Intel hoping to get Larrabee within next-gen Xbox?

Quite honestly, there's absolutely zero proof that Intel and Microsoft are yapping it up in regard to Larrabee, but considering just how logical it sounds, we couldn't help but pass along the latest whispers surrounding the two. According to those fabled "industry insiders," The Inquirer has it that Intel is currently on bended knee asking Microsoft to integrate Larrabee into the next-generation Xbox. If it were to land such a deal, developers would -- in essence -- be forced to develop for Larrabee given the need for console titles, which could then make things all the easier for Intel on the PC side. Like we said, this is all speculatory for now, but so long as the dots continue to connect, we'll at least give this one a fighting chance at materializing.

[Via Joystiq]

[andrew @ Sep 6th 2008 8:15AM]

IMO I don't like the sound of this. Larrabee is untested in the real world and Microsoft should not let it be on the next Xbox. Think we should stick to ATI and Nvidia as they have been well proven, it could be suicide for the Xbox.

[CJ @ Sep 6th 2008 8:36AM]

Well, considering that the timeline for the Xbox 3 (or what have you) is not even fleshed out, there will be a decent amount of time to test and re-test Larrabee to it's full extent

Considering the weight the Xbox has in modern console gaming, it would be very loathe of Microsoft to put a sub-parr chip in what should thoretically be a much more powerful console.

[Daza @ Sep 6th 2008 9:20AM]

The original Xbox has an Intel 733MHz Celeron-based CPU, and I'd say that things worked out pretty well with that.

The fact that it's a GPGPU means it can handle the graphics and video related tasks but also handle computational tasks like a traditional CPU. The thought of having both GPU and CPU combined could also reduce costs significantly, as well as space, heat and power requirements (all big factors when considering a chip for a console).

Since it's untested and unproven, it's hard to say what it will perform. But with Intel hoping to launch by the end of this year, we'll have enough time to gauge it's performance. If Microsoft are looking towards a 2010 launch for the next-gen Xbox, there's still plenty of time to suss out what to use. Plus, with Larabee launching soon there should be plenty of stock and prices should settle by the time Microsoft are thinking about the new console.

[JohnTitor @ Sep 6th 2008 11:59AM]

how well did it work?
it was expensive to make, and easy for modders
a GPGPU is a GPU acting as a CPU not the other way around, on the original Xbox there was a modified Celeron AND a modified nVIDIA GeForce 3 for graphics
all current gen systems with an IBM CPU and either ATi or nVIDIA graphics, and if they want easy backwards compatibility they'll probably build on that

[Zeus.:God @ Sep 6th 2008 3:56PM]

Sorry, but I must correct you on one thing... The original Xbox's CPU was based on the Pentium 3, not the Celeron, but yes, it did fare very well.

[Plothole @ Sep 7th 2008 12:22AM]

@Zeus.:God

The Xbox had a stripped down Pentium 3, otherwise known as the Celeron.

[Platinum_Skeet @ Sep 7th 2008 4:28PM]

I say throw this in with a good GPU and you'll have the power of having 2 cards...

[white_ultras @ Sep 6th 2008 8:20AM]

"logical it sounds" lol ?

Larabullshit is non-existent tech and Intel has a long trackrecord of previous GPU failures with architectures like Intel-GMA, they have proven many times they can't really compete with nVidia or ATI.. with GPU stuff, they are only best in CPU due to x86 patent entrenchment... get a clue

[indeed @ Sep 6th 2008 8:56AM]

You are an idiot. Intel's integrated graphics do what they're supposed to do, they're a cheap and good enough solution to 90% of their market.

And, if you had read a bit about Larrabee, you'd know that it has a lot of potential and flexibility.

[matt merritt @ Sep 6th 2008 12:37PM]

Jen-Hsun Huang?

You post on Engadget now?

[sam @ Sep 6th 2008 8:22AM]

ms should wait for 6 months after larrabee has been released until they decide, no point say yes to unproven technology. We dont want RROD with next xbox!

[Dusty Nipples @ Sep 6th 2008 12:03PM]

Come on, they will come up with another way to tell us our console has died in the next gen. Maybe a sprung fist with a boxing glove on it straight to the groin.

[Ignatius @ Sep 6th 2008 1:25PM]

Well, at least it'll be less painful than seeing the RROD.

[Eric @ Sep 6th 2008 1:53PM]

The RRoD on the 360 isn't from chips dying. I believe it was mostly from bad solder connections. Still using Larrabee would be quite a risk because we don't know how it performs.

[Zeus.:God @ Sep 6th 2008 3:57PM]

Wow, you're all wrong. The RRoD is caused by overheating in which the solder softens and the mounting of the heatsinks pulls them away from the board as the board flexes from heat. It's not bad soldering, bad chips, or a bad mainboard, it's poor cooling- that is really the only flaw of the 360.

[Another take @ Sep 6th 2008 8:37AM]

I don't know much about how good this product will be in comparison. But my take is that MS must consider strategic implications. If intel manage not ONLY to beat AMD but ALSO Nvidia the future might not look bright for consumers or MS. If Intel will become the ONLY choice in the gen after that MS will have a weak position against intel since they be the only ones to deliver. A deal with a smaller producer like AMD or Nividia will help keep the market competitive for a longer time since they will have market shares from the console market. I think MS should consider go all AMD since it's crucial that they survive for the future of the CPU and GPU market.

[Dirk Hooley @ Sep 7th 2008 1:58AM]

Oh you're absolutely right. We should all accept inferior tech so that someone else can enjoy better tech. MS should sign contracts with, and create profits for, the company in last place cause its important to the market. BULL SH!T If that is how the world works, i would like to announce my new lines of graphic processors. The Pentel #2! Capable of lowest resolutions of LD (low definition) graphics. They come in one color, a sort of yellow orange blend. I will be expecting MS to contact me shortly on the details of my contract for the next gen X-box.

[Kurian @ Sep 6th 2008 9:06AM]

I have this feeling that Larrabee is NOT going to be targeted at competing with nVidia or ATi. Its just going to be some thing that comes onboard and will be good enough for the idiots that buy GeForce 8600s and the like.

[Skyride @ Sep 6th 2008 9:19AM]

Larrabee is meant for that. Their project to make GPU's comparable to the ATI and Nvidia gaming cards is a totally seperate project. And like you say, it probably will do onboard. They have already taken steps to ensure the ease of that with 1160 socket (the lower end nehalem socket).

[Skyride @ Sep 6th 2008 9:24AM]

The suggestion of this is silly. GPGPU on a gaming console? pfft. It wont happen on the xbox 3. the PS3 has a geforce 7800GTX in it (question me if you want but you know im right) that is connected directly to the cell processor. They both share memory. The PS3 already essentially uses what larrabee is designed to do except it does it on 3 chips as opposed to a single die. You simply could not make it work on 1 die due to the heat output.

[Bageloid @ Sep 6th 2008 3:18PM]

They are similar, but the RSX is less powerfull than a 7800GTX
http://en.wikipedia.org/wiki/GeForce_7_Series#GeForce_7800_GTX
http://en.wikipedia.org/wiki/RSX_%27Reality_Synthesizer%27

Less ROP's and almost half the memory bandwith(128 bit memory interface and slower speed gddr3)

[Zeus.:God @ Sep 6th 2008 4:00PM]

That, and the fact that the PS3 doesn't share memory, it has it's own dedicated sets of RAM. Oh, and if you mean directly connected by on the same mother board and "connected" to eachother in a fairly standard fashion, then yes, they are. If you were thinking something special, then no, not at all.

Also, I'm pretty sure Larrabee is supposed to combine the GPU and CPU functions onto one chip- not the RAM as well.

[d0x360 @ Sep 6th 2008 5:56PM]

The Cell and RSX do not share memory. They each have 256 mb of ram to use and can never dip into the others pool. On the other hand the 360 has unified memory which is shared by both gpu and cpu so they can divert as much or as little of the 512mb to either as its needed.

Oh and the RSX is based on the 7600 series so question I will my friend. There are plenty of technical documents that prove both your comments wrong.

[Zeus.:God @ Sep 6th 2008 11:17PM]

No, man, the RSX is, in fact, based on the 7800GTX's G70, but the only difference is slightly higher clocks and a rebadging to NV47 status.

[MarcBot @ Sep 6th 2008 10:08AM]

I'm sorry,but i bet 1 million dollars on microsoft it will happen. Will have to wait and see.
Will xbox 720 run directx 10? probably not..........

[Rollins @ Sep 6th 2008 10:41AM]

It'll probably have a modified version of DX11, to be honest.

[broli @ Sep 6th 2008 10:18AM]

This would be a dream come true for game developers. Not only do will they have extra x86 cores in there but they can write their own graphical pipeline from scratch to their own desires in any language they desire. Xbox was already easy to develop for but this will bring it to a whole new level. Sony just has to go for lerrabee as well.

[nerdinaction @ Sep 6th 2008 10:46AM]

Screw u console Noobs! us pc gamers are getting direct x 11.your way behind in tech. move along please.

[Phoenix @ Sep 6th 2008 11:17AM]

Erm... no.

[Konata @ Sep 6th 2008 11:09AM]

What about AMD Fusion? That could work also...

If you hadn't notice, most Intel integrated graphics aren't that good. I doubt they'd be able to play any, if none at all of the upcoming games.

If Intel does take over the market, i wonder what the prices will be...

[futurepastnow @ Sep 6th 2008 11:46AM]

And Larrabee has nothing at all to do with Intel's integrated graphics.

[Cal @ Sep 6th 2008 2:24PM]

Well the X4500 is only slightly worse than an 8400M G, so games will definitely be playable on that...

[decapitor @ Sep 6th 2008 11:10AM]

Seems like they would be better off pitching this to Sony seeing as how they are all about proprietary non x86 chips in their game consoles and M$ is all about ease of development and using pretty standard hardware.

[ethana2 @ Sep 6th 2008 11:44AM]

x86 isn't as open as you think.

I love the fact that game consoles are PPC. It helps keep certain groups of people away...

[ethana2 @ Sep 6th 2008 11:55AM]

Actually, the 360 uses PPC also so that's probably not what you were talking about.
Still, the CBE SPU ISA is a subset of PPC, which I think is licensed more openly than x86 is, so even while it's a bit weird, I don't know if I'd call it proprietary-- although almost everything is at /some/ point. You can't just go download the transistor maps for their chips yet.

[Labrador @ Sep 6th 2008 11:34AM]

This is very important for Intel and they have deep enough pockets to offer Microsoft a very sweet deal. It will happen.

[futurepastnow @ Sep 6th 2008 11:43AM]

People complaining about this don't understand. Larrabee is a GPU... but it is also an x86 CPU. A console using it would not need a separate processor and graphics chip. One hot little piece of silicon to handle it all.

That has the potential to be cheaper, especially after a couple of years. It greatly simplifies design, because only one hot chip needs to be cooled. And it means Microsoft would only have to deal with one vendor instead of two.

[ethana2 @ Sep 6th 2008 11:51AM]

I'd like a tiered approach-- Quad core nehalem, an array of 16 PIII equivalent cores, 64 high end shaders, and 256 simple shaders..

...then again, maybe for a developer that would be hades and is simply a Bad Idea. I don't know. --but if you use the PS3 as a desktop system you may have felt the effect of the fact that the cell requires code to be rewritten for it, which really puts a crimp in software development... It's part of why everyone wants to hack the RSX instead of writing an OpenGL driver for the CBE.

Things like PhysX and OpenGL are vital as we move forward, in part because they could allow for some substantial hardware modifications userspace software wouldn't even have to worry about. Native backwards compatibility and all that. --course, part of the complication for that was the fact that the PS2 was MIPS..

[John @ Sep 6th 2008 12:37PM]

There's really no special effort to running stuff on the CBE's PPC core (the PPE). Using the SPUs requires a lot of special effort, but since a 3.2GHz core is good enough for most apps you'd run anyhow, it doesn't make much difference. I do use my PS3 as a desktop machine (or, more often, as an SSH target), and it works great.

[nimblesquirrel @ Sep 6th 2008 10:47PM]

"People complaining about this don't understand. Larrabee is a GPU... but it is also an x86 CPU."

Umm... No. It seems you don't understand. Intel are designing this as a standalone GPU with a huge lean towards it being a GPGPU. Yes, it is based on the x86 instruction set, but it is based on the original Pentium design (P54C) and includes none of the features like branch prediction and out of order execution found in today's Core2 processors. Intel have also stated that the Larrabee is a GPU not a CPU.

[jaysins @ Sep 6th 2008 12:44PM]

Well I would expect them to get a second iteration of Larrabree because of the time line and who knows how much improved it could be. Also, they could use Larrabree as the cpu primarily and if need be dedicate a portion of it as a gpu when needed. Who knows how flexible the next generation of it will be. Though something like this does bode well with the current way the 360's architecture is setup. It's very unified compared to most other systems out there with the gpu and cpu sharing a memory pool and the gpu having direct access to the L2 cache of the cpu. Certainly something like this would be a big step forward in having a completely unified system and one allowing great flexibility to programmers and designers. Big question is, will it be ready in time? I can also see MS going with AMD/ATI and their fusion architecture if they can get it off the ground in time. Another note is that compared to today's cpu's the triple core one in the 360 is so slow for gaming. A game like crysis doesn't max out a current high end dual core, imagine a quad or octocore nehalem or whatever is next down the pipeline in one of these consoles. It would be so superior and if MS does go with a cpu and gpu from intel, you better believe they are going to get a good deal from intel as being put into a console maybe be the most painless way for intel to get developer support for their architecture.

[Todd @ Sep 6th 2008 2:50PM]

With the compatibility nightmare they had when switching between nVidia and ATI, I think MS would much rather stick with ATI. To tell you the truth I do not think either Sony or MS will shoot for a huge increase in power. Just look what both can do already. They just need a little boost to handle higher frame rates in 1080p and add some DX10.1 or 11 features for better graphics effects. I think the same will be true for the CPU. I do not think more than the 3 cores would be needed, but would not be hard to add. Maybe integrate a few more features and the chipset to the CPU

I think they are going to shoot for smaller packages, lower noise, lower heat, and less power draw. More than likely a big focus on features.... possibly DVR features, lots of internet connectivity... browsing, photo editing (with uploads to printers and social networks), video editing (with tools for uploads to MSN/YouTube/Google video sites), instant messaging, e-mail... Likely motion interfaces along with updated controllers. Also a smaller more stylish case.

I also expect MS to make a portable unit, but the big question is going to be what other features it has in it... like make it small enough to make it a cell phone and Zune too, or will it be a larger unit with HD PMP features and larger gaming capabilities. It would be nice if the portable unit worked well with the larger unit for transfering songs and movies and possibly saved games and profiloes to your friend's house.

[Yipcanjo @ Sep 6th 2008 8:41PM]

What if the next Xbox technology isn't a full-blown console like the Xbox / 360? Perhaps a handheld unit...

Just saying.

[nimblesquirrel @ Sep 6th 2008 11:40PM]

I can't see how Microsoft are going to throw billions of dollars on a GPU that is completely untried in the real world, especially one that is going to need specialised software development to produce games for.

With the Cell, the PS3 is already difficult to develop for, but if MS went for an equally difficult architecture it would put alot of pressure on game studios to develop for one or the other - something that MS is trying hard to stop with their efforts to get previously Playstation
exclusive titles in the Xbox.

MS have already had huge issues with heat and the RRoD with the existing XBox360. Larrabee has the potential to make this even worse: the current chips (fabbed at 45nm in their labs) are reported to have a TDP as high as 300W, much higher than any or ATI and nVidia's offerings (currently fabbed at 55nm).

Aside from showing data on how well the performance scales as more cores are added to the die, there has been no real performance data for the Larrabee. There are suggestions that it preforms as well as current offerings from ATI and nVidia, but that is what they are currently selling, not what is hidden away in their labs.

There is only one area where the Larrabee hopes to outperform ATI and nVidia's current offerings, and that is true ray tracing as opposed to rasterized rendering. However, both ATI and nVidia are already researching ray tracing (and have already demonstrated it in current hardware) and it is highly likely they will have something ready by the time Larrabee ships.

It is possibly that Intel are begging MS to put the Larrabee in the next XBox, but Intel really do have to be careful about how they do this, as they still have an Anti-Trust case to settle. If they start paying MS to take their chips, or sell their chips to MS at a loss, I'm sure ATI and nVidia will be crying "foul" to the DOJ. I'm sure MS do not want to be dragged into another anti-trust/anti-competition case in either the US or the EU.

[Meatgortex @ Sep 7th 2008 2:42AM]

It seemed obvious from the first LRB papers that the chip was targeted towards getting into a console. An MS console would make the most since due to the strong existing relationship and the use of intel hardware in the original xbox.

For those complaining about the GPU being unproven. Outside of the original xbox, consoles always have unproven tech. The 3xPPC setup in the 360 was new, as was the GPU design. Cell was new, and all the previous generations were custom chips as well. This is why there tends to be a ramp up in game quality over the life of most consoles as developers come to terms with what the machines can and can't do well.

Default interaction with LRB won't be that difficult as it will have a DirectX wrapper, again making it a good choice for the next MS console, however once people get comfortable the programmable nature of the chip means there are lot more viable options for different rendering techniques.

[Alvira Khan @ Sep 7th 2008 9:41AM]

If there is an opportunity to improve and enhance existing technologies, appropriate research should be carried out to investigate the possibilities. The eventual reward could outweigh the initial cost.

Alvira Khan
Florida Atlantic University
FAU Alumni

[Jeffrey @ Sep 7th 2008 7:28PM]

Larrabee in next XBOX? Good chance it's YES.

As an Electronics Technician, computer/gaming nerd and amateur physicist when I first heard about Larrabee I was extremely skeptical, but mostly because of misinformation of the Larrabee GPU being mode of entire CPU cores.

After reading all of the current relevent information I did a 180 in my opinion and now think it has incredible potential.

Untested? Yes and No. It isn't out there yet but the technology it's based on is well understood and games have already been modded to run on it. They have a pretty good idea of how it will scale.

Scalable? Yes. Very.

How is coding?
Writing code for this is easy. In fact, it's the biggest draw. This is very important as it not only brings down game development cost but it can decrease the time to market.

I was recently thinking, how would Intel bring Larrabee to market? It's supposed to be fully compatible with games but has to be specifically coded to for full optimization. Coming out on the XBox would be a perfect way to transition onto the desktop and servers.

In my humble opinion, Intel has the clout to get Larrabee working. The ease of coding will greatly assist its adoption and the ONLY cause concern is power efficiency.

I really like the fact that features that normally are done in hardware can be done on the Larrabee in software. This might seem like a step backwards but if it can be done efficiently it isn't. Physics encoding? Yes. Transcoding? Yes.

SLI and Crossfire.
Adding a 2nd card in SLI or Crossfire mode for more gaming power can be hit and miss. The game has to be specifically coded for it as well as having driver support. Larrabee on the other hand isn't quite the same. We can get 2X the performance with two cards. There's plenty of room for discussion here but the bottom line is games need to get away from SLI type coding and code so that GPGPU's like Larrabee's design can scale almost perfectly by adding more cores.

If Intel can get Larrabee on the XBox I think it has a strong chance of dominating the PC GPU market.

(Just to be clear, and it was mentioned before, Larrabee is a GPU. A traditional CPU will be in a console or desktop.)

[nimblesquirrel @ Sep 7th 2008 11:56PM]

How is the coding going to be easy? Intel haven't released an SDK for this yet, so there is no way to tell. Programmers today have a hard enough job effectiovely using two or four processors, yet alone a 32 core GPGPU. Just because it is an x86 instruction set, doesn't automatically mean it will be easy to program. Going massively parallel does seem to scale well, acording to Intel's figures, but the associated coding is by no means easy. Have a chat to a PS3 developer and ask about how hard it is to program for a Cell processor and it's 7 SPEs.

Offloading hardware specific features into software is an interesting way of doing things, and it is clear that Intel intend the Larrabee to be more than just a GPU. However, this extra layer of software translation comes at a significant performance hit, and adds another layer of complexity to coding.

As far as SLI and Crossfire, the performance would not double. SLI or Crossfire does not allow the Larrabee chips to share their coherent cache. It is that coherent cache that is a key part to the scalability of the cores on the Larrabee die. Because the chips do not share that high speed cache (and associated interconnects), and because there is some overhead to both SLI and Crossfire, the performance will not simply double.