Print 50 comment(s) - last by Builder15.. on Nov 29 at 3:48 PM

AMD execs talk purchase considerations for physics acceleration hardware maker AGEIA

Few would claim the AMD/ATI merger this point has gone well. John Peddie Research claims ATI only commands a 17.5% share of the graphics market currently, representing a loss in market share. This comes at the same time that NVIDIA, ATI’s largest rival, is posting record profits with its first billion dollar quarter.

AMD’s Richard Huddy said last week that GPU physics is dead for now after the Intel acquisition of Havok. AMD may consider the purchase of physics hardware manufacturer AGEIA; a topic that comes up every three months apparently.

A purchase of AGEIA would allow direct competition between AMD and Intel in the physics world and could open up the ability for hardware based physics acceleration on ATI graphics cards and in AMD processors.

The biggest reason the purchase of AGEIA by AMD is not likely to happen is one of cash.

Huddy recently told Custom PC, “Our biggest problem is that Havok reputedly cost in excess of $100 million. If I'd been valuing Havok, I'd have valued it at probably something like 10 per cent of that because they were in so much trouble in the marketplace, but realistically they did have some valuable IP, and you really can capitalize on that if you're Intel in this situation."

"If AGEIA want to command a comparable price," said Huddy, "then that's a pretty significant problem for AMD. No one would think of us as cash rich at the moment, so splashing an extra $100 million just to get physics, which is a niche market, is quite an issue for us."

Few would consider AGEIA to be flourishing in the current market place with the adoption of their hardware physics accelerators waning. However, that could possibly change in the future with the mobile physics accelerator AGEIA announced in August of 2007.

AGEIA, which has minimal product support right now, might not be in that bad of shape.  Even competitor Havok agrees hardware-physics competition from GPU manufacturers (including AMD) may not take form until Microsoft releases its next-generation API layer

Comments     Threshold

This article is over a month old, voting and posting comments is disabled

By Slaimus on 11/27/2007 2:45:07 PM , Rating: 2
Why doesn't AMD instead spend their money buying innovative CPU startups instead? I remember how they had a chance to buy Stexar, but Nvidia gobbled them up instead.

Aegia does not really help their core business.

By porkpie on 11/27/2007 2:51:34 PM , Rating: 3
I think AEGIA could help them...but not for the price they'd have to pay. When you're swimming in debt, you don't go looking for things to spend money on.

By mlau on 11/27/2007 3:55:12 PM , Rating: 2
I'm curious: How can ageia help amd?

Only an insanely tiny fraction of hardcore gamers buy this completely useless tech (pay 100$ just so that the demo particle fountain can render 500 particles more? And I doubt
anyone would notice a difference in the trajectory of a falling box in a game, or even care)

Floating point offload? That's what DSP's and vector cpus like NEC's SX-9 are for.
Ageia: a solution in search of a problem (and stupid people
with too much money, but they're mulitplying every day ;))

By clovell on 11/27/2007 4:15:02 PM , Rating: 4
It could help with AMD's plans for Fusion - but there hasn't been much said about that lately.

By theapparition on 11/28/2007 7:57:37 AM , Rating: 2
They should drop 100mil on the real future technology...Bigfoot's Killer NIC card. That would certainly solidify AMD's the ground.

Seriously, if they get caught trying to go blow-for-blow with Intel, they are going to lose. Buying Agena puts them one step closer to where Intel want's them.

By Clauzii on 11/27/2007 8:08:51 PM , Rating: 1
Considering the pure performance of the latest offerings from AMD/ATI they are obviously not the fastest out there, which saddens me a bit.

At the same time, I don't think AMD/ATI are totally nuts and lost the knowledge of doing things overnight, and that bying AGEIA is a smart move, especially considering the latest Havoc-buy by intel.

Competition rules upped :)


What the new company would have is a complete platform with CPU, GPU and PHYSICS (which will probably be the know-how regarding Physix but running on ATI hardware which will be a significant speedbump compared to current AGEIA GPUs.

Now I'm only waiting for the intel-nvidia merger to show up.

PS: I'll bet "AAA MicroSystems" could be considered a name for the company as a whole, or is it a mere coincidence that all three's name start with an "A".... :)

By vignyan on 11/28/2007 7:03:04 AM , Rating: 2
Sorry to disappoint you but there is absolutely no reason why Intel has to even consider buying NVidia... probably will buy when their stock drops to $10 (dont think thats going to happen either!) or so when its cheaper to buy than invest in new product technologies! :)

By Clauzii on 11/28/2007 4:23:09 PM , Rating: 2
So intels lineup of GPU-stuff and some Havoc physix will get them up to level with nVidia/ATI?

Or did intel buy Havoc, to earn money on the Havoc licensies?

By subhajit on 11/27/2007 10:24:04 PM , Rating: 2
DirectX 11 supposed to have physics APIs. Having the knowledge of AGEA would help AMD/ATI with the implementation of that.
Besides, AMD can also use it with their Fusion technology or with R700 (which is supposed to have a multi core architecture).

By mlau on 11/28/2007 7:53:57 AM , Rating: 2
That's nice for Microsoft, but really, does *anyone* besides
hardest-core gamers (which are an insignificantly small amount
of people who buy PC hardware) care about physics?

When I play a game, the LAST thing on my mind is whether the
enemies are going down physically correct when their head is
blown off. If you care about such things then go get help asap. Seriously.

By CyborgTMT on 11/28/2007 12:28:11 AM , Rating: 2
One big advantage AMD would have is changing the hardware/software so it would could accelerate both Ageia's physics and Havok. Add it as a part of the GPU and they will have a card that would support all games that are out. IMO this is where Ageia went wrong to begin with, they should have had Havok support in their card. But with Ageia it would have cost them for the licence, with AMD in control a cross licencing deal with Intel between Havok IP and Ageia IP is possible. Then both companies can stick it to nVidia and make them pay to play along....

By slickr on 11/28/07, Rating: 0
By Polynikes on 11/27/2007 5:51:53 PM , Rating: 3
I don't think Aegeia can help.

A GPU-based or separate hardware-based physics solution is not going to work long-term, because in a couple years we'll have more CPU cores than we know what to do with, and no one will buy it. It would probably be easier for developers to simply offload the physics processing to a CPU core than to have to work with another completely different set of instructions for a piece of hardware not very many will have. Honestly, I'm surprised Aegeia has managed to get ANY developers to use their physics solution.

AMD customers won't be real inclined to pay more for a GPU with a physics chip on it when they can buy a cheaper Nvidia solution and let their CPU handle the physics.

By Hydrofirex on 11/27/2007 10:53:54 PM , Rating: 2
A valid counter-point, but isn't there something to be said for specialization? At a certain level of complexity won't even a PPU 'accelerator' be a big help? General purpose CPU's are often many orders of magnitude slower at certain intensive operations than a specialized core.

I've said it here before, but I imagine a 10 core CPU in the future as follows: 1 GPU, 1 physics, 1 AI, and 1 OS/TCP-IP accelerator coupled with 6 general purpose CPU's. From there it really takes off - and this definitely won't have the power you'll be able to give to these specialized areas with a dedicated add-on card. However, I see this as an integrated solution and not 2 or 3 separate cards (Unless you're running a 'multi-GPU' configuration). Further, I would imagine (hope) AI acceleration to be more useful for use in the GUIs of the future than gaming.


By kelmon on 11/28/2007 5:50:47 AM , Rating: 2
A valid counter-point, but isn't there something to be said for specialization?

Honestly, no. The last thing that we need is a load of specialised chips to keep upgrading when we're not even using the resources that we already have. Sure, a general purpose CPU will be slower but it's almost certain to be available in the future and it will be faster than the current solutions so it will be a benefit. Make use of what's already available before adding more.

The sooner this stupid idea is killed the better.

By afkrotch on 11/28/2007 7:44:50 AM , Rating: 2
So why do we have GPUs? Specialized chips give you more performance. With tech nowadays, I'm sure they could easily jam a cpu, gpu, and ppu onto the same die. Would I want it? No. I'd end up having one proc that would cost a buttload to replace, as opposed to individual components. Not to mention, each with their own memory.

If you aren't using the resources that's available, that's your own problem. I have a C2D and I utilize the crap out of it. Am I going C2Q? No, cause I don't have a need for it. It's not Intel or AMD's fault that you bought a proc that you can't use. No one told you to buy it.

By calyth on 11/28/2007 6:04:10 PM , Rating: 2
Honestly, no. The last thing that we need is a load of specialised chips to keep upgrading when we're not even using the resources that we already have.

Unless you don't play 3D games, you do pay for a specialized chip that you hardly use all that often. Sure, Windows uses it to render 2D everyday, but you can probably power down 3/4 of that chip and still have standard 2D performance.

Also, when you make everything GP, you lose a lot of optimization. It's the reason why a 575MHz/1.35GHz 8800GTX can perform so much better in 3D than using your CPU at 2GHz+ rendering the same scene. If the PPU is sufficiently cheap, and not requiring constant upgrades like your GPU, what's the harm in having it to offload some of the more common calculations in an FPS?

By semo on 11/27/2007 2:53:44 PM , Rating: 2
why should they buy anyone.

look at consoles. the big 3 don't own every company that makes parts or assemble their products.

By mendocinosummit on 11/27/2007 4:00:14 PM , Rating: 2
No, but they sell a ready product such as Dell and HP, there is a big difference in why AMD would buy Ageia and why Sony would buy IBM.

By Clauzii on 11/27/2007 8:15:14 PM , Rating: 1
Agree and also, what company would say no to producing 20+ million CPUs to ANYONE? - Not counting countries with restrictions on computer-speed being imported, of course.

By jskirwin on 11/27/2007 2:56:11 PM , Rating: 2
AMD needs to produce better CPUs. Period. Instead of buying something that might work in the future, they need to focus on what they do best - and that is making CPUs and GPUs.

Until they beat Intel in two of the three better/cheaper/faster areas, they are just wasting investor cash.

By wordsworm on 11/27/2007 8:41:45 PM , Rating: 2
Ah... I see, so you think that AMD focusing on the server market, 2-4 CPU that is, so that they can maintain their recent dominance in selling chips for as much as $1,000 each, was a poor decision, and they should instead invest more heavily in the consumer market which is expensive to market in and they're trailing in both brand name recognition and performance. Of course, anyone worth their weight in beans in the server industry will have done research into the options rather than make a hasty decision based on an expensive commercial.

By jhinoz on 11/27/2007 9:36:24 PM , Rating: 2
Maybe they could spend some of that 100 million on a decent product management team instead then?

By theapparition on 11/28/2007 7:45:09 AM , Rating: 2
Ah... I see, so you think that AMD focusing on the server market, 2-4 CPU that is, so that they can maintain their recent dominance in selling chips for as much as $1,000 each, was a poor decision.......

Well, considering how much cash they are bleeding each quarter, with no real positive outlook in sight.....yes, it was a very poor decision.

By KristopherKubicki on 11/27/2007 3:01:51 PM , Rating: 2
Why doesn't AMD instead spend their money buying innovative CPU startups instead?

Because AMD makes it money by selling innovative CPU startups.

I jest though. AMD just recently put some money into Transmeta.

By sxr7171 on 11/27/2007 7:02:36 PM , Rating: 2
They should consider spending money on something to compete with either Intel or Nvidia (take your pick). Phenom - what a joke. After all this time they came up with something that is barely price competitive with Intel's low to mid-end offerings. On the GPU front, they are pretty much floundering and allowing Nvidia to take a vacation. Now they want a company that basically has no product that anyone wants. Yeah sure, buy into more crap and make more stupid decisions and then sell parts of yourself to another middle eastern oil baron.

How sad this story has become since the A64.

By MandrakeQ on 11/27/2007 10:41:52 PM , Rating: 2
AMD needs more resources to develop processors specifically tailored for each market. All of their designs lately have been derivatives of K8, and the original design requirements for that architecture was for servers, not desktops or mobile. If they could've stripped out the IMC and the DDR component from their desktop and mobile parts in favor of larger caches, they would've been much better off in terms of performance.

Not to mention they also dropped the ball with "quad core". This is probably because they don't have the MCM resources to follow Intel's Clovertown example.

With AMD's narrow focus of the server space, it seems they are going to be following Sun's lead in the future with less focus on ILP and more focus on thread level parallelism and better memory technology.

By Phynaz on 11/27/2007 4:15:16 PM , Rating: 2
Why doesn't AMD instead spend their money buying innovative CPU startups instead?

What money?

By ahkey on 11/27/2007 7:05:56 PM , Rating: 2
Or even, god forbid, *GPU* startups, since that's what this concerns. Hell, why not fund better GPU designs that won't necessitate their current practise of price gouging to stay remotely competitive?

By tcsenter on 11/28/2007 3:09:44 AM , Rating: 3
Why doesn't AMD instead spend their money buying innovative CPU startups instead? I remember how they had a chance to buy Stexar, but Nvidia gobbled them up instead.
I'm not sure Nexar would qualify either as an innovative or CPU startup. Nexar was founded in 2005 and closed its doors in 2006 without a single product to its credit, before NVIDIA acquired what was left of the company. It is true Nexar was started by some former Intel x86 engineers, but they reportedly planned to focus on advanced DSP architectures, not general purpose x86 processors.

Nevertheless, why would AMD acquire more of what it already has in spades? AMD's expertise in x86 architecture is second only to Intel. It is worth noting that AMD also has considerable expertise in RISC architecture. e.g. AM29k, Nx586 (NexGen), K6, K7, former DEC Alpha design team, et. al.

NVIDIA didn't have any particular expertise in x86 CPU architecture, making a lot more sense for NVIDIA to acquire some.

AGEIA undoubtedly would complement AMD's core business, but I don't see how AMD can afford it. AMD seems to need every last nickel just to operate day-to-day.

By Clauzii on 11/28/2007 4:32:51 PM , Rating: 2
Someone just (16. November) threw 600+ millions in their pocket:

Seen this before?
By Mitch101 on 11/27/2007 3:32:27 PM , Rating: 1
Where have I seen this suggestion before?

Posted by me back on September 17 @ 08:22:33 MST

I feel NVIDIA and ATI need to work together on a replacement for HAVOK rendering HAVOK’s future a useless buy for Intel. They seriously need to think about working with AGEIA or one of them purchasing them. Either way NVIDIA and ATI need to work together or Intel will eventually roll physics over them and eventually GPU’s.

RE: Seen this before?
By intogamer on 11/28/2007 4:19:41 AM , Rating: 2

But really CPUs are getting faster and powerful all the time. Physics doesn't prove much of an advantage(price) on discrete processing. What if you don't render games? More CPU power is always greatly benefited in mostly every situation. This is a niche for the GPUs.

There is and advantage for Nvidia by having TSMC to produce their chips. CPUs and GPUs are always getting a die shrink, so older fab plants aren't needed. TSMC would produce chips for other companies with Nvidia's older hardware. Didn't AMD just sell their old 120nm fab to some company in Russia?

I don't think Intel will go too narrow on their chipsets. Nvidia's SLI is a brand statement, niche perhaps? More like an optional standard feature. You could SLI, it's just that technology does what it does best. Which in the end defeat's SLI.

Intel starting a GPU division is just nonsense. It would be too much of a monopoly if it dominates with the CPUs. Plus that will be R&D from scratch here.

"Intel won't make lousy graphics forever" is a saying nothing here. Intel's integrated graphics aren't designed to do anything other than to feed out a display. Anything higher is replaced by a mobile chip from ATI/Nvidia

CPUs and GPUs are pretty much in the same nanometer process. Sure you can fab them at the same process, but there will always be a need for a new and bigger facility for the next die shrink. Relating back to TSMC, AMD will won't have much use for the older facilities other than to sell them..

Intel is just sitting on a boat load of money and the Havoc acquisition will be a bonus. Just in case situations?

And lastly Nvidia is not going to push through x86. CPUs are a whole other R&D as it has to be designed to work with everything. They already have something a few years ago with mobile GPUs. Lightweigh mobile CPUs is something nvidia can push through. Whereas AMD has little and Intel has sold off.

Maybe Intel can whip out an even better ultra low watt Centrino/ Solo? Intel is probably avoiding doing stuff to rise too high. Havoc purchase could possibly factor a secret weapon when AMD rises?

It is all a speculating game here guys. This is a whole new generation of technology. The new era of High Def, Fios, 3G, E85 Cars. Another bubble?

RE: Seen this before?
By afkrotch on 11/28/2007 8:14:36 AM , Rating: 2
Not to be rude or anything, but doesn't this whole post seem like incoherent babbling? It's like the subject changes every paragraph, with no seamless transition.

RE: Seen this before?
By Mitch101 on 11/28/2007 10:20:12 AM , Rating: 2
I equate Physics in games currently to something like having Smartshader or Astropic filtering. Currently it adds a level of realism and depth to gameplay. It is the way of the future for gaming. Ageia is a little early but probably not by much. The problem being most of todays benchmarks are based upon framerate and not visual quality. Adding physics usually causes a framerate reduction because of the additional complexity or additional details physics adds to the scene to be drawn.

Visually games are going to run into a roadblock with detail levels. We are getting close to being able to render Jurrasic Park in real time. It may not look like it today but were only a few years off when you can do Quad graphics chips and soon multi core then it will all come down to eye candy. We can move around physics for now but eventually it will be a marketing tool used against the competition. Every company brags about thier radtime physics but they would love it 10 time more if they had a PPU in every PC to make life easier. If NVIDIA has special physics instructions built into their cards next year they will push that as gospel all over ATI. Marketing could drive us into the physics age. Right now AGEIA Marketing cant do that because there isnt much using physics now only minor patches not full blown physics games but Ageia is working on producing a game to show everyone what they are missing out on.

Sadly Physics might not take full swing until DX11 or AMD Fusion. Either way its coming and its going to be good.

CPU's are not good for physics no matter how many current cpu cores you have.

amd should do it. ha!
By semo on 11/27/2007 2:49:29 PM , Rating: 2
common just partner up, share resources, learn from each other and offer a package worth buying that has real uses.

just because amd bought ati doesn't mean it will buy every other company connected to personal computers.

RE: amd should do it. ha!
By bupkus on 11/27/2007 5:52:41 PM , Rating: 2
Yah, just who do they think they are, Google?

I'm just sayin'....

RE: amd should do it. ha!
By Gul Westfale on 11/27/2007 10:54:55 PM , Rating: 2
if they integrated next-gen ageia tech into motherboard chipsets then that would give them a definite advantage in the enthusiast market, as the physics processing would be off loaded from the CPU and thus overall system performance would increase.

BUT: current ageia tech seems far from the marvel of modern technology it was made out to be, and i'm not buying a $200 add-in card if i can put the same amount towards a faster dual or quad core CPU.

thus, buying ageia does not make sense to me, unless ageia have something up their sleeve that AMD thinks it can use...

Are they high?
By jiteo on 11/27/2007 4:13:37 PM , Rating: 3
I can see how such a purchase would open up interesting possibilities in the future, a bit like Fusion with the ATI acquisition, but where the crap are they going to get the money? They were probaly drinking their troubles away, then one of them saw a dime on the floor, and said "Duuuude, let's buy something!" Then the next morning after the hangover, "Oh man, that was not a good idea." "So should backtrack on our press release sir?" "We did press release?!"

RE: Are they high?
By bupkus on 11/27/2007 5:55:01 PM , Rating: 2
Too bad I'm out of votes.

To buy something you need funds ...
By 2ManyOptions on 11/28/2007 3:26:12 AM , Rating: 2
Considering the current scenario, which is, AMD is into losses quarter on quarter, will it be a good thing to do to go and buy? Also, where will they get the money from? They have a substantial debt I believe?

By intogamer on 11/28/2007 4:32:57 AM , Rating: 2
I would say AMD is a long term gamble. Even when AMD comes out and succeed. They still have a huge debt under their ass. That means less money for investing on themselves. AMD is going to keep dropping. I would buy and depending on anymore information on AMD's future. Possibly just a phase of teh semiconductorz.

Stupid, stupid, stupid AMD....
By sxr7171 on 11/27/2007 6:57:34 PM , Rating: 3
One mistake after another. You'd think that they'd have learned by now.

By knowom on 11/27/2007 6:10:53 PM , Rating: 2
Seems like a waste of money for AMD since they're all ready in a heap of debt. It could see it as a great buy for Nvidia who has zero debt and like a billion in spending cash.

By amanojaku on 11/27/2007 7:21:17 PM , Rating: 2
For God's sake, no...

Useless for the fun games?
By tjr508 on 11/27/2007 8:59:30 PM , Rating: 2
I was wondering how such hardware may affect multiplayer games. It seems to me that physics affects gameplay. As the number one, above all else, most important factor in online gaming is a level playing field, would games have to be split into one group with added physics hardware and one without?

I have been thinking this ever since I first heard of physice hardware and am suprised it doesn't come up more often.

Let it go.
By EglsFly on 11/27/2007 9:54:39 PM , Rating: 2
Purchasing ATI seemed to really weigh heavy on AMDs money and resources, they need to let this one go.

By Silver2k7 on 11/28/2007 3:37:58 AM , Rating: 2
Ageia just recently slashed its prices to $99 for the cards.. but when I ask a local computer shop to check for price its 1400 SEK or the same as $220 !!

If I could find it for $99 in Sweden I would have ordered one when I purchased Unreal Tournament 3.. also the only avalible in PCI is a bit of a turndown when PCI is beeing phased out for PCIe slots.. if I purchase a PCI card now my next computer might not even have a slot for the card.

By crystal clear on 11/28/2007 3:38:44 AM , Rating: 2
Just how can you believe this news item about AMD buying AGEI when it(AMD) can barely pay its way through the next 6 months.

Read this -

AMD Shares Hit Year Low on Downgrade
Associated Press 11.27.07, 11:19 AM ET

NEW YORK - Shares of Advanced Micro Devices Inc. on Tuesday dropped to a year low for the second day in a row after an American Technology Research analyst downgraded the semiconductor maker, predicting no catalysts to boost the share price within the next three to four months.

In a note to investors, analyst Doug Freedman downgraded the stock to "Neutral" from "Buy." He cited the lack of "clear cut positive catalysts" until the company introduces its new 45 nanometer-based chip in mid-2008.

"Management has been reluctant to take steps we believe would have added near-term value for shareholders, and there are no signs of a strategy change," he added.

He also suspects that investors will view with skepticism any announcements on the company's "fab-lite" strategy, under which AMD hopes to outsource more of its chip fabrication to third parties.

Freedman said that he was not certain investors will view AMD's fabless strategy as positive, regardless of the strategy's merits, because of comparisons with AMD's larger rival Intel Corp. (nasdaq: INTC - news - people )

"For one, investors continue to believe that Intel's key differentiating factor is its in-house design and manufacturing excellence," Freedman wrote.

AMD shares dropped 35 cents, or 3.4 percent to $9.92 in morning trading. Earlier, they traded as low as $9.80, below a 52-week low of $10.21 set Monday

52wk High 23.00 - - - - - - - 52wk Low 10.21


Investors are worried that AMD's sale of an 8.1 percent stake to an Abu Dhabi investment company for $622 million last week won't be enough, analysts including Jefferies & Co.'s John Lau said. Another stock sale would dilute investors' holdings and lower earnings per share.

"Investors fear another round of financing may be forthcoming," Lau said. "The recent secondary offering still isn't enough for them to fund their future plans."

It could be also that AMD has a secret commitment from the
Emir Sheikh Mohammed bin Rashid al-Maktoum to finance it even more.......

Why doesn't Intel buy them?
By Builder15 on 11/29/2007 3:48:34 PM , Rating: 2
I would like to know why Intel doesn't buy Ageia. They already purchased Havok. This would provide them with basically all the physics.

"We’re Apple. We don’t wear suits. We don’t even own suits." -- Apple CEO Steve Jobs
Related Articles

Most Popular Articles5 Cases for iPhone 7 and 7 iPhone Plus
September 18, 2016, 10:08 AM
Laptop or Tablet - Which Do You Prefer?
September 20, 2016, 6:32 AM
Update: Samsung Exchange Program Now in Progress
September 20, 2016, 5:30 AM
Smartphone Screen Protectors – What To Look For
September 21, 2016, 9:33 AM
Walmart may get "Robot Shopping Carts?"
September 17, 2016, 6:01 AM

Copyright 2016 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki