Jump to content

AMD: "CUDA is doomed"

EChondo

You didn't read it. You didn't watch it. You fail. Check again:

Did you see it that time? 

I'll say it in case you didn't: That video is a 3 year old PhysX, not TressFX, Tech Demo.

PhysX has been in games for a very long time as compared to TressFX. :|

Until we see PhysX-powered hair like this in an actual game, this is a moot point.  Tech demos mean nothing.

Intel Core i7-7700K | EVGA GeForce GTX 1080 FTW | ASUS ROG Strix Z270G Gaming | 32GB G-Skill TridentZ RGB DDR4-3200 | Corsair AX860i

Cooler Master MasterCase Pro 3 Samsung 950 Pro 256GB | Samsung 850 Evo 1TB | EKWB Custom Loop | Noctua NF-F12(x4)/NF-A14 LTT Special Edition

Dell S2716DGR | Corsair K95 RGB Platinum (Cherry MX Brown) | Logitech G502 Proteus Spectrum | FiiO E17 DAC/Amp | Beyerdynamic DT990 Pro

Link to comment
Share on other sites

Link to post
Share on other sites

Until we see PhysX-powered hair like this in an actual game, this is a moot point.  Tech demos mean nothing.

You seem to be confusing what we are talking about here. 

What we are talking about is the technological prowess of PhysX vs TressFX. Not the implementation. 

PhysX could do something that TressFX only recently put into a game 3 years ago. Whether such a feature is implemented or not is irrelevant. PhysX is the older and more advanced of the two. Even if it is gimped on CPU's.

As for actual implementation in a real game...

Alice-ChessboardDestruction-620x.png

That's some lovely hair. And this is from 2011. 

I personally think the best part is about at 0:23-0:28

 

† Christian Member †

For my pertinent links to guides, reviews, and anything similar, go here, and look under the spoiler labeled such. A brief history of Unix and it's relation to OS X by Builder.

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

If some company came out and made a universally compatible GPU based physics solution, that does both hard object/liquid and hair physics, they would have SO much more traction than stuff like TresFX and PhysX. I like the implementation of TresFX in Tomb Raider, if they could put that in all Gaming Evolved titles and get it expanded to more than hair, i'd be sold as a fantastic competitor to PhysX.

"Don't make a girl a promise, if you know you can't keep it" | "Didn't you hear? Spartans never die."

-Gaming- FX 8150 @ 4.2GHz, HD 7870 Tahiti LE 2GB @ 1.1GHz, ASRock 990FX Extreme3, 8GB DDR3 1333 MHz, WIndows 10, Custom Watercooled, NZXT Switch 810

-Audio Editing/Mixing- Mac Mini Late 2014, Core i7 3.5Ghz, 8GB RAM, 2TB Fusion Drive, OSX 10.12 El Captain -Phone- HTC One M7 32GB

Link to comment
Share on other sites

Link to post
Share on other sites

If some company came out and made a universally compatible GPU based physics solution, that does both hard object/liquid and hair physics, they would have SO much more traction than stuff like TresFX and PhysX. I like the implementation of TresFX in Tomb Raider, if they could put that in all Gaming Evolved titles and get it expanded to more than hair, i'd be sold as a fantastic competitor to PhysX.

Huh, sounds like Havok.

Link to comment
Share on other sites

Link to post
Share on other sites

Huh, sounds like Havok.

 

The only thing i know about Havok is that it was in Fallout New Vegas, and that was hardly a physics breakthrough. I mean an engine that is over the top and crazy powerful on GPU's.

"Don't make a girl a promise, if you know you can't keep it" | "Didn't you hear? Spartans never die."

-Gaming- FX 8150 @ 4.2GHz, HD 7870 Tahiti LE 2GB @ 1.1GHz, ASRock 990FX Extreme3, 8GB DDR3 1333 MHz, WIndows 10, Custom Watercooled, NZXT Switch 810

-Audio Editing/Mixing- Mac Mini Late 2014, Core i7 3.5Ghz, 8GB RAM, 2TB Fusion Drive, OSX 10.12 El Captain -Phone- HTC One M7 32GB

Link to comment
Share on other sites

Link to post
Share on other sites

Huh, sounds like Havok.

 

Like a CryEngine 3 or Unreal 4, but for physics. Completely out of proportion but you get what i mean,

"Don't make a girl a promise, if you know you can't keep it" | "Didn't you hear? Spartans never die."

-Gaming- FX 8150 @ 4.2GHz, HD 7870 Tahiti LE 2GB @ 1.1GHz, ASRock 990FX Extreme3, 8GB DDR3 1333 MHz, WIndows 10, Custom Watercooled, NZXT Switch 810

-Audio Editing/Mixing- Mac Mini Late 2014, Core i7 3.5Ghz, 8GB RAM, 2TB Fusion Drive, OSX 10.12 El Captain -Phone- HTC One M7 32GB

Link to comment
Share on other sites

Link to post
Share on other sites

This is the only thing AMD has said recently that i actually agree with, cuda blows and so does physx.

 

If it was open source then hell why not, but seriously nvidia when you buy into a game it runs shit on AMD

because you bribe the game devs to build it for physx (borderlands 2) but when AMD buys into a game

at least it runs decent on every platform. 

CPU: AMD FX-8350 - Noctua NH-D14 | Asus Crosshair V Formula | 16GB Kingston HyperX | Plextor M5P Extreme 256GB SSD

GPU - Asus AMD Radeon 6950 Direct CU II 2GB |

Seasonic 1000 W Platinum | Cooler Master 690 ii Advanced | Asus VE248H + VE248Q

Link to comment
Share on other sites

Link to post
Share on other sites

It is my belief that both technologies need to change or a new implementation needs to emerge that standardizes the GPU driven performance.

 

While OpenCL is a good concept, from what I have seen of it there is one major flaw.  The reliance on passing the code openly to the dll, or else having specific compiles for multiple different platforms/video cards.  From what I know, openCL doesn't have a very good way to protect the openCL portion of the code.  (vs cuda which at least compiles to opcode)

 

CUDA is a good concept, but needs more open standards.  While I give props to nvidia for starting out this, having more open standards imo is the only way to go.

 

Who knows maybe one day we will get a "gpu" type of chip which can execute a hybrid type of openCL and cuda.  (CPU + GPU + heavily paralleled chip = awesome)

0b10111010 10101101 11110000 00001101

Link to comment
Share on other sites

Link to post
Share on other sites

im far frominformed on the cuda vs opencl debate so i just wanted to as you guys who seem to know what your talking about this, will opencl make a difference in newer games. We all know next gen consoles are amd so does that mean game developers will start to take advantage of opencl more if they havent already....people like it when games look amazing and they want great performance so will opencl make a difference in perfirmance and eye candy for gamers in the coming years?

 

-Anubikai

| i5-4670k at stock until I do a custom WC loop | H80i | Asus Matrix Platinum 7970 at stock until I do a custom WC loop  | MSI Z87 GD65 | 1TB CAVIAR BLACK | 250gb Blue Scorpio | HX750 80 PLUS GOLD modular | 16GB INTEL EXTREME EDITION PATRIOT 1866MHZ | Coolermaster Storm Enforcer with two Red 20cm fans intake and exhaust |

Link to comment
Share on other sites

Link to post
Share on other sites

You seem to be confusing what we are talking about here. 

What we are talking about is the technological prowess of PhysX vs TressFX. Not the implementation. 

PhysX could do something that TressFX only recently put into a game 3 years ago. Whether such a feature is implemented or not is irrelevant. PhysX is the older and more advanced of the two. Even if it is gimped on CPU's.

As for actual implementation in a real game...

I personally think the best part is about at 0:23-0:28

 

PhysX may very well be more advanced; I wasn't implying otherwise.  If we're solely comparing technological capabilities of the two, then yes, it's better.  But IMO raw power truly doesn't mean much unless it's utilized.

 

So there's a game with PhysX hair effects, okay, cool, thanks.  Looks pretty good, but TressFX hair effects look just as good, if not better, in Tomb Raider.  Yes, that's comparing a two-year-old game to a five-month-old one, but until a game releases that features PhysX-based hair effects as good as those in the tech demo, I'd hardly call it a win for PhysX.

 

Anyway, that's not so important; the greater concern is the openness and compatibility of these technologies.  PhysX is great, but needs to be able to run on an AMD GPU before it can really flourish and give gamers an overall greater experience.  TressFX is compatible with both green and red, so as it stands now, it'll go farther.

Intel Core i7-7700K | EVGA GeForce GTX 1080 FTW | ASUS ROG Strix Z270G Gaming | 32GB G-Skill TridentZ RGB DDR4-3200 | Corsair AX860i

Cooler Master MasterCase Pro 3 Samsung 950 Pro 256GB | Samsung 850 Evo 1TB | EKWB Custom Loop | Noctua NF-F12(x4)/NF-A14 LTT Special Edition

Dell S2716DGR | Corsair K95 RGB Platinum (Cherry MX Brown) | Logitech G502 Proteus Spectrum | FiiO E17 DAC/Amp | Beyerdynamic DT990 Pro

Link to comment
Share on other sites

Link to post
Share on other sites

You didn't read it. You didn't watch it. You fail. Check again:

Did you see it that time? 

I'll say it in case you didn't: That video is a 3 year old PhysX, not TressFX, Tech Demo.

PhysX has been in games for a very long time as compared to TressFX. :|

 

 

Not sure how I failed... but if that's your view then I have nothing to say, but at least I compared tech that IS being fully used in an actual game engine. What you showed from the Alice game was nothing compared to what was done with TressFX... or the "Tech Demo" you're backing.

MOBO - Asus Sabertooth 990FX R2.0   CPU - AMD Phenom II X6 1055T @ 3.5GHz - Corsair H80i   RAM - 2x Corsair Vengence 4GB DDR3 @ 1666 MHz   

GPU - SAPPHIRE 100362-3L Radeon R9 290 4GB Tri-X OC  Sound - ASUS Xonar DSX  PSU - Thermaltake SMART M Series SP-850M 850W

Link to comment
Share on other sites

Link to post
Share on other sites

Not sure how I failed... but if that's your view then I have nothing to say, but at least I compared tech that IS being fully used in an actual game engine. What you showed from the Alice game was nothing compared to what was done with TressFX... or the "Tech Demo" you're backing.

Like MdX MaxX mentioned, comparing a 2 year old game to a 5 month old one. :|

Actually, I would argue that what is in Alice is comparative to what is in that Tech Demo. In that demo, nothing else is happening. There is just hair, a model for it to be on, and wind. Alice is a full blown game with a lot more going on in it. Flowing hair existing in it alongside all the other Physics effects, imo, is pretty awesome. I would say that Tomb Raider is a more polished version of that and the only reason PhysX hasn't done much more than TressFX in this regard is because it's proprietary and developers basically have to be paid to put it in the game. 

† Christian Member †

For my pertinent links to guides, reviews, and anything similar, go here, and look under the spoiler labeled such. A brief history of Unix and it's relation to OS X by Builder.

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

i want a third company to just pop out of no where and throw down a gpu that sh*ts on nvidia and amd... for a great price.....like undeniably renders their gpus useless.....then i think then the red and green team would get their sh*t together and bring out something that really screams performance.....nvidia seems to have done something great with the 780 as its a vast improvemnt over the 680...just have to see how the 9970 compares to the 7970....i couldnt care less about the little extra particle stuff tressfx or physx gives. even without these technologies games look freaking great and will only get better, whats more important is which implementation can deliver better pefromance to the user mainly in gaming but also in other applications...opencl or cuda?

 

-Anubikai

| i5-4670k at stock until I do a custom WC loop | H80i | Asus Matrix Platinum 7970 at stock until I do a custom WC loop  | MSI Z87 GD65 | 1TB CAVIAR BLACK | 250gb Blue Scorpio | HX750 80 PLUS GOLD modular | 16GB INTEL EXTREME EDITION PATRIOT 1866MHZ | Coolermaster Storm Enforcer with two Red 20cm fans intake and exhaust |

Link to comment
Share on other sites

Link to post
Share on other sites

I hope both "nVidia the way it's meant to be played" and "AMD gaming evolved" will die in horrible death lol. At the end, gaming evolved was response for something that shouldn't exist in the first place.

 

For PhysX, it is not nVidia invention anyway, same as some GPU advances are not AMD invention, but companies that are bought by them (Ageia, ATI).

 

In my opinion, it is really sad to see companies who actually put resources in something that potentially can be revolutionary go under.

 

I never liked nVidia strategy, but i can pretty much vouch that anyone would do same (or similar) thing if they are in same position....

 

At the end, all that nonsense will hurt consumers and companies on the long run, and will potentially open the door for another ones.

 

Oh, and i wouldn't want to see nVidia drivers messing with AMD GPU's, nor other way around, in that case, i wouldn't know who to blame if something doesn't work as it should.

Link to comment
Share on other sites

Link to post
Share on other sites

I agree, OpenCL is much better than CUDA, that's why the Mac Pro uses AMD cards, it is much better for rendering and is open. The problem with CUDA is that it is proprietary and is only exclusive. I wonder how Direct compute stacks up against PhysX.

My PC specs; Processor: Intel i5 2500K @4.6GHz, Graphics card: Sapphire AMD R9 Nano 4GB DD Overclocked @1050MHz Core and 550 MHz Memory. Hard Drives: 500GB Seagate Barracuda 7200 RPM, 2TB Western Digital Green Drive, Motherboard: Asus P8Z77-V , Power Supply: OCZ ZS series 750W 80+ Bronze certified, Case: NZXT S340, Memory: Corsair Vengance series Ram, Dual Channel kit @ 1866 Mhz, 10-11-10-30 Timings, 4x4 GB DIMMs. Cooler: CoolerMaster Seidon 240V

Link to comment
Share on other sites

Link to post
Share on other sites

i want a third company to just pop out of no where and throw down a gpu that sh*ts on nvidia and amd... for a great price.....like undeniably renders their gpus useless.....then i think then the red and green team would get their sh*t together and bring out something that really screams performance.....nvidia seems to have done something great with the 780 as its a vast improvemnt over the 680...just have to see how the 9970 compares to the 7970....i couldnt care less about the little extra particle stuff tressfx or physx gives. even without these technologies games look freaking great and will only get better, whats more important is which implementation can deliver better pefromance to the user mainly in gaming but also in other applications...opencl or cuda?

 

-Anubikai

Thats the Point where at a Time where its not just more Particles, but where it could Change the gameplay and the hole Feeling of the Game.

AMD and Nvidia should get there **** Together and make a PC exclusiv Engine thats what we need to stand out from Next Gen Consoles.

Here PhysX Ocean Simulation :

A PC only GPU Physics Engine could make PC gaming way bigger.

RTX2070OC 

Link to comment
Share on other sites

Link to post
Share on other sites

i find it funny how AMD behaves like a little kid trying to make fun of NVidia. First the Never Settle graphic card destruction and now this. If I was someone looking to buy a GPU atm, this would only show me that AMD is unprofessional and needs attention.

Case: 650D CPU: i5 4670K GPU: GTX 770 Gaming @1306MHz Motherboard: MAXIMUS VI Hero PSU: AX760 CPU Cooler: H100i RAM: 8GB Vengeance Pro @1866MHz Storage: 840 250GB SSD / 2TB Seagate Barracuda 7200.14

Link to comment
Share on other sites

Link to post
Share on other sites

i find it funny how AMD behaves like a little kid trying to make fun of NVidia. First the Never Settle graphic card destruction and now this. If I was someone looking to buy a GPU atm, this would only show me that AMD is unprofessional and needs attention.

*cough* fanboy *cough*

 

From a developer's point of view I agree with AMD, developers want to stand out to as broad an audience as possible and using a proprietary system excludes people which is why OpenCL is better... PhysX is beautiful and so is CUDA programming the way it can easily be convert to from standard C++ programming but still proprietary...

Console optimisations and how they will effect you | The difference between AMD cores and Intel cores | Memory Bus size and how it effects your VRAM usage |
How much vram do you actually need? | APUs and the future of processing | Projects: SO - here

Intel i7 5820l @ with Corsair H110 | 32GB DDR4 RAM @ 1600Mhz | XFX Radeon R9 290 @ 1.2Ghz | Corsair 600Q | Corsair TX650 | Probably too much corsair but meh should have had a Corsair SSD and RAM | 1.3TB HDD Space | Sennheiser HD598 | Beyerdynamic Custom One Pro | Blue Snowball

Link to comment
Share on other sites

Link to post
Share on other sites

*cough* fanboy *cough*

 

From a developer's point of view I agree with AMD, developers want to stand out to as broad an audience as possible and using a proprietary system excludes people which is why OpenCL is better... PhysX is beautiful and so is CUDA programming the way it can easily be convert to from standard C++ programming but still proprietary...

How the hell am I a fanboy? As a matter of fact NVidia doesn't do these things and if they do that would make them seem equally ridiculous. (If they do please link me since I like to have a laugh :) )

Case: 650D CPU: i5 4670K GPU: GTX 770 Gaming @1306MHz Motherboard: MAXIMUS VI Hero PSU: AX760 CPU Cooler: H100i RAM: 8GB Vengeance Pro @1866MHz Storage: 840 250GB SSD / 2TB Seagate Barracuda 7200.14

Link to comment
Share on other sites

Link to post
Share on other sites

How the hell am I a fanboy? As a matter of fact NVidia doesn't do these things and if they do that would make them seem equally ridiculous. (If they do please link me since I like to have a laugh :) )

The never settle thing was amazingly funny but was indeed unprofessional but what they're saying about CUDA is true.

Console optimisations and how they will effect you | The difference between AMD cores and Intel cores | Memory Bus size and how it effects your VRAM usage |
How much vram do you actually need? | APUs and the future of processing | Projects: SO - here

Intel i7 5820l @ with Corsair H110 | 32GB DDR4 RAM @ 1600Mhz | XFX Radeon R9 290 @ 1.2Ghz | Corsair 600Q | Corsair TX650 | Probably too much corsair but meh should have had a Corsair SSD and RAM | 1.3TB HDD Space | Sennheiser HD598 | Beyerdynamic Custom One Pro | Blue Snowball

Link to comment
Share on other sites

Link to post
Share on other sites

The never settle thing was amazingly funny but was indeed unprofessional but what they're saying about CUDA is true.

Yes CUDA will Die, but AMD hasn't even got OpenCL to work that you can use it on a Professional Level.

I want OpenCL and I will change as fast as I can to AMD when they bring it, but first they should get it to Work before making such Statemants.

This is why Blender has Problems with it and there not the only ones:

"NVidia hardware and compilers support true function calls, which is important for complex kernels. It seems that AMD hardware or compilers do not support them, or not to the same extent."

"Note that V-Ray RT at this time also does not support running their full OpenCL kernel on AMD (only an older and simpler version), and Luxrender with OpenCL is also running into kernel size issues when adding more features. So that's a good indication kernel size is the main issue here."

 

http://wiki.blender.org/index.php/Dev:2.6/Source/Render/Cycles/OpenCL

RTX2070OC 

Link to comment
Share on other sites

Link to post
Share on other sites

Yes CUDA will Die, but AMD hasn't even got OpenCL to work that you can use it on a Professional Level.

I want OpenCL and I will change as fast as I can to AMD when they bring it, but first they should get it to Work before making such Statemants.

This is why Blender has Problems with it and there not the only ones:

"NVidia hardware and compilers support true function calls, which is important for complex kernels. It seems that AMD hardware or compilers do not support them, or not to the same extent."

"Note that V-Ray RT at this time also does not support running their full OpenCL kernel on AMD (only an older and simpler version), and Luxrender with OpenCL is also running into kernel size issues when adding more features. So that's a good indication kernel size is the main issue here."

 

http://wiki.blender.org/index.php/Dev:2.6/Source/Render/Cycles/OpenCL

xD OpenCL isn't AMD's own project it's open-source.

Console optimisations and how they will effect you | The difference between AMD cores and Intel cores | Memory Bus size and how it effects your VRAM usage |
How much vram do you actually need? | APUs and the future of processing | Projects: SO - here

Intel i7 5820l @ with Corsair H110 | 32GB DDR4 RAM @ 1600Mhz | XFX Radeon R9 290 @ 1.2Ghz | Corsair 600Q | Corsair TX650 | Probably too much corsair but meh should have had a Corsair SSD and RAM | 1.3TB HDD Space | Sennheiser HD598 | Beyerdynamic Custom One Pro | Blue Snowball

Link to comment
Share on other sites

Link to post
Share on other sites

xD OpenCL isn't AMD's own project it's open-source.

Im aware of that, but there GPU's don't even Work right with OpenCL and there doing like they're blowing Nvidia with Cuda away and that's not the case right now.

Till they get there GPU's working with it they shouldn't make such statemants.

 

RTX2070OC 

Link to comment
Share on other sites

Link to post
Share on other sites

i find it funny how AMD behaves like a little kid trying to make fun of NVidia. First the Never Settle graphic card destruction and now this. If I was someone looking to buy a GPU atm, this would only show me that AMD is unprofessional and needs attention.

If you see any of Nvidia's graphs which they put up against other products you will see what is unprofessional, many of the results are fakes such as the 760 vs 7950 and now the Tegra 4 against other SOCs.  I agree with you, however, on the never settle bundle, i would much prefer to save money and buy just the graphics card than get the games bundled with it as many people would most probably own them games anyway.

My PC specs; Processor: Intel i5 2500K @4.6GHz, Graphics card: Sapphire AMD R9 Nano 4GB DD Overclocked @1050MHz Core and 550 MHz Memory. Hard Drives: 500GB Seagate Barracuda 7200 RPM, 2TB Western Digital Green Drive, Motherboard: Asus P8Z77-V , Power Supply: OCZ ZS series 750W 80+ Bronze certified, Case: NZXT S340, Memory: Corsair Vengance series Ram, Dual Channel kit @ 1866 Mhz, 10-11-10-30 Timings, 4x4 GB DIMMs. Cooler: CoolerMaster Seidon 240V

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×