Monday, October 5th 2015

Microsoft Acquires Havok Physics from Intel

Microsoft acquired Havok Physics, the industry's most popular in-game physics API, from Intel. Microsoft intends to add Havok's IP to its existing tools and platforms, including DirectX 12, Visual Studio, and Azure. Havok will continue to remain accessible to all its existing licensees and partners, including Activision, EA, Ubisoft, Nintendo, and Sony. It currently features in more than 600 AAA game titles across major platforms, such as the PC, PlayStation, Wii, and Xbox.
Source: Microsoft
Add your own comment

38 Comments on Microsoft Acquires Havok Physics from Intel

#1
Jack1n
Why would Intel sell IP?
Posted on Reply
#2
RejZoR
Because they relaized they can't really do anything with it. Where Microsoft can do plenty...
Posted on Reply
#3
john_
Nvidia wanted to lock PhysX - and I don't mean lock it on running on their GPUs, but lock it on running only when their GPU is primary. I would love to see Havok implemented in DirectX 12 and PhysX go down the toiler.
Posted on Reply
#4
HumanSmoke
john_
Nvidia wanted to lock PhysX - and I don't mean lock it on running on their GPUs, but lock it on running only when their GPU is primary. I would love to see Havok implemented in DirectX 12 and PhysX go down the toiler.
Not all PhysX is GPU based (rather obvious since it is used in conjunction with AMD-based consoles) Developers are free to use it for non-commercial use with a royalty paid for commercial use, so CPU PhysX and Havok aren't actually that different. Both are licensed middleware.
The second part of the equation is that MS will implement Havok via cloud-based gaming. Microsoft recently announced that their cloud service, Azure will utilize Nvidia's GRID virtualization platform (just as Amazon's AWS already has). Given that MS has a stake in Havok, and PhysX already has a pervasive ecosystem(UE4-based in particular), I'd think that Nvidia would have a piece of the pie either way.
Posted on Reply
#5
Recus
john_
Nvidia wanted to lock PhysX - and I don't mean lock it on running on their GPUs, but lock it on running only when their GPU is primary. I would love to see Havok implemented in DirectX 12 and PhysX go down the toiler.
Are you Sony Xperia S brother? I mean, know nothing and want monopoly.

Also knowing MS Havok mass destruction will be locked on XOne cloud.


Physx-Windows, PS3, X360, Wii, Android, PS4, XOne, WiiU.
Posted on Reply
#6
john_
Recus
Are you Sony Xperia S brother? I mean, know nothing and want monopoly.

Also knowing MS Havok mass destruction will be locked on XOne cloud.


Physx-Windows, PS3, X360, Wii, Android, PS4, XOne, WiiU.
I don't mind with a monopoly that will work on every hardware... brother. I do mind with a monopoly that dictates what hardware I will put in my system and punishes me even if I am a customer of that monopoly for not being loyal enough. Also if Havok is locked at Xbox One what kind of monopoly are you talking about? A MS monopoly on the Xbox One platform?

The only one who knows nothing and loves monopolies is that guy that looks at you when looking at a mirror.... brother.
Posted on Reply
#7
john_
HumanSmoke
Not all PhysX is GPU based (rather obvious since it is used in conjunction with AMD-based consoles) Developers are free to use it for non-commercial use with a royalty paid for commercial use, so CPU PhysX and Havok aren't actually that different. Both are licensed middleware.
The second part of the equation is that MS will implement Havok via cloud-based gaming. Microsoft recently announced that their cloud service, Azure will utilize Nvidia's GRID virtualization platform (just as Amazon's AWS already has). Given that MS has a stake in Havok, and PhysX already has a pervasive ecosystem(UE4-based in particular), I'd think that Nvidia would have a piece of the pie either way.
No it's not, it is only killing performance when not running on an Nvidia GPU. Developers are free to use it for non commercial use you say. Like NOT in games? Interesting, and pointless for us.
The second part of the equation will either make Havok something that we do care about, or something that we don't. As for UE4, if Microsoft makes Havok part of the Dx12 ecosystem and a viable option for developers, you can be certain that UE4 will be updated to use it. The question now is, if there was any chance for Nvidia to get the contract for the next XBOX, why would MS want Havok?
Posted on Reply
#8
HumanSmoke
john_
No it's not, it is only killing performance when not running on an Nvidia GPU. Developers are free to use it for non commercial use you say. Like NOT in games? Interesting, and pointless for us.
The point is, that for non-Nvidia GPU systems, both Havok and PhysX run on CPU. Both require a paid license for commercial use (Havok required a paid royalty for any game that retails at $10 or more as Intel IP. The link I posted earlier surmises that the license fee will increase now that MS has Havok).
john_
The question now is, if there was any chance for Nvidia to get the contract for the next XBOX, why would MS want Havok?
Why would Nvidia even tender for a console??? That really has no bearing whatsoever on what is obviously taking place - which is MS and Nvidia teaming up to promote cloud-based gaming. MS providing the service, consumer hardware, and an SDK, and Nvidia providing the cloud hardware and an in-place software ecosystem.
MS want Havok because it is a drawcard for cloud gaming - I thought that was made relatively clear. They can gain IP revenue at the same time, just as Intel did before them. It's called investment.
john_
The second part of the equation will either make Havok something that we do care about, or something that we don't.
I doubt it will be anything earth shattering, any more than PhysX is. It will just be another incentive dangled in front of gamers to get them integrated further into MS's business model.
john_
As for UE4, if Microsoft makes Havok part of the Dx12 ecosystem and a viable option for developers, you can be certain that UE4 will be updated to use it.
Probably. But I'm guessing most developers are just lazy and also realize that an SDK is just an email away, so it will ultimately come down to who offers the devs the best deal and best time to market. MS will probably offer some incentives to increase the attractiveness of Azure. But the same is true for Microsoft as is the case for Nvidia I made in the previous post. When MS integrate Havok into Azure, then I'm pretty sure Nvidia will do the same for PhysX (I'd be surprised if it wasn't already in development)- after all, Azure is utilizing Nvidia GPU hardware...and MS won't care too much whether the game in question uses Havok or PhysX - just so long as it benefits their cloud-based gaming program- which is by far the main focus. It is what is generally described as a mutually beneficial relationship. Both companies have a vested interest in the technology as a growth market.
Posted on Reply
#9
Prima.Vera
Smart move. Wanna bet that in a couple of month you will see marketing slides for the new games something like, "bla bla bla, only on XBone with realistic physics and bla bla bla..."
Posted on Reply
#10
HumanSmoke
Prima.Vera
Smart move. Wanna bet that in a couple of month you will see marketing slides for the new games something like, "bla bla bla, only on XBone with realistic physics and bla bla bla..."
That's exactly what will happen. Microsoft will seduce the console crowd with smoke particle effects and mirrors to entice a few more into the glories of Gaming with Latency, and once MS starts to reach an inflection point, it adds legitimacy to their partners (Nvidia) own cloud gaming program. One hand washes the other - American business in action.
Posted on Reply
#11
truth teller
here is to hoping that this will become an free api module of directx (like direct3d, dirctinput, directsound, etc), having a directcollision or something, even if not gpu accelerated by default, would certainly help.
and if this turns out to be a flop, developers can always start to adopt vulkan + bullet (at least for the pc crowd)
Posted on Reply
#12
john_
HumanSmoke
The point is, that for non-Nvidia GPU systems, both Havok and PhysX run on CPU. Both require a paid license for commercial use (Havok required a paid royalty for any game that retails at $10 or more as Intel IP. The link I posted earlier surmises that the license fee will increase now that MS has Havok).
Yes, but Havok under Intel wouldn't run on a GPU and PhysX under Nvidia, would never run efficiently on a CPU. But on MS's hands, we can hope that it would run everywhere. Now if they decide to limit it on XBOX, OK, the PC will be wide open for anyone with a new Physics engine to take on PhysX. I don't see something bad with it. Do you?
Why would Nvidia even tender for a console??? That really has no bearing whatsoever on what is obviously taking place - which is MS and Nvidia teaming up to promote cloud-based gaming. MS providing the service, consumer hardware, and an SDK, and Nvidia providing the cloud hardware and an in-place software ecosystem.
MS want Havok because it is a drawcard for cloud gaming - I thought that was made relatively clear. They can gain IP revenue at the same time, just as Intel did before them. It's called investment.
Nvidia would love to have consoles. We can of course take for granted their excuses about low margins and stuff, but a company that tries to make everything and everywhere proprietary would LOVE to have the consoles even for ZERO income. But they didn't have x86 cores, so they couldn't produce an APU with enough CPU power and Windows PC compatibility. Intel had the cores, didn't had the GPU. Cloud based gaming is nice, but you can't replace the local computer. Not yet. Havok can also be a drawcard for DX12, one more advantage over Vulkan.
I doubt it will be anything earth shattering, any more than PhysX is. It will just be another incentive dangled in front of gamers to get them integrated further into MS's business model.
It doesn't have to be earth shattering, just people see it in a much more positive way than PhysX. Win7 wasn't earth shattering compared to Vista. The proprietary and locked nature of PhysX and the need for Nvidia to make it look more important than it is, didn't helped it really. People who don't want to be locked to a hardware manufacturer and people who are fed up with paper debris and smoke everywhere, wouldn't mind to see PhysX to go away. A Havok engine that uses the power of both CPU and GPU and doesn't ruin the games by putting unnecessarily too many Physics effects on screen just to impress, will be accepted instantly.
Probably. But I'm guessing most developers are just lazy and also realize that an SDK is just an email away, so it will ultimately come down to who offers the devs the best deal and best time to market. MS will probably offer some incentives to increase the attractiveness of Azure. But the same is true for Microsoft as is the case for Nvidia I made in the previous post. When MS integrate Havok into Azure, then I'm pretty sure Nvidia will do the same for PhysX (I'd be surprised if it wasn't already in development)- after all, Azure is utilizing Nvidia GPU hardware...and MS won't care too much whether the game in question uses Havok or PhysX - just so long as it benefits their cloud-based gaming program- which is by far the main focus. It is what is generally described as a mutually beneficial relationship. Both companies have a vested interest in the technology as a growth market.
We will have to wait and see. But yes, some developers do like the cake full and eat it too. The money from customers and the money from a good deal with a manufacturer. As for MS, it is doing business with everybody and that's good. I like their attitude lately where they don't try to favor any of the hardware competitors.
Posted on Reply
#13
FordGT90Concept
"I go fast!1!11!1!"
If Microsoft rolls all of Havok into DirectX, PhysX will be relegated to the history books because developers won't use it (unless NVIDIA pays them to). That's what should be done but as I said in the other thread, I have my doubts Microsoft wants to cut off profits from Havok by making it a free product. Ultimately the question becomes what will Microsoft include in DirectX and what will they exclude for profit? I don't know.
Posted on Reply
#15
Ikaruga
john_
Nvidia wanted to lock PhysX - and I don't mean lock it on running on their GPUs, but lock it on running only when their GPU is primary. I would love to see Havok implemented in DirectX 12 and PhysX go down the toiler.
I love to read these Nvidia hater comments, about pretty much everything Nvidia does from physx to 3D glasses. There was no widespreadly used accelerated physics engine for PC gamers, so Nvidia put in time and effort and decided to bring life to their "own" solution using the stuff Ageia had to offer, and yes, they made it with their hardware in mind. They did not chop of the hands of the guys at AMD or Intel or Microsoft, etc.. to do the same. How is that a bad thing is simply beyond my comprehension.
Posted on Reply
#16
FordGT90Concept
"I go fast!1!11!1!"
Because TWIMTBP. They coerce developers into using their proprietary software that only accelerates on their own hardware. NVIDIA has NVIDIA's interests in mind, no one else's (especially gamers).
Posted on Reply
#17
HumanSmoke
Ikaruga
I love to read these Nvidia hater comments, about pretty much everything Nvidia does from physx to 3D glasses. There was no widespreadly used accelerated physics engine for PC gamers, so Nvidia put in time and effort and decided to bring life to their "own" solution using the stuff Ageia had to offer, and yes, they made it with their hardware in mind. They did not chop of the hands of the guys at AMD or Intel or Microsoft, etc.. to do the same. How is that a bad thing is simply beyond my comprehension.
Some people just prefer to reduce everything down to a simplistic level. Unfortunately, it does tend to trample over the facts. Case in point
john_
Yes, but Havok under Intel wouldn't run on a GPU and PhysX under Nvidia, would never run efficiently on a CPU.
Two points made. Two points fundamentally incorrect.
1. Intel introduced GPU accelerated Havok as HavokFX after aquiring Havok specifically to entice ATI and Nvidia to get on board. Intel opened the door. Developers decided to be lazy (whats new). ATI made big noises. AMD acquired ATI and canned it's involvement. Nvidia acquired Ageia because the impetus had been lost with HavokFX and Nvidia needed GPU physics for their GPGPU ecosystem and saw the potential for marketing GPU game physics that HavokFX was intended to bring. Intel had no GPU hardware and without AMD on board lost interest very quickly.
2. "PhysX under Nvidia would never run efficently on a CPU". CPU PhysX has always played second fiddle to GPU PhysX. Hardly surprising, since the company makes GPUs, not CPUs, but if you really wanted to find out about the coding you'd read the articles by the original NovodeX/PhysX coder - but then, that would mean revising a lot of preconceived bullshit. And that just doesn't fit the paradigm that some people desperately cling to.
john_
Nvidia would love to have consoles.
Yet you said it yourself. Nvidia doesn't have the tools to provide console hardware at the moment, so what's the point of even bringing it into a discussion that has absolutely nothing to do with the article? The console hardware is immaterial to Havok, or PhysX, or Bullet, or any game engine specific physics engine - ALL OF THEM RUN ON THE SAME X86 PROCESSORS.
john_
It doesn't have to be earth shattering, just people see it in a much more positive way than PhysX.
No, its only people like you that see everything as World vs Nvidia. The industry is built on mutually beneficial partnerships. If this weren't the case, why did Microsoft choose to go with Nvidia GPU hardware and software for Azure rather than integrate an AMD solution which already exists in their console?
john_
Cloud based gaming is nice, but you can't replace the local computer. Not yet.
Exactly. NOT YET. Some companies move the industry, and some companies follow it around like a flock of seagulls hoping for a tidbit to float their way after the ship has sailed past. Microsoft and Nvidia are where they are in their respective markets because they are proactive.

Unless MS roll Havok into D3D, which is very, very, very unlikely (I'm guessing it's just another piece of the ecosystem for MS's cloud platform. Big picture beats narrow focus every time), Havok will remain proprietary middleware - it is no different to CPU PhysX or any game engine specific physics implementation. It is actually a pretty safe bet that if MS alter anything about Havok's implementation - whether licensing or functionality, that affects Nvidia's business model, the latter will rejig PhysX/Flex to accomodate the change - it isn't any different to how companies reassess in the face of competition - but until that happens it is pretty much moot
john_
people who are fed up with paper debris and smoke everywhere, wouldn't mind to see PhysX to go away
Except that some people - the ones who don't have such a narrow-minded view of the industry, realize that gaming PhysX is (and has always been)- just one facet of what the technology is used for. For example (PDF) its scientific use for Tesla/Quadro deployments. Even in a gaming-only scenario, developers are happy to incorporate it if it doesn't cost them anything to do so - that is unlikely to change if Havok requires extra coding by the developers to implement. i.e. it is no different to what has been happening in the gaming industry since its inception. If this weren't the case, AMD and Nvidia wouldn't be sponsoring game titles.
FordGT90Concept
Because TWIMTBP. They coerce developers into using their proprietary software that only accelerates on their own hardware. NVIDIA has NVIDIA's interests in mind, no one else's (especially gamers).
Nvidia paid $150 million for Ageia to acquire PhysX. How many companies buy IP just to give it away? Does this seem like a sound business model to you?
Nvidia offers licensing, and the only other GPU vendor declined to do so- repeatedly. The first time supposedly because they put their faith in HavokFX being their future. AMD came along and swallowed up ATI and promptly canned everything - including ATI's gaming development program.

So yeah, Nvidia is a big ol' nasty for charging a licensing fee (just as Intel/MS are doing with Havok incidentally), but I've never heard a word from the people railing against Nvidia's proprietary model actually calling out the company that helped put Nvidia in the cat bird seat to start with. AMD crushed not only PhysX's only credible alternative back in the day, but elevated TWIMTBP by cutting off their own inherited GITG gaming development program.
Posted on Reply
#18
FordGT90Concept
"I go fast!1!11!1!"
It's not the software I have a problem with, it's the fact they didn't port it to DirectCompute and deliberately make it exclusive to CUDA cores). I suspect that's exactly what Microsoft will do: make Havok run on DirectCompute and license the code to developers. That makes it hardware agnostic like PhysX should have been and Ageia likely would have gone in that direction if they weren't bought out (February 13, 2008) before DirectCompute debuted (July 28, 2008).
Posted on Reply
#19
HumanSmoke
FordGT90Concept
It's not the software I have a problem with, it's the fact they didn't port it to DirectCompute and deliberately make it exclusive to CUDA cores). I suspect that's exactly what Microsoft will do: make Havok run on DirectCompute and license the code to developers. That makes it hardware agnostic like PhysX should have been and Ageia likely would have gone in that direction if they weren't bought out (February 13, 2008) before DirectCompute debuted (July 28, 2008).
Nvidia (nor any company that has a market locked down without competition) wouldn't leverage anything outside its own ecosystem while it is still building it - remember that even AMD locked down Mantle until its advantage in the market evaporated with the advent of DX12. Also remember that GPU physics pretty much required a more flexible pipeline to do anything other than eye candy - at least as far as gaming scenarios are concerned, which is why DC really only took off with actual DirectX11 adoption rather than the 2008 preview.
For the record, Nvidia are on record as having PhysX/Flex implementation in DC on their roadmap. Nvidia have in the past made good on their roadmaps for PhysX ( multithreading support, SSE and AVX support etc.) so I don't see any reason to doubt it will happen - I generally only reserve pessimism for company roadmaps if they don't follow through. With Azure and AWS likely to give the platform a wider audience, I don't think the company would pass up the opportunities it could offer, especially as the tools, software/API base, and revenue opportunities are now all in place.
Is FLEX purely GPU accelerated library, or will it support CPU execution? Is it plausible to see FLEX ported to OpenCL or DirectCompute?

Miles Macklin (physics programmer at NVIDIA and lead-developer of the FLEX system): Right now we have a CUDA implementation and a DirectCompute implementation is planned. We are considering a CPU implementation.

I have also built FLEX for Linux (Ubuntu 12.04 64bit) and it works great, in some cases it is faster than Windows.
Posted on Reply
#20
john_
Ikaruga
I love to read these Nvidia hater comments, about pretty much everything Nvidia does from physx to 3D glasses. There was no widespreadly used accelerated physics engine for PC gamers, so Nvidia put in time and effort and decided to bring life to their "own" solution using the stuff Ageia had to offer, and yes, they made it with their hardware in mind. They did not chop of the hands of the guys at AMD or Intel or Microsoft, etc.. to do the same. How is that a bad thing is simply beyond my comprehension.
Nvidia had only to keep their drivers unlocked and do not dictate what hardware you should have in your PC.
Let me give you an example.
What would have been your reaction if Intel's platform was promoted as only compatible with Intel SSDs? How about if Intel was saying "I cannot warranty the performance and stability of any other company's SSDs in my platform, so anything else than Intel SSDs, like your Kingston, should run in IDE mode. What would have been your reaction then? What if there was a patch showing that every SSD could run without problems on an Intel platform and Intel's excuses where just BS to force you buy their SSDs? How about if their own SSDs where going back to IDE mode if another manufacturer's SSD was in the system, even as a secondary drive?

That's what Nvidia is doing. I don't care if PhysX runs only on Nvidia GPUs. But disabling it when the primary GPU is an AMD is just BS. Disabling it because USB monitor's drivers are treated by Nvidia's drivers as a competitor's GPU in the system is not just BS, it is also hilarious. Yes they do that. Nvidia could leave PhysX unlocked so people could buy midrange Nvidia cards as PhysX/CUDA cards. But they don't want to be secondary hardware in a PC. So, they lock it. You want it? You have to pay more money to Nvidia than their competitors. In fact they forbid you to pay ANY money to their competitors if you want hardware Physics. You could give $700 for a 295X2 as a primary card and $330 for a 970 to use for PhysX and CUDA. No, no NO. You will be punished. You payed them full price, but no PhysX for you, no CUDA for you. And you seem to love it. To enjoy reading people's comments who do not like the idea of kissing Nvidia's behind. Nvidia could have left PhysX unlocked and just give you warnings that it can not support nd will not be responsible for any kind of problems in a system where AMD cards are primary or secondary. They chose to lock it. There are patches there that prove you can use PhysX with AMD as a primary card.
Posted on Reply
#21
FordGT90Concept
"I go fast!1!11!1!"
HumanSmoke
Nvidia (nor any company that has a market locked down without competition) wouldn't leverage anything outside its own ecosystem while it is still building it - remember that even AMD locked down Mantle until its advantage in the market evaporated with the advent of DX12. Also remember that GPU physics pretty much required a more flexible pipeline to do anything other than eye candy - at least as far as gaming scenarios are concerned, which is why DC really only took off with actual DirectX11 adoption rather than the 2008 preview.
For the record, Nvidia are on record as having PhysX/Flex implementation in DC on their roadmap. Nvidia have in the past made good on their roadmaps for PhysX ( multithreading support, SSE and AVX support etc.) so I don't see any reason to doubt it will happen - I generally only reserve pessimism for company roadmaps if they don't follow through. With Azure and AWS likely to give the platform a wider audience, I don't think the company would pass up the opportunities it could offer, especially as the tools, software/API base, and revenue opportunities are now all in place.
All games that implemented Mantle also had alternative hardware acceleration options like Direct3D 11. All Mantle gained were some more frames per second. NVIDIA users weren't out much. There's some games (Crazy Machines 2 DLC comes to mind) that are completely unplayable without an NVIDIA card.

It should have been priority number one for NVIDIA to port PhysX to DirectCompute. Now they're likely only doing it because they know PhysX will be dead in the water when the inevitable happens (DirectCompute physics engine). I'm sure we're going to see NVIDIA rush that roadmap now before Microsoft slams the door on them.

Your link only talks about FleX. Havok/PhysX at its core is rigid body. If there's a link that says NVIDIA is considering porting the entire PhysX library to DirectCompute, it's not that.
Posted on Reply
#22
HumanSmoke
john_
They chose to lock it. There are patches there that prove you can use PhysX with AMD as a primary card.
Well, I'm sure Nvidia would sell a few cards to a handful of people, but I'd also guess that the vast majority of the minuscule AMD users that would consider using an Nvidia card would do like just everybody else does, and just source a secondhand card for PhysX processing. Most people with a dedicated PhysX card don't go out and buy them, they either re-purpose one of their older cards or just buy a cheap secondhand board. The former is much more likely than the latter. I can point you to dozens of threads concerning AMD owners who would use a dedicated PhysX card - but only if it meant they didn't give a cent to Nvidia to achieve it. If you can find me some threads where AMD card owners would be happy to buy a new Nvidia card as a dedicated PhysX board I could see the viewpoint as valid, but it just doesn't occur
john_
You could give $700 for a 295X2 as a primary card and $330 for a 970 to use for PhysX and CUDA.
Can you find anything other than isolated examples of anything similar occurring even when the primary is a comparable priced Nvidia primary card where the option IS available? Most people realize that a mainstream card ( GTX x50/x60) is ample for PhysX - and that a dedicated PPU, while beneficial (it does make a PhysX enabled game run smoother) doesn't scale with core count/ability like a GPU does for gaming in general.
So, why bother? AMD don't care enough to license or maintain driver compatibility. Nvidia don't care enough about PhysX to give it away. And AMD card owners don't care enough to pay to use it....and we have at least one poster here (and he's not alone) stating that PhysX is useless anyway. Strangely enough the person screaming for AMD users to be able to use PhysX is an AMD user who also thinks PhysX doesn't offer anything :rolleyes: go figure...
john_
people who are fed up with paper debris and smoke everywhere, wouldn't mind to see PhysX to go away.
Hardly a situation that warrants the effort. The ONLY upside is PhysX would have a wider audience - but it is only a single selling point. If people don't buy a shitty game, they aren't going to change their mind because the shitty game has PhysX support.
FordGT90Concept
All games that implemented Mantle also had alternative hardware acceleration options like Direct3D 11. All Mantle gained were some more frames per second. NVIDIA users weren't out much.
Firstly, I didn't quantify how much Nvidia (or Intel for that matter) lost out because Mantle wasn't available to them. All I stated was that companies in general don't release IP for free and deprive themselves of revenue/PR unless they have to.
I ask what colour is a Lemon, and you tell me a Potato isn't a fruit. A+ for obfuscation.
It should have been priority number one for NVIDIA to port PhysX to DirectCompute.
And yet Havok - the supposed subject of this thread- has been around 4 years longer than PhysX... and also isn't DC ?
FordGT90Concept
Now they're likely only doing it because they know PhysX will be dead in the water when the inevitable happens (DirectCompute physics engine). I'm sure we're going to see NVIDIA rush that roadmap now before Microsoft slams the door on them.
Very probably. When your product is under threat, you adapt or die. Not a particularly revolutionary concept.
FordGT90Concept
Your link only talks about FleX. Havok/PhysX at its core is rigid body. If there's a link that says NVIDIA is considering porting the entire PhysX library to DirectCompute, it's not that.
I didn't say Nvidia was planning to port the entire PhysX library to DC, so it may be an idea not to try to put words into my mouth. Flex is designed to supercede PhysX, so I doubt Nvidia would worry about porting PhysX in any case. The expense wouldn't be worth the effort - porting isn't a trivial matter, which is why I supplied links to Pierre Terdiman's PhysX site.
Posted on Reply
#23
FordGT90Concept
"I go fast!1!11!1!"
I can't name anyone that has a dedicated PhysX card. That was a fad half a decade ago. Not enough games use it and what they use it for isn't significant enough to bother.

PhysX implementations are largely left to eye candy because of the aforementioned problem with games that rely on PhysX as a function of the game. Because PhysX is isolated to one hardware vendor (remember, Intel HD users are hung out to dry too), it is unsuitable as a core feature because of the number of support tickets it generates. As far as PhysX is concerned, NVIDIA is NVIDIA's own worst enemy. Havok is far more popular where the physics simulations actually matter (vehicles, fluids, etc.).
Posted on Reply
#24
HumanSmoke
FordGT90Concept
I can't name anyone that has a dedicated PhysX card. That was a fad half a decade ago. Not enough games use it and what they use it for isn't significant enough to bother.
I have had them in the past, but only because I had the card sitting around. Generally it isn't worth the power it takes to run it. I've had PhysX cards, so I'm pretty confident that my argument that opening PhysX to AMD-based systems wouldn't add to Nvidia's bottom line. I've had AMD based systems alongside Intel ones and plenty of graphics cards sitting around for new owners or to go into the junk pile (I wonder if Jay from Jayztwocents wants to pay shipping? I can send him a box of AMD/ATI cards for the cost of shipping only since he seems to have trouble finding them). I never once bothered with a PhysX hack to enable it for use with my AMD systems. Apart from the limited use, the next driver would just mean I'd have to go through the whole routine again.
Hard to argue the need for PhysX when the guy arguing of its wider availability also claims how useless it is. :rolleyes:
FordGT90Concept
PhysX implementations are largely left to eye candy because of the aforementioned problem with games that rely on PhysX as a function of the game.
Mostly agree. I'm not a PhysX advocate in its gaming sense (scientific utilization is a different situation), but I'm also not blind to the marketing involved or the wider industry implications either. This thread was supposed to be about MS buying Havok - which makes sense from a cloud gaming, licensing, and MS's virtualization points of view. But quickly devolved into an Nvidia bashing thread at the earliest opportunity - which turned out to be post #4 in this instance, and focused entirely upon a tangential technology that has little to do with Havok and is in the process of being superceded by a platform that addresses most of the issues that its detractors had aimed at it (from PhysX's 3.4 SDK by all accounts)
FordGT90Concept
Havok is far more popular where the physics simulations actually matter (vehicles, fluids, etc.).
Which is the reason that PhysX is evolving into Flex. Adapt or die.
Posted on Reply
#25
FordGT90Concept
"I go fast!1!11!1!"
Which is blatantly stupid. PhysX's problem is the lack of AMD support. That's truly not hard to fix.

Again, FleX is a subset of PhysX; it doesn't appear poised to replace it at all. From what I gather, it's just more eye candy but performed live instead of through modeling. It's more of the same fad/niche PhysX has already been relegated to and, yet again, why would any developer use it when they can use the tools they already have available (like Maya) to do the same damn thing and have work identically on AMD, Intel, and NVIDIA? NVIDIA might as well put PhysX out to pasture.

This move by Microsoft likely proves that NVIDIA won't cooperate with anyone on hardware physics. Microsoft's only option was to approach Intel and luckily Intel agreed.
Posted on Reply
Add your own comment