Hard Light Productions Forums
Off-Topic Discussion => General Discussion => Topic started by: Fury on July 24, 2006, 08:25:03 am
-
AMD press release. (http://www.amd.com/us-en/0,,3715_14197_14198,00.html?redir=goBG01)
Edit: AMD/ATI merger FAQ. (http://www.amd.com/us-en/0,,3715_14197_14208,00.html)
-
Interesting. going after (partly) intels integrated gpu market?
-
:jaw: Now we're one step closer to an effective integrated all in one pc off-the-shelf, *eventually :nervous:* Well once they buy soundblaster i mean :lol:
The motherboards of the next few years are gonna be V V V Nice.....
-
I don't like this. I forsee great troubles when using nVidia cards with AMD/ATI boards/chips.
-
I don't like this. I forsee great troubles when using nVidia cards with AMD/ATI boards/chips.
AMD is not Intel, you forsee nothing.
especially since it's hardly possible to cause problems like that on a PCIe bus without specifically designing problems into the CPU and then it would be antitrust and nVidia could sue AMD
oh and someone posted this on /.
http://catb.org/jargon/html/W/wheel-of-reincarnation.html
wheel of reincarnation: [1968] Term used to refer to a well-known effect whereby function in a computing system family is migrated out to special-purpose peripheral hardware for speed, then the peripheral evolves toward more computing power as it does its job, then somebody notices that it is inefficient to support two asymmetrical processors in the architecture and folds the function back into the main CPU, at which point the cycle begins again.
Several iterations of this cycle have been observed in graphics-processor design, and at least one or two in communications and floating-point processors. [...]
-
antitrust
I love that term, A board is a board, They will accept NVIDIA no probs (i hope).........
-
they would be absolutely out of their freaking minds to do something to piss off nvidia - the CPU and GPU markets both NEED two codominant manufactuerers to drive innnovation - and it would be a huge profit loss to AMD to piss off nvidia
-
they would be absolutely out of their freaking minds to do something to piss off nvidia - the CPU and GPU markets both NEED two codominant manufactuerers to drive innnovation - and it would be a huge profit loss to AMD to piss off nvidia
They won't bend over backwards to make it easy, though........... I think AMD / ATi have already pointed out that part of this is a tactic to split open the imarket, even at the expense of short term profit (i.e. relating to intel support / mobos). It's probably too risky to try and use that against both nVidia and Intel of course, but I wouldn't be surprised if they had their eye on, um, 'optimizing' for ATi proprietary tech in a manner somewhat akin to MS' apparent sidelining of OGL with Vista.
-
AMD isn't dumb enough to engage in anticompetative practices - hardware isn't software, it's really easy to prove anticompetative measures have been taken
-
AMD isn't dumb enough to engage in anticompetative practices - hardware isn't software, it's really easy to prove anticompetative measures have been taken
Surely what you mean is AMD isn't dumb enough to engage in anticompetative (sic) practices and get caught? ;)
-
Since AMD is significantly more friendly towards OSS developers, hopefully this merge will result in AMD releasing ATI's GPU specs to the driver developers. Linux users have been suffering from ATI's horrible binary drivers for long enough.
-
this will either be a very good thing, or a very bad thing. only time will tell.
-
this will either be a very good thing, or a very bad thing. only time will tell.
It could be a 'meh' thing.
-
Capitalism.
[Chris Lambert]THERE CAN BE ONLY ONE![/Chris Lambert]
-
When AMD has paid their loan to buy ATI, AMD should get another loan to buy C-Media. If you didn't know it, C-Media produces the best audio chipsets after Creative. Would certainly make both integated and dedicated sound card market much more interesting.
Since AMD is significantly more friendly towards OSS developers, hopefully this merge will result in AMD releasing ATI's GPU specs to the driver developers. Linux users have been suffering from ATI's horrible binary drivers for long enough.
No kidding. I really hope that ATI linux drivers will rapidly improve.
-
hopefully this merge will result in AMD releasing ATI's GPU specs to the driver developers
I'm sure both Linux and Windows users would appreciate that very much. :p
-
I wonder if AMD will decide to keep the ATI name for products currently in ATI's portfolio in 6 months time. I'm looking forward towards what this merger might bring, hopefully a completely updated platform.
I've noticed mergers are quite common now as well :nod:.
-
See the FAQ, ATI's name will change to AMD.
-
amd radeon just looses that ring to it doesnt it :D
-
i bet you the radeon line will remain "ATI Radeon" but they'll start a new line post merger completion
-
See the FAQ, ATI's name will change to AMD.
I already knew that :p, but Nuke pointed out what I didn't want to know
-
I doubt this will have much effect on the GPU market for at least a year, but it's hard to see how the RD600 chipset is going to go through now. Shame too, as I was planning on getting the DFI RD600 board when it came out. :(
-
hm I forsee ATi and AMD being built to run faster with eachother, and some sort of Intel/nVidia alliance. :P :eek2:
-
hm I forsee ATi and AMD being built to run faster with eachother, and some sort of Intel/nVidia alliance. :P :eek2:
id rather nividia stay in the game by itself. that way their priority to make good gpus remains. i have a feeling ati is gonna focus on chipsets and integrated video rather than video cards. essentially it will mean amd will have its own line of graphics processors like intel, except because its based off of ati hardware, it could be better than those ****ty extreme graphics integrated video thingies.
-
heh Intel Extreme is the worse graphics chip I know, even more than Geforce 2 if I remember right.
-
heh Intel Extreme is the worse graphics chip I know, even more than Geforce 2 if I remember right.
How shall I say this... it's an integrated graphic card. It's dirt cheap and not meant for heavy graphic applications.
-
My little bros computer is running an ATI Rage Pro AGP 2X graphics card now, Intel Extreme is not horrible compared to that.
-
Here's a chart :
Geforce 7950 SLI / 7950 QUAD-SLI (Total Ownage...for now)
....
Geforce 7800 GT / Radeon x1800XTXwhatever (Ownage)
...
..
Geforce 6800 / Radeon x800 (Standard Solid Gaming)
...
...
Geforce 4 Ti (Sucks but... respect)
Geforce FX (Sucks)
...
...
Geforce 3 (Sucks, Classic !)
..
..
Geforce 2 GTS (Really Really Sucks, Uber Classic !)
..
..
..
Geforce MX 2/4 (Really Really Really Sucks, Not a Classic)
..
..
TNT2/Rage Pro(Really Really Really Sucks, Not a Classic, Not Even Playable)
..
..
..
Intel Extreme Graphics(Really Really Really Sucks, Not a Classic, Not Even Playable, Do not Pass Go, Do Not Collect $200)
-
Geforce 7950 SLI / 7950 QUAD-SLI (Total Ownage...for now)
Yeah, total ownage with screen tearing and stuttering.
(sorry, had to say that after getting suckered into SLI and finding out about all its problems :p)
-
i think that windows vista is the main reason behind the amd-ati merger. intel's graphics solution will not only be obsolete (it is already) but it be totally worthless when vista comes out unless they really supercharge that bastard. if what everyone says about vista is true even for office applications will require some acceleration. so in that market intel's solution is as good as dirt. as for the graphics card market its hard to say. they might not put suffietient r&d into their graphics card line to keep up with nvidia.
-
There's an Intel integrated card out there (I think it's the Extreme 955, or something) that actually beats the crap out of a Radeon 9700.
I doubt nVidia and Intel will merge, but they might work a little closer. nVidia's market includes stuff like military HUD hardware and the like, so they're not likely to be unduly worried about AMD and ATI merging to better control the PC gaming market.
-
I just noticed that FiringSquad has tossed together a six-page long speculation article (http://www.firingsquad.com/hardware/ati_amd_merger_faceoff/) about the merger.
-
Here's a chart :
Geforce 2 GTS (Really Really Sucks, Uber Classic !)
Guess this one is mine LOL
-
Sooner or later they'll have to move away from integrated chipsets. They just don't provide as much power as add-on graphics cards, and with innovative ideas like AGEIA’s PhysiX becoming available theres no doubt they won't be able to squeeze everything onto a motherboard.
-
Here's a chart :
ETC>
Wheres my Voodoo 2 ?
[j/k srsly]
-
Sooner or later they'll have to move away from integrated chipsets. They just don't provide as much power as add-on graphics cards, and with innovative ideas like AGEIA’s PhysiX becoming available theres no doubt they won't be able to squeeze everything onto a motherboard.
You fail to see the reason integrated cards exist. They don't exist to compete with the existing cards but to ensure that the target computer can get suficient graphic "acceleration" (read, not heavy graphics aplications).
Oh, AGEIA's sucks. It will actually slow down your pc if you use one of those cards.
-
Sooner or later they'll have to move away from integrated chipsets. They just don't provide as much power as add-on graphics cards, and with innovative ideas like AGEIA’s PhysiX becoming available theres no doubt they won't be able to squeeze everything onto a motherboard.
You fail to see the reason integrated cards exist. They don't exist to compete with the existing cards but to ensure that the target computer can get suficient graphic "acceleration" (read, not heavy graphics aplications).
Oh, AGEIA's sucks. It will actually slow down your pc if you use one of those cards.
yeah its true, the physX card is designed to do maths... and guess what the CPU is designed to do with direct feed to the RAM?!? Even that PhysX requring game (CellFactor I think) runs without it if you just add something to the shortcut path. Waste of silicon.
Oh yeah and integrated cards are for running windows, not Far Cry. I can only seeing them getting a little better when vista comes out because it requires a fairly good card.
-
Ghostavo
You fail to see the reason integrated cards exist. They don't exist to compete with the existing cards but to ensure that the target computer can get suficient graphic "acceleration" (read, not heavy graphics aplications).
Apathy
Oh yeah and integrated cards are for running windows, not Far Cry. I can only seeing them getting a little better when vista comes out because it requires a fairly good card.
Actually I understand completely, but with the next Windows OS becoming available soon and applications demanding more grunt, an integrated chipset just doesn't have the power to run this next-gen software and hence why integrated cards will become a thing of the past.
Oh, AGEIA's sucks. It will actually slow down your pc if you use one of those cards.
It might, but I didn't say the card was any good, I simply said the idea was. I do believe that certain processes should be handle independently of the CPU and GPU, they're designed to handle many different calculations and having a dedicated chip to processing specific calculations should provide better results overall.
yeah its true, the physX card is designed to do maths... and guess what the CPU is designed to do with direct feed to the RAM?!? Even that PhysX requring game (CellFactor I think) runs without it if you just add something to the shortcut path. Waste of silicon.
From what I have read the PPU runs physics calculations much faster than any CPU.
-
But that's no use at all if data flows to and from it at a quarter of the speed.
-
hence why integrated cards will become a thing of the past
No they won't, what will happen is that integrated graphics chips will become better and small costs and passive cooling is achieved from smaller die sizes. ATI and NVIDIA have already produced integrated graphics chips in the past and they will continue to do so in the future. More than likely Intel will also have two integrated graphics chip lines, one low-end and one high-end.
ATI's Xpress 1100 series integrated GPU's and NVIDIA's GeForce Go 6150 are more than capable of running Windows Vista in full hardware acceleration including any home/work applications. If you don't buy a computer for gaming, you don't need a dedicated graphics card. Thus wasting your money for anything else than an integrated graphics chip is pointless.
It's not the fault of hardware manufacturers and OEM's if customers don't know what they're buying and expect to play games with hardware not designed for the task. If anything, the responsibility of guiding a customer into buying a computer for his or her own needs is on resellers.
-
You know they are reaching the limit on how small they can make wafers. I checked out ATI's Xpress 1100 series and it does seem promising, but the clock speed could be higher. As for it saying that it'll be able to run Windows Vista in full hardware acceleration mode only means that it will, it doesn't say how fast or how long loading times will be. Your not going to run some commercial applications smoothly with an integrated chip either. Another point worth mentioning is that on most boards the northbridge and southbridge already have heat sinks, soon they'll need fans, its just a matter of time for integrated chips.
But that's no use at all if data flows to and from it at a quarter of the speed.
Where did you get that from? The only limit for speed between the graphics card and the cpu is the AGP/PCI-E speed(transfer rate), since I'd imagine that the speed from the northbridge, the FSB and the memory are faster.
-
the smallest theoretical size you can cut a chip is to a point where wires on the chip are one or two atoms thick. it will take many advances in nanorobotics to acheace that though, but its theoretically possible. that wall is still a ways away. the wall were at now is the 3 ghz barrier. increasing the clock is pointless if you cant feed the chip data fast enough. bus technolody needs to be improved. all the speed boosting advances lately have been in improving the flow of data to the core. its why using multiple processors doesnt increase the performance in perportion to the number of cpus. they bipass this limitation with dual cores, by transfering data between them at the chips clock rate and by sharing cache. still its not a 2 x increase. i look forward to the high end quad core chip intel is working on, but if its a 3x speed increase over a single core then it will be a good thing. im curious how far they can push multiple cores before they hit another wall.
-
OK, let's take a look at what you've just said.
1. The 3GHz barrier. We've passed that barrier.
2. Why increasing the clock speed is useless is not because of the bus, but rather because to increase the clock speed they increase the length of the pipeline, making brand prediction errors more costly and lowering the IPC (Instructions Per Clock cycle) of the CPU.
3. Increasing the bus is a concern, but more for transferring data to and from the RAM, not for transferring data between the processors.
4. You fail at microelectronics. It's not nanorobotics that are creating a wall to decreasing feature size on chips, it's things like pattern transfer, index of refraction, etc.
In conclusion, use real facts, not ones you made up.
-
facts have a tendancy to be misleading :D
-
But that's no use at all if data flows to and from it at a quarter of the speed.
Where did you get that from? The only limit for speed between the graphics card and the cpu is the AGP/PCI-E speed(transfer rate), since I'd imagine that the speed from the northbridge, the FSB and the memory are faster.
Actually. it was a general comment on the tendancy of gimmickery to ignore the real bottlenecks, but now that you mention it... what's the bandwidth of PCI-E 4x? Memory bandwidth is something like 3-4GB/sec these days. Then we have to consider latency; it's no good shifting 16GB/sec if it's taking so long the CPU's already working on the next frame by the time the PPU is finished with the current one.
So for the PPU to be worthwhile, it must not only be able to do the calculations bust must also be connected to a suitable bus. I'm guessing PCI-E 4x is adequate; graphics cards need to shift a lot of data and they always worked fine on AGP.
Even so, the CPU's bottleneck these days is memory speeds. That is going to be worse for a device on the PCI-E bus. To make the PPU worthwhile, it'd need a substantial amount of on-board memory for caching objects, but since the CPU needs access to those objects as well that stuff still needs writing back to memory.
I can see the PPU being useful when the CPU is close to maximum load, but otherwise it's just adding latency to the loop.
-
The answer to high bandwidth and low latency needs is HTX (http://en.wikipedia.org/wiki/HyperTransport#HTX_and_Co-processor_interconnect). Many major players are supporting HTX, so Intel will either have to license HyperTransport including HTX or develop their Common Systems Interconnect (CSI) to include 100% compatible connector similar to HTX. But Intel's CSI is still in development and will probably be launched sometime 2008.
-
Theres an Anaandtech article for this too...
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2807&p=1
Honestly, Im totally lost. I guess I can see why AMD is doing this.
In short, AMD is losing money because of Intel's Core Duo 2, and AMD is running out of resources, so they decided to pur 5.4 billion dollars into buying ATI and theyre hoping it will all pay off when "GPUs and CPUs become one". I dont see that happening for at least 5 years, maybe 10. Thats a total waste of money, time and investment.
Intel seems totally fine on this, they have market share, and enouch cash to continue, and with the Core 2 Duo just released, and with all the stuff Dell has bought from them, theyre good so far. (I did hear that Dell was contemplating buying AMD parts because Intel wasnt delivering for them, but thats changed).
Nvidia is im assuming at this point probably going to be making more for Intel, as AMD is acquiring ATI. Apparently there is supposed to be an evening out of chipset sales per company sometime around, from what I can see in the article.
ATI is all game for this i guess. I think that they are going to lose all ties with Intel because AMD is (or at least was) prime competition for Intel. If ATI joins up with AMD, I dont think intel would want and ATI chips OR support, since that would be like feeding directly to your enemy (AMD).
Man times have changed... All I know is I'm totally happy with Intel's performance lately and if they end up working more with Nvidia then they have with ATI, im game. Though I do favor ATI cards, I was never really an AMD fan and now that Intel has control of the road for at least 5 years or so and AMD is probably not going to gain much more control, Nvidia working with Intel would probably work better for both companies, as both are top notch, compared to their counterpart companies battling them.
I'll still buy ATI Cards.
Someone said that ATI is going to be renamed? EWWW...
AMD Radeon? That sounds horrible..
-
If you're buying a card based on it's name, you deserve to get shafted. ;)
-
If you're buying a card based on it's name, you deserve to get shafted. ;)
no i didnt say that.
I meant that I like ATI cars better than Nvidias.
but the name AMD Radeon is lame...
-
All I can say is, vroom? :D
-
AMD is strongly considering open-sourcing at least a functional subset of ATI’s graphics drivers. It’s time for X Window System, OpenGL, and client virtualization for which ATI binary drivers aren’t available to escape the ghetto of the 1980s-era framebuffer. And what a boon for PR. If AMD’s graphics cards were the only ones with open device drivers, it might affect a buying decision or two.
http://www.infoworld.com/article/06/08/02/32OPcurve_1.html
-
http://lwn.net/Articles/195061/ (http://lwn.net/Articles/195061/)
Boy, this merger must be fun.
-
This isn't a merger. AMD bought ATI. That's not the same thing as two companies merging, FYI.