and ONLY doing the physics for a piece of cloth rippling, we've still not seen this level of detail in game, and won't for years upon years, because you've actually got to render the game and deal with vastly more on screen than a single item. The theoretical applications of physx are pretty widespread, the problem is, gaming won't use pretty much any of them, its a complete waste.

Destructible environments are NOT hard to implement, they are a design issue, not a physics issue. Look at skyrim, how massive it is and how long it would take to make each individual wall and house and castle/keep/whatever, now ask the designer to quadruple his design time by making another version of it that is fully destructible, its simply not feasible. There isn't and hasn't really been anything complicated or impossible in terms of game physics for a decade. its simply about design time, implementation and where to focus the dev's efforts. Fully destructible walls, on a game 1/5th the size or less destructible walls, and a game 5 times as big, the choices are pretty simple most of the time.

The biggest limit in game design, by a MASSIVE margin, is man hours in the game, both because more hours worked increases costs, reduces profit and increases time to market, meaning, when the game comes out with great added features, its also 2 years to late and looks dated.

There are a few key examples of games that try to do too much, end up behind and keep trying to catch up, Duke Nukem Forawful is the best example of that.
 
drunkenmaster said:
Most games don't use physx, and most games that do use physx don't use hardware accelerated physx, and I'd say most of the destructible environment games around at the moment don't use physx.

People keep getting caught up in physx like it was something new and could bring new levels of something or other to gaming. It can't.

It CAN, but often doesn't, bring new levels of accuracy in physics in a game. Problem is, computers that search for oil or predict the weather need ultra accuracy, ALL gaming does NOT need any such level of accuracy. With physx you're paying the performance penalty of accuracy, without the slightest need for it.

In real time you CAN NOT tell if that shattered piece of glass would have fallen a bit to the left, or a bit to the right, or bounced half an inch or 0.4498 inches off the ground. You can't tell the difference between falling with 10g acceleration or 9.81g acceleration.

Games DO NOT need high levels of accuracy and that is ALL physx brings. I think its what, 6-7 years ago now that Ageia were showing off video's of a single screen NOT in a game and ONLY doing the physics for a piece of cloth rippling, we've still not seen this level of detail in game, and won't for years upon years, because you've actually got to render the game and deal with vastly more on screen than a single item. The theoretical applications of physx are pretty widespread, the problem is, gaming won't use pretty much any of them, its a complete waste.

Destructible environments are NOT hard to implement, they are a design issue, not a physics issue. Look at skyrim, how massive it is and how long it would take to make each individual wall and house and castle/keep/whatever, now ask the designer to quadruple his design time by making another version of it that is fully destructible, its simply not feasible. There isn't and hasn't really been anything complicated or impossible in terms of game physics for a decade. its simply about design time, implementation and where to focus the dev's efforts. Fully destructible walls, on a game 1/5th the size or less destructible walls, and a game 5 times as big, the choices are pretty simple most of the time.

The biggest limit in game design, by a MASSIVE margin, is man hours in the game, both because more hours worked increases costs, reduces profit and increases time to market, meaning, when the game comes out with great added features, its also 2 years to late and looks dated.

There are a few key examples of games that try to do too much, end up behind and keep trying to catch up, Duke Nukem Forawful is the best example of that.
Click to expand...

I really do enjoy your posts man. Articulate, concise and insightful as usual. Thanks!
 
drunkenmaster said:
Most games don't use physx, and most games that do use physx don't use hardware accelerated physx, and I'd say most of the destructible environment games around at the moment don't use physx.

People keep getting caught up in physx like it was something new and could bring new levels of something or other to gaming. It can't.

It CAN, but often doesn't, bring new levels of accuracy in physics in a game. Problem is, computers that search for oil or predict the weather need ultra accuracy, ALL gaming does NOT need any such level of accuracy. With physx you're paying the performance penalty of accuracy, without the slightest need for it.

In real time you CAN NOT tell if that shattered piece of glass would have fallen a bit to the left, or a bit to the right, or bounced half an inch or 0.4498 inches off the ground. You can't tell the difference between falling with 10g acceleration or 9.81g acceleration.

Games DO NOT need high levels of accuracy and that is ALL physx brings. I think its what, 6-7 years ago now that Ageia were showing off video's of a single screen NOT in a game and ONLY doing the physics for a piece of cloth rippling, we've still not seen this level of detail in game, and won't for years upon years, because you've actually got to render the game and deal with vastly more on screen than a single item. The theoretical applications of physx are pretty widespread, the problem is, gaming won't use pretty much any of them, its a complete waste.

Destructible environments are NOT hard to implement, they are a design issue, not a physics issue. Look at skyrim, how massive it is and how long it would take to make each individual wall and house and castle/keep/whatever, now ask the designer to quadruple his design time by making another version of it that is fully destructible, its simply not feasible. There isn't and hasn't really been anything complicated or impossible in terms of game physics for a decade. its simply about design time, implementation and where to focus the dev's efforts. Fully destructible walls, on a game 1/5th the size or less destructible walls, and a game 5 times as big, the choices are pretty simple most of the time.

The biggest limit in game design, by a MASSIVE margin, is man hours in the game, both because more hours worked increases costs, reduces profit and increases time to market, meaning, when the game comes out with great added features, its also 2 years to late and looks dated.

There are a few key examples of games that try to do too much, end up behind and keep trying to catch up, Duke Nukem Forawful is the best example of that.
Click to expand...

+1
 
drunkenmaster said:
People keep getting caught up in physx like it was something new and could bring new levels of something or other to gaming. It can't.
Click to expand...

It absolutely can - the API can provide physics waaay beyond what we have seen in anything to date... its unlikely to happen for a number of reasons but its perfectly capable of it.

drunkenmaster said:
It CAN, but often doesn't, bring new levels of accuracy in physics in a game. Problem is, computers that search for oil or predict the weather need ultra accuracy, ALL gaming does NOT need any such level of accuracy. With physx you're paying the performance penalty of accuracy, without the slightest need for it.
Click to expand...

You really don't understand PhysX.

drunkenmaster said:
Destructible environments are NOT hard to implement, they are a design issue, not a physics issue. Look at skyrim, how massive it is and how long it would take to make each individual wall and house and castle/keep/whatever, now ask the designer to quadruple his design time by making another version of it that is fully destructible, its simply not feasible.
Click to expand...

You really really don't understand PhysX... the whole point of using a physics engine like PhysX is that it makes it a lot easier to implement physics wholesale, rather than spending time building specific instances of scripted physics with simple rigid body calculations or even more primitive physics simulations. Granted even with ApeX making something like skyrim with a high level of destructable physics throughout the gameworld would still be a fairly big effort.

I'll quickly link to ApeX as its a lot easier than trying to explain it all in depth myself http://developer.nvidia.com/apex-destruction hopefully that gives you enough of an idea without me having to make 2 pages worth of posts explaining PhysX.

By using these kind of physics engines doing something like: http://www.youtube.com/watch?v=9lCkB77it-M&feature=player_embedded perfectly possible on a large scale, circumventing the design issues that would exist with traditional approaches.

EDIT: While you post isn't without its merit, you seem to be entrenched in an extremely outdated view of PhysX that hasn't really been true since the Novodex days - super accurate simulation was given up a long time ago for what works best in a game environment.
 
Last edited:
Rroff said:
It absolutely can - the API can provide physics waaay beyond what we have seen in anything to date... its unlikely to happen for a number of reasons but its perfectly capable of it.



You really don't understand PhysX.



You really really don't understand PhysX... the whole point of using a physics engine like PhysX is that it makes it a lot easier to implement physics wholesale, rather than spending time building specific instances of scripted physics with simple rigid body calculations or even more primitive physics simulations. Granted even with ApeX making something like skyrim with a high level of destructable physics throughout the gameworld would still be a fairly big effort.

I'll quickly link to ApeX as its a lot easier than trying to explain it all in depth myself http://developer.nvidia.com/apex-destruction hopefully that gives you enough of an idea without me having to make 2 pages worth of posts explaining PhysX.

By using these kind of physics engines doing something like: http://www.youtube.com/watch?v=9lCkB77it-M&feature=player_embedded it perfectly possible on a large scale, circumventing the design issues that would exist with traditional approaches.

EDIT: While you post isn't without its merit, you seem to be entrenched in an extremely outdated view of PhysX that hasn't really been true since the Novodex days - super accurate simulation was given up a long time ago for what works best in a game environment.
Click to expand...

+1000
 
Drunkenmaster is wrong on many points. I have no time to delve into details but its not merely a matter of accuracy and There is a real need fOr hardware accelerating physx in games. But as rrOff observed this ont gain enough traction for game physx when the API and thus the community=market is split in two. On an aside age of Conan uses an interesting new approach with serversidee physx so all can benefit


Most typos cost of damn iPhone and fudging Autocorrect
 
love the butthurt amd fanboys that believe amd is a saint company when they are just as bad as anyone else plus they NEVER get anything done just bullsh!t around waiting for others to do the job for them with their so called "open" apis.

xsistor proved how ignorant you all are so just apologise and leave the thread like gentlemen.

this section needs another clear-out starting with the biggest troll of the forum in disguise (drunkenmaster) as he is now attracting followers that sit and read his walls of text full of fanboyism written in a way that attracts people with lower IQ.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom