Mafia 2 will utilize PhysX and 3D Vision

[Manicdan]

^ so your saying i need to spend 100$ on a spare gpu, to enjoy 2 games that combined cost 100$?

[Carfax]

i wasnt trying to start a fight, just trying to point out the perspective that many of us have.

I didn't think you were trying to start a fight

Xoulz on the other hand.....

go look up steam hardware, lots of people have quads. and physics can be easily scaled across threads, so whats their excuse? not worth the investment, or were they paid by nvidia to reduce what they wanted for cpu physics to make physx look better? until the answer to that is known, its all just speculation. if my assumption is right, nvidia is hurting the gaming market. if your assumption is right, then developers are just flat out lazy.

According to Nvidia, devs have access to the Physx code base, so they could have optimized it themselves if they wished, but they didn't.

I doubt it would be worth it though. Modern CPUs don't have the FP capability and bandwidth to run the advanced physics effects seen in games like Mafia 2, without severely bogging down the rest of the game...

btw after all the BS nvidia has pulled with physx, i honestly doubt that 3.0 will make everyone happy. i expect some kind of marketing circus again.

I guess we'll have to see whenever they release it.

[El Mano]

Physx could be perfectly run on the CPU if it was properly coded and not artificially capped.
There's a reason for you not telling us how many threads does this game use and you know it.

Obvious troll is obvious.

http://www.techpowerup.com/127565/AMD_Surpasses_NVIDIA_in_Discrete_Graphics_Shipment s.html


Then shut up you troll.


I'm so bored of this physx discussion. You guys are way beyond fanboy-ism, you are true Nvidia talibans.
JHH is simply stupid: if he capitalized Physx on developers just like any other software creation tool he would still make lots of money from cards and most of all, developers could exploit Physx to its limits since everybody could run it great.

Oh boy, all this hype about 10.000 particles (NOT objects. Objects have several vortex while particles are just dots) and this game is just more DX9 crap.
Wow 10.000 particles, there they are for you to count them one by one. Enjoy.

[Carfax]

^ so your saying i need to spend 100$ on a spare gpu, to enjoy 2 games that combined cost 100$?

You don't have to.

You can run PhysX on one GPU, if it's powerful enough to do both that and rendering.

I use a dedicated card purely because of the performance benefits.

And if you have enough money for a hexcore processor, then I doubt 100 USD will be much of a concern.

[Manicdan]

According to Nvidia, devs have access to the Physx code base, so they could have optimized it themselves if they wished, but they didn't.

I doubt it would be worth it though. Modern CPUs don't have the FP capability and bandwidth to run the advanced physics effects seen in games like Mafia 2, without severely bogging down the rest of the game...

I guess we'll have to see whenever they release it.

you say according to nvidia, which means they will NEVER say something negative. if the contract states they have to maintain a low quality for cpu usage, who would tell us? we might as well just ask everyone trying to be a president, did you ever do drugs or cheat on your wife, cause we know they wont just come out and say it out of kindness

i see no reason NOT to optimize. a developer trying to keep a high quality standard would take that little extra time and optimize physics for all cores, then let there be a few options for medium-high, high, high-ultra, and hardware.

we all know that gpus can do physics better, i will never argue against it. but we also know that cpus built in the last 4 years are all being underutilized in games. physics can fill that gap, but is it laziness, or someones hand in the cookie jar that gives us a sub-par gaming experiences.

[Eastcoasthandle]

and u think steam is more obvious?

Well, he wouldn't want it to be obvious that nvidia has lost a few pecent through steam HW survey. Now would he?

[Manicdan]

You don't have to.

You can run PhysX on one GPU, if it's powerful enough to do both that and rendering.

I use a dedicated card purely because of the performance benefits.

And if you have enough money for a hexcore processor, then I doubt 100 USD will be much of a concern.

my hexi was 125$, i got it because my PII 940 was not much of an overclocker without alot of volts being pushed into it.

i got a second 4850 for 65$ after mail in rebate. i use 4 sticks of DDR2 1GB each for 35$, im a very thrifty shopper. my monitor was a bit of splurge, 1920x1200 S-IPS that ive had almost 5 years now, 750$.

100$ for phsyx
or
100$ for a steam pack of games half a dozen good titles i never played. so for me its easier to just avoid trash that comes out due to physx hardware being "needed" to enjoy a few select titles.

[FischOderAal]

While the cloth simulation looks great, all in all I can live without APEX. Mafia 2 seems to use multicore processors quite well, although not to the fullest extend. My Core i5 750 was barely going above 60 percent per core.

Thanks, it worked!

You're welcome

By the way, now there is a way to disable that annoying 10 minute time limit in the demo. I don't think we're allowed to reveal it in here so just google it. Don't use the so called "patch" it's not working. Just look for the "hex editor" solution. It takes longer but it works. Now I can go around the map and explore the stores or just drive around in different cars. Makes the demo little bit more enjoyable.

The patch worked for me

Which one did you use? I used the one in the attachment.

[madcho]

Steam data is great and interesting.

But watch more in detail.

In GPUs the most DX11 are ATI. 460 and 480 numbers are very low to 5870 and 5770, and even 4870.

There is a lot of nvidia GPUs ? why, because most of nvidias GPUs are old.

The most 3D graphic GPU ( in all gpu even dx9 ) used is 48xx with 7.21% of all steam user. And You can see 57xx and 58xx are 2.98% and 2.82%. ( 7 & 8th in order even it's mainstream and high end, old than less a year. ).

The 460 is not in top 50 ... so ...

I think evrybody that use a less 18 months GPU mainstream or high end is build at 80% by ATI. It's just a reality. And if your not conveinced, just set windows only and check DX11 GPUs.

[Vardant]

http://www.techpowerup.com/127565/AMD_Surpasses_NVIDIA_in_Discrete_Graphics_Shipment s.html


Then shut up you troll.

AMD's discrete graphics products held 51% of the discrete graphics market in the last quarter.

vs

59% user base owning NV cards.

And yet you can't seem to know the difference and you're even defending his moronic statement? Speechless.

I think evrybody that use a less 18 months GPU mainstream or high end is build at 80% by ATI. It's just a reality. And if your not conveinced, just set windows only and check DX11 GPUs.

Can't be. This is the first quarter ATI was better than NV overall IIRC. Maybe it was matched when it comes to high-end sooner, but that would still mean, that the numbers wouldn't change at all.
It is too soon for the GTX 460 to show up there btw. We'll see in a couple more months how successful it really is.

[Manicdan]

Steam data is great and interesting.

But watch more in detail.

In GPUs the most DX11 are ATI. 460 and 480 numbers are very low to 5870 and 5770, and even 4870.

There is a lot of nvidia GPUs ? why, because most of nvidias GPUs are old.

The most 3D graphic GPU ( in all gpu even dx9 ) used is 48xx with 7.21% of all steam user. And You can see 57xx and 58xx are 2.98% and 2.82%. ( 7 & 8th in order even it's mainstream and high end, old than less a year. ).

The 460 is not in top 50 ... so ...

I think evrybody that use a less 18 months GPU mainstream or high end is build at 80% by ATI. It's just a reality.

yeah i do wish steam survey gave out ALOT more info. its incredible as is, but some of us just want an excel sheet with all 100 million computers lol

we do know that intel has like 50% of the gpu market share for units, and we know those are all IGPs, so if we take the intel number alone at 6.22%, its pretty safe to say that 10-14% of all steam users are on IGP computers. we sure cant include them as "gaming" machines just yet (fusion chips coming out i hope puts a swift end to this)

also looking at some of the most popular games, may of them can be run on dx9 gpus from 5-8 years ago. (counter strike 1.6 is the 3rd most popular game on there taking up about 20% of the current gamers!!!)

steam is a great way to look at a very big overall picture, but the details in depth are a little random feeling. its pretty tough to try and determine things like "how many people who play newer graphics intensive games have idle cpu cores, or spare gpu power"

based on an article about a development team that was switching to using steam for their distribution, i think theres alot more to the hardware survey for business use than what we get for free

[FischOderAal]

You keep saying PhysX is irrelevent, yet big titles keep using it. First Metro 2033, then Mafia 2..

What game will be next I wonder?

PhysX is irrelevant because it just adds a few nice effects. It's not a deal breaker. A game with ty graphic and a bad story doesn't turn into a cash cow because of PhysX.

Mafia 2 shows this very good imho. The cloth simulation looks fine, but I can live without it. And as far as the debris is concerned, I'm sure this would look basically the same if you use scripted animations. Those are very small improvements which I certainly won't notice while playing the game.

Imho definitely not worth spending 100 €/$ for a GTS250.

[Eastcoasthandle]

Steam data is great and interesting.

But watch more in detail.

In GPUs the most DX11 are ATI. 460 and 480 numbers are very low to 5870 and 5770, and even 4870.

There is a lot of nvidia GPUs ? why, because most of nvidias GPUs are old.

The most 3D graphic GPU ( in all gpu even dx9 ) used is 48xx with 7.21% of all steam user. And You can see 57xx and 58xx are 2.98% and 2.82%. ( 7 & 8th in order even it's mainstream and high end, old than less a year. ).

The 460 is not in top 50 ... so ...

I think evrybody that use a less 18 months GPU mainstream or high end is build at 80% by ATI. It's just a reality. And if your not conveinced, just set windows only and check DX11 GPUs.

Good observation . Also take a look at the physx results for Mafia 2 in this video. Here I noticed physx most for debris and the movement of the coat tail (which sways unrealistically IMO). Later you see the swaying of the trench coat (or whatever they call it) with physx. But wait, it still sways without physx. There are other examples such as normal physics in game but in all contradicts the talking points for physx. IE: the only real clothing physx being done in the black suite (at the beginning of the video) is just the swaying of the coat tail, etc. Furthermore, the amount of PC processing power overall for such little physx effects show me that the software is inefficient and needs a complete overhaul IMO.

[madcho]

yeah i do wish steam survey gave out ALOT more info. its incredible as is, but some of us just want an excel sheet with all 100 million computers lol

we do know that intel has like 50% of the gpu market share for units, and we know those are all IGPs, so if we take the intel number alone at 6.22%, its pretty safe to say that 10-14% of all steam users are on IGP computers. we sure cant include them as "gaming" machines just yet (fusion chips coming out i hope puts a swift end to this)

also looking at some of the most popular games, may of them can be run on dx9 gpus from 5-8 years ago. (counter strike 1.6 is the 3rd most popular game on there taking up about 20% of the current gamers!!!)

steam is a great way to look at a very big overall picture, but the details in depth are a little random feeling. its pretty tough to try and determine things like "how many people who play newer graphics intensive games have idle cpu cores, or spare gpu power"

based on an article about a development team that was switching to using steam for their distribution, i think theres alot more to the hardware survey for business use than what we get for free

This is the best hardware survey available, and it's for gamers, and casuals.

I think this is a good tool for watch nvidia/ATI market, but not really IGP and AMD/intel market. Because IGP is not a gamer market.

And i think most IGP are tool PC for server ( non graphic ) or SDK. So it's 'gamer' rigs too.

[Eastcoasthandle]

This is the best hardware survey available, and it's for gamers, and casuals.

I think this is a good tool for watch nvidia/ATI market, but not really IGP and AMD/intel market. Because IGP is not a gamer market.

And i think most IGP are tool PC for server ( non graphic ) or SDK. So it's 'gamer' rigs too.

Actually it's not. Steam HW survey may give you some idea of what some steam users have but it doesn't give you the whole picture. Remember, it only relates to those who participate in providing that information to "just steam". It's not a complete representation of the market. There are other sources for that. The best we could ever say is that those who participated may use that hardware. And that's if they've actually submitted/updated their hardware information.

[madcho]

And the cloth on gpu is a joke :

http://www.youtube.com/watch?v=4amtGiPpvhQ

Max payne I, is the same kind of game that mafia 2. Same cloth thing, even i think it's not real time but scripted, but very nice to see.

PhysX was a nice thing with multicore feature before nvidia bought it.
Now it's a marketing bullsh*t.

Mafia II, looks great, even without that useless effects.

[Vardant]

Show me this multicore game before Ageia was bought, that I keep hearing about so much.

[madcho]

Show me this multicore game before Ageia was bought, that I keep hearing about so much.

check cellfactor.

The demo was running as well on multicore than on Ageia PPU with a cheat. Was as smooth and complex too.

[Vardant]

I think the cheat was actually a hoax and the load was considerably lower on the CPU. Without anything, the game barely runs over 30fps on todays quad core computers, IIRC.

And it was far from multicore.

http://www.pcper.com/article.php?aid=403&type=expert

[Carfax]

you say according to nvidia, which means they will NEVER say something negative. if the contract states they have to maintain a low quality for cpu usage, who would tell us? we might as well just ask everyone trying to be a president, did you ever do drugs or cheat on your wife, cause we know they wont just come out and say it out of kindness

Come on man, just because Nvidia says something, it doesn't mean it's automatically false or suspicious..

i see no reason NOT to optimize. a developer trying to keep a high quality standard would take that little extra time and optimize physics for all cores, then let there be a few options for medium-high, high, high-ultra, and hardware.

This is contingent on whether multicore processors can actually handle the advanced levels of physics found in these games. I don't think they can, based on what I saw with the Velocity engine demo; which is perhaps the most optimized physics engine for the CPU..

we all know that gpus can do physics better, i will never argue against it. but we also know that cpus built in the last 4 years are all being underutilized in games. physics can fill that gap, but is it laziness, or someones hand in the cookie jar that gives us a sub-par gaming experiences.

I agree that CPUs aren't utilized to their full potential, but having them run physics calculations that peg them to 100% with far less dynamic objects (like the Velocity engine demo I referred to earlier) isn't exactly a solution.

[Carfax]

And as far as the debris is concerned, I'm sure this would look basically the same if you use scripted animations.

Yes, but scripted animations take extra time and resources to implement.

Using the physics engine not only saves time and money, but looks better and behaves more realistically.

[FischOderAal]

Do you know how long it takes to implement PhysX? Doing it via scripted animations has the advantage that everyone will be able to use them as well.

[Manicdan]

Come on man, just because Nvidia says something, it doesn't mean it's automatically false or suspicious..



This is contingent on whether multicore processors can actually handle the advanced levels of physics found in these games. I don't think they can, based on what I saw with the Velocity engine demo; which is perhaps the most optimized physics engine for the CPU..



I agree that CPUs aren't utilized to their full potential, but having them run physics calculations that peg them to 100% with far less dynamic objects (like the Velocity engine demo I referred to earlier) isn't exactly a solution.

i found this
http://www.viddler.com/explore/HardOCP/videos/35/

4000 objects on an i7 running real time, compared with 10k in mafia 2

i dont see where 10k objects in a single scene go honestly. i saw a bunch of debris which is a few hundred. smoke and dust i can understand being alot, but that scales nicely if they reduced that to about 1/4th the particles.

i think the game would look pretty close to physx if they let people with quad core cpus and better from either brand use it to 100%. give us a version that allows 2-3k particles on cpu, and i would be 100000x happier with nvidia

[Hu1kamania]

I'm with ya Carfax. One of the best CPU based physics engines that I have seen was Infernal's Velocity engine, used for the Ghostbusters game. While it is nice, it still doesn't compare to what can be done with GPU based physics.

A room full of boxes is nice, but only 2500? Impressive by CPU standards but not GPU. Also, factor in volumetric fog and fluid calculations and the CPU is done for.

[Final8ty]

Originally Posted by drunkenmaster
Night and day, yup, performance maybe, the effects themselves are crap. So a bullet hits the "glass wall", and each time the glass debrie effect happens, its basically an explosion, where glass goes off in a spherical explosion with glass going in each direction, including opposite the direction of the bullet.

This is the problem with physx, its wasted on effects that simply aren't realistic. Same with the wood, its not calculated, its not realistic and its really not interacting with the enviroment, if it was it would be interacting with the bullet and the glass would all be shattering on the side opposite the bullets impact(in general) also the glass is mostly shattering into identically sized little pieces and doesn't seem to actually be taking into account the structure or size of the original piece. IE the bullet hits, the glass shatters, the original piece of glass is gone and the debrie "appears" at the site of the bullet and goes in a premade "explosion" type effect where it spreads out evenly in all directions.

Honestly, thats not even anywhere near close to realistic. Its different, because they put effort into making it different, its not the slightest bit more realistic, and considering its a very very poor effect, and really shouldn't be taking more cpu time, so it feels like more Nvidia pointless inclusion to damage performance and cheat the user out of a better experience rubbish.

Lets be honest, those bits aren't any more interactive, and the "performance cost" of truly accurate physics, is keeping track of many individually sized pieces, and calculating each piece of debries movement individually. The debrie is pre rendered, not moving realistically and therefore can't possible be being heavily calculated each time, especially as each glass shattering effect is all but identical to the last, despite bullets not hitting each piece in the same place or same angle, IE theres no way they should be so similar each time.

Just watched a bit more, its really really awful. THe glass effects don't even centre around the bullets explosion and infact seem to be very slightly "delayed", half the "glass explosion" type effects seem to be in the slightly long location, delayed, and again, the glass will explode evenly in all directions.

THe actual glass panes in the next scene are also "exploding", one bullet hits and a huge area of glass shatters into identical pieces each time, with some going up, and some going forward, and some huge fragments going up, which simply wouldn't be possible.

There isn't a sane person that would defend that as night and day difference, when its actually worse in many bits, and completely unrealistic.

Its different, not more realistic and shouldn't cost ANY more performance, if it does, Nvidia is once again absolutely cheating both Nvidia and AMD users. IE add in a new physics api, make the effects different, but exactly no better, call it realistic, its not, kill performance so it only runs well when hardware accelerated.

Its an abomination on the gaming industry to be honest. Its more realistic debrie than Mafia 1, because its the best part of, what a decade newer, nothing to do with physx, and another attempt for Nvidia to buy off a dev to try to prove how good their physx is. If it was remotely realistic, they'd have a point, unfortunately, the glass wall spherical blowout effect is like something from 5-10 years ago itself. The glass pieces of debrie don't even look like pieces of glass it looks so bad.

Ill wait & see if the game is any good.