• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

AMD Ponders Potential AGEIA Purchase

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
More acquisitions should be the lastest of the last last things on AMD's minds.
 
I saw a report today that this is a run away rumor that started when one of the execs was speaking hypothetically about if AMD were to purchase Ageia. That somehow quickly boiled into a full accusition intent.
 
I doubt AMD would fish in troubled waters at this point. Their GPU devision is just beginning to come out of a slump while AEGIA are struggling. If AMD can pick them up at a bargain basement price, it may be o.k, but otherwise...
 
I don't have any hard numbers, but I would assume running physics on most newer ATI cards would be faster than on an Ageia PPU. Unless AMD knows of some sort of break through at Ageia then AMD is already well equipped to compete in that market.
 
I would think Fusion is not meant to be a powerhouse gaming solution, so adding physics to it to me seems pointless. Fusion will be for mobile devices primarily from what I've read.
 
I saw a report today that this is a run away rumour that started when one of the execs was speaking hypothetically about if AMD were to purchase Ageia. That somehow quickly boiled into a full accusation intent.

Yeah, the Inquirer published the following which is supposed to be an offifical statement from AMD.

Richard Huddy @ AMD said:
I think my words have been taken to mean more than they really do:

I said that I wouldn’t rule it out and I said we’re not likely to splash $100M on a PPU vendor, and I said that we go through the arguments every few months.

That’s certainly not something that I would have headlined as “AMD considers buying Ageia”. The thought has certainly passed through our minds - but then humans think about a lot of things, and if you look at what I said I also point out that the cost is crazy high, so actually I come very close to ruling it out…

As far as I can see there’s really no news here. Everyone in the industry understands that Ageia’s primary aim is to just to be bought, and we’re one of the companies that needs to work out whether we think it makes sense.
 
This is a chance chance for AMD to put in a set of sideband processors on DIE with FUSION. The purpose of these SPs if to so what stream procs on GPUs do as GPGUPs. Think of it as a bank of 32 to 250 stream processors with FPU abilities and basic operations. Data is streamed in with a small inline cache. These could be used for physics, encoding, topography management, micro engines and other features where all you are doing is feeding numbers to be crunched on a timely basis. The core would be almost the size of one standard core but require to load store unit much like a north bridge.

If AMD is backing down from GPGPU does this effect Fusion?
Who would believe Intel would go for such throat cutting measures as buying up companies that play an active role in the industry. Does the dire situtation with Havoc indicate that PC gaming in in trouble vs the Game Box industry?
If AMD falls back on the ATI side and the desktop CPU, they may end up as just a server chip maker with a line of low end descktop CPUs. We can now look forward to 300-500 as being the norm for moderate prices for Gameing Rig CPUs. Cheak overclocking is about to become limited to low cache, low core count chips.
 
Beacuse now they don't have a physics engine to run on their GPUs?

Like Alabama said, stream processors can do just about any floating point based calculations you want them to do. This is why I bought a G80 based Nvidia card because it was the first stream processor based chip from Nvidia. Once the software side catches up, a stream processor can dynamically change how much of what it processes per cycle, including physics. The "engines" are now software based. This adds the possibility that when a new version of pixel shaders is released, then in theory all Nvidia needs to do is release a driver update that will make a G80 compatible with it.

I haven't kept up with the ATI side of things but as far as I know most of their current cards are stream processor based. Thus why take the extra die space for a chip that can only do physics when you can use single stream processor for many different things. The more all in one parts you have is ideal for an integrated solution like fusion.
 
Yes I know but Havoc for GPUs is most likely dead because Intel purchased it. Havoc is (or was) THE big player for game physics, so regardless of how wonderful AMD or Nvidia's stream processors might be, they need to find a physics engine to run on them.
 
Yes I know but Havoc for GPUs is most likely dead because Intel purchased it. Havoc is (or was) THE big player for game physics, so regardless of how wonderful AMD or Nvidia's stream processors might be, they need to find a physics engine to run on them.

I didn't know they were bought out. Interesting...
 
After thinking about this, looks like physics processing is going to go stale. You now have one monopoly controlling it and that means only one engine common engine that only the big game companies will use. Microsoft shoved directX into the graphics market cutting Open GL down to almost nothing. The latest Open GL runs on most OSes. Direct X10 only works on Vista (minus a hacked XP). This means if you want physics in the future you will have to buy Intel or AMD implements it's own. As developers we end up having to support different standards. These monopolies think they have to control the standard just to make more money of control the market. As consumers we suffer from incompatibilities and the software companies foot the bill for supporting all the problems it causes. We can't control the eager buyer who buys the product without reading then installs it on the wrong type of machine then does a chargeback on their CC.
 
Back