The Texture Units
As we said earlier, Larrabee doesn’t look much like a GPU, but while Intel was able to eliminate the setup engine and ROP by implementing those functions directly at the level of the Larrabee cores, the same isn’t true of the texture units. These units handle a very specific job that can be done much more efficiently in a dedicated unit. Texture decompression (DXTC), for example, is very simple to do in hardware, which is what guaranteed its success, but it requires enormous resources to do it on the software side. Intel calculated that performing texture operations on the processors would be between 12 and 40 times slower than on dedicated units, depending on the configuration (such as filtering quality and compressed or uncompressed texture format).
The texture units are fairly classic, and Intel doesn’t really provide a lot of detail about them, except that they will support all standard Direct3D 10 operations and compression modes. In fact, that’s the only point that might limit Larrabee’s ability to evolve. As we’ll see later, software entirely handles Larrabee’s rendering pipeline and so could address the future functionalities of Microsoft's application programming interface (API) via a simple update of the 3D engine executed by Larrabee’s cores. On the other hand, the texture units will be stuck at a certain level of functionality, which will determine that of the chip as a whole.
One interesting specificity of these texture units is that they handle translation of virtual addresses into physical addresses, which, in more concrete terms, means that there’s no longer a need to load an entire texture with its mipmap stack into local memory, as only the portions needed for display will be stored in the form of pages of a few kilobytes in size. If a page isn’t available in memory (a page fault), the texture unit notifies the processor, which then calls for it. This mechanism would make it very simple for programmers who aren’t afraid to get their hands dirty to implement, for example, an algorithm like id Software’s MegaTexturing.
Latest Graphics Cards News
- 20/05 – Qualcomm Hires Former AMD CTO Demers
- 18/05 – Nvidia Responds to GeForce 600 Series V-Sync Stuttering Issue
- 18/05 – Behold: Here Are The First Unreal Engine 4 Screens
- 17/05 – Nvidia Debuts GK110-based 7.1 Billion Transistor Super GPU
- 17/05 – GE Announces Kepler Graphics Card for Military and Aviation
Latest Graphics Cards reviews
- 18/05 – Best Gaming CPUs For The Money: May 2012
- 10/05 – Best Graphics Cards For The Money: May 2012
- 10/05 – GeForce GTX 670 2 GB Review: Is It Already Time To Forget GTX...
- 09/05 – Video Teaser: Radeon HD 7800 Series Remains A Solid Value
- 08/05 – Benchmarking AMD's 768-Shader Pitcairn: Not For Public...
When push comes to chase, who thinks it will outperform my 8800gt?
hrm.. first time i ever do TL DR to a article, sinc eit sfuckign intel i DONT belive it will perfrom, but i hope it DOES and smacks me behidn the head with super fast performance
(btw great article im sure you explained why it might or not kick ass)
also i own an intel cpu, but i DO not belive intel witll get it right (at first)
An earlier article mentioned this is initially aimed at replacing the onboard graphics parts Intel uses i.e. GMA950, etc. So don't expect it to come anywhere near an 8800gt, at least not initially and perhaps not ever as that may not be the point of it. Perhaps Intel are just thinking of a super cheap onboard graphics part that can be updated to match any future DX/OGL spec or video codec with just a patch as well being available for accelerating specialist apps that need number crunching.
fezztah,
If you read the article it is more more expensive to produce than the cell. So really I don't think that hoping it outpaces the now old 8800GT is a lot to expect.
Also it mentions that this is intels stab at DISCRETE graphics, meaning not on a motherboard.
"start-ups that offered first-rate 3D performance (3Dfx, Nvidia, and PowerVR) and heavy hitters who seemed to think that 3D acceleration was just a gadget (Matrox, S3, and ATI before AMD purchased it"
I have to disagree on this as Matrox, before going business only, was high-end.
Can't say i have much faith in Intel producing a good graphics solution.
Maybe this will just end up as an addin card for applications that require parallel processing. Will they be able into interconnect like SLI???
Ok the addin card is just an idea but surely from what i read for some applications it could be an ideal solution. Wouldn't it work well for decryption like the Nvida cards have been used?
If it's PCI-E then I might buy it. Or if my new mobo works with it.
@Belinda, doubt the dual gpu idea will get to intel.
@core i7 ownage, it'll probably be pci-e but with an x86 base(quickly reminding everyone how the 1ghz snapdragon comes close to the atom)it won't perform great, unless it's basically an i7 cpu on a card because then you'd have tesla league power.
Got mixed feelings about Intel's hybrid solution on account of its x86-based cores:
I don't know if Intel is capable of pulling this off, since there's a lot more to it than simply having a better product on paper; Betamax was better than VHS but look what happened in the end. Its one of several solutions, some of which are already in mass production.
I see nothing stopping say AMD or nVidia taking this design and adding their own tweaks, in the meantime using their discrete technology to accelerate applications that are in dire need of a speed hike. Ask anyone who uses video transcoding software: do you care whether it's 15x or 50x faster..whether it's AMD, Intel or nVidia..as long as it's available now, and at the right price?
Keeping all the options open was a good idea on the part of AMD, since competing directly with the fastest Intel Core i7 architecture isn't going to be the future of desktop computing.
Thanks for this interesting article, Abi-Chahla. There is a lot of information to take in (some of which is beyond my understanding), but it helps put things in perspective.
Is it just me or is that an AGP card in the picture?
I Lol'd.
If Intel stick with it and do produce a quality piece of kit down the line maybe Nvidia and AMD will take note.
Competition = lower prices, that's the most i can hope for from Intel with this card to be honest.
Like most people who play games I doubt this is going to be worth looking at when it's launched. Even compared to other cards in the same price range.
In my opinion its not really the chipset, gpu, or cpu in this day an age, there all super capable of much much more than current programming capacities give credit. Give any instruction set a first glance and give the actual use of instruction sets a glance and the lists parallel themselves.
The fact is, there not even usen a grain of sand in comparison to the capabilities of the current platforms in any direction you go in, but to think that an instruction set so used and abused as x86 can deliver anything new to the programming angle of the equation,,,, I just dont see it.
Flexible, yes, manageable, definitely, a solution, no. GPU is my opinion of a solution within itself. It is capable of so so much more than it is currently used.
Why no mention if AMD+ATI Fusion?
AMD is leveraging its acquisition AIT to combine CPU and GPU on one chip. This could give advantages over an Intel CPU + any add-on graphics card. Who knows how any of this will work out, but my guess is that Intel has realised that is cannot afford to be left out in the cold if the AMD Fusion chips prove to be very successful. Therefore, Intel is getting into the graphics market with a very CPUish graphics chip to cover itself.
In any case, competition can only be good for the consumer.
P.S. I thought what happened a decade ago with Intel graphics chips was that Intel was hoping to re-used older fabrication equipment to make the chips, but what happened is GPUs used the latest, smallest dies so the Intel strategy just did not produce anything competitive on their old fab gear.
This is Intels stepping-stone towards real time ray tracing, this will require many x86 cores. This is Intels way of providing them.