Sign in with
Sign up | Sign in
Your question

So, Xbox 720: Larrabee Inside?!

Last response: in Graphics & Displays
Share
August 6, 2008 2:26:24 PM

http://www.fudzilla.com/index.php?option=com_content&ta...

Quote:
The latest talk that we received from trusted sources is that Intel is considering the possibility of trying to reach out to market a variation of Larrabee’s architecture and design that would be targeted at the next generation of gaming consoles, and specifically at the one already known in design discussions as the 720 numbered unit.

The time frame for release of Larrabee does fit the timeline that a new console might be on the drawing board, but while these whispers have started, we just don’t think that Intel will want to disrupt what they are doing to produce a specialized version of Larrabee targeted at consoles.


Interesting times, indeed.
August 6, 2008 3:32:47 PM

Intel sure seems like they're trying to be a good old fashioned monopoly.
a b U Graphics card
August 6, 2008 10:52:12 PM

IMO this doesn't make much sense shorter term, but would make more sense longer term.

Currently M$ is the keeper of DX, and Intel's plans upset that dominance with their Ray-tracing view of the future.

Eventually M$ mmay go that route but it doesn't seem that it would be before 2011+ if 2010 is set to be all about DX11.

I would suspect that the other two conteders are better suited for Larabee than M$.
Related resources
August 6, 2008 11:54:04 PM

TheGreatGrapeApe said:
IMO this doesn't make much sense shorter term, but would make more sense longer term.

Currently M$ is the keeper of DX, and Intel's plans upset that dominance with their Ray-tracing view of the future.

Eventually M$ mmay go that route but it doesn't seem that it would be before 2011+ if 2010 is set to be all about DX11.

I would suspect that the other two conteders are better suited for Larabee than M$.

It could make porting games (in both directions) easier. With that in mind, it make sense for MS to adopt it as soon as possible.
a b U Graphics card
August 7, 2008 1:48:03 AM

However for M$ of all people they already have very easy porting using XNA (especially compared to the problems for the Wii and PS3), so I don't think that the loss of their HLSL and DirectX fiefedoms would be a good motivation to simply make cross-platform development a little easier for... anyone using intel.... but not necessarily M$. This would also open the door for other compilers and other apps further challenging M$' role.

It would be like intel adopting someone else's processor IP as a basis and giving up on their X86 dominance, instead of keeping your dominance while forcefully moulding the future.

I see it as a long term potential for M$, however abandoning their DX domination so easily on platforms they control doesn't make much sense, and to me it makes more sense for those that see M$ & DX as a barrier to entry to open up to Larrabee. Considering the talk of 8-10 year lifecycles for the consoles, perhaps this is talk about the future of the future, and thus a 2012 and beyond idea.
August 7, 2008 3:34:30 AM

With the massive, uncanny and unstopable momentum that Nintendo has right now, I think they would be the ideal participant. It could serve as a "told you so" for both companies. Intel proving they can deliver on graphics on a non-"**** onboard" method, and Nintendo proving to the world that it can still deliver a console with top tier HD graphics. Throw refined Wii controls version 2 in there and you got some money.
August 7, 2008 11:03:09 AM

Cool stuff. Larrabee is promising.
August 7, 2008 12:45:19 PM

TheGreatGrapeApe said:
However for M$ of all people they already have very easy porting using XNA (especially compared to the problems for the Wii and PS3), so I don't think that the loss of their HLSL and DirectX fiefedoms would be a good motivation to simply make cross-platform development a little easier for... anyone using intel.... but not necessarily M$. This would also open the door for other compilers and other apps further challenging M$' role.

It would be like intel adopting someone else's processor IP as a basis and giving up on their X86 dominance, instead of keeping your dominance while forcefully moulding the future.

I see it as a long term potential for M$, however abandoning their DX domination so easily on platforms they control doesn't make much sense, and to me it makes more sense for those that see M$ & DX as a barrier to entry to open up to Larrabee. Considering the talk of 8-10 year lifecycles for the consoles, perhaps this is talk about the future of the future, and thus a 2012 and beyond idea.

Nothing stops them from using Larrabee AND sticking with DX. Actually they could use that to build a strong DX base for Larrabee which would reinforce their position on the PC market. MS has absolute control over all things that are released for the Xbox. Absolute.
Anonymous
a b U Graphics card
August 7, 2008 3:22:46 PM

looks like this upcoming console war in 2 years will be bigger than most... Playstation will have a lot to prove... it better have a better game line up this time or it won't take off

COMPARED to the 360... the playstation 3 had very respectable sales and even out sold the 360 in the last quarter... but it pales in comparison to the overall sales of the 360
a b U Graphics card
August 7, 2008 4:42:37 PM

Slobogob said:
Nothing stops them from using Larrabee AND sticking with DX. Actually they could use that to build a strong DX base for Larrabee which would reinforce their position on the PC market. MS has absolute control over all things that are released for the Xbox. Absolute.


Yeah, but that to me is why it would be the LONG Term strategy. There I agree, and that's where I think it would work well, but IMO that means either that A) This X720 is 3-4 years off, or B) They are talking about the succesor to the succesor if anything is about to be released. Either way IMO we're talking about a mature Larrabee part after the spread of DX11. There is little chance that M$ will support Larrabee and a turn away from DX11 before either launches. Right now like I said, it makes more sense for M$ to weild their mono-power and shape the Larrabee future, whereas jumping into it right now would jeopardize their position and put all the power in intel's hands.

IMO, this is probably a rumour of the type that hapens when someone asks;
"Hello MS, are you going to replace the Power PC core of the X360 with something more adaptable, scalable and more multi-cored than the current 3HT?"
MS - Probably somethig similar cutting edge of the time.
"What about intergrating Larrabee?"
MS - We are considering everything, all options are open at this point


"Hello intel, what are you plans for Larrabee outside of desktop computing/graphics?"
intel - Well we're loking at server applications and thin clients, and any other opportunity that we thiink would be a good match for our solution(s).
"What about in in consumer appliance area, like media players, or game consoles?"
intel - Sure we are open to any opportunities that present themselves to us.
"Even something like an Xbox 720"
intel - Certainly, if there is a role for our product in that area we would definitely be willing to explore that opportunity.

Etc....

Right now I'd say it's something talking about gen 2 Larabee solution and a console product way off into the future.
August 7, 2008 4:52:23 PM

+1 for TGGA.

This is a no-brainer. M$ wont migrate to a GPGPU or any other hardware that only runs well on "his" propietary code, and just cuts backwards compatibility. This of course knowing that M$ develops DX. Thinking Larrabee for the next gen of Xboxs is just wrong in too many ways.

I might see Sony doing it faster than MS. Cell is cool, but it is horrible to code for it.
a b U Graphics card
August 7, 2008 6:55:47 PM

Yeah, short term Larrabee to me would work against the benefits of XNA not for it. To me M$ jumping into Larrabee before seing a functioning first gen actually do some real gaming is questionable.

Even if you consider the Xenos as a radical design departure it's still using the traditional components and developed by a company with the proven expertise, whereas intel still has yet to prove that Larrabee isn't more than just an interesting idea. It's not overly radical and has a good fallback strategy, but where you want to get maximum performance out of a smalll efficient package, it's not looking like a good solution. Larrabee is slated to be more power consumptive than a GTX or HD4KX2, but that's no good for a console. They can't have them running 500W and hot like a PC, they've learned that from the X360 / PS3 problems.

On so many levels it seems to simply be a strange choice near term.

An offshot product that is more Fusion combining parts of Larrabee with with multicore terascale stuff, that has variable loading options, does make sense for a far future product from both production (reduce costs, shrink all parts per refresh) and flexability issues, but that's one future the PC starts preparing and maturing the software in the near term.
a b U Graphics card
August 7, 2008 7:08:59 PM

Quote:

COMPARED to the 360... the playstation 3 had very respectable sales and even out sold the 360 in the last quarter... but it pales in comparison to the overall sales of the 360


Yeah, X360 sales are slowing, but when you have such a huge lead that's not as much an issue. 350,000 vs 300,000 sale doesn't meant much if overall you're 10 vs 3 million. Install base is still huge.

The most surprising of course is the incredible success of the Wii, which has outdone both competitors in both areas.

I always thought they would have good sales as most people's "Second cheap console" bought for legacy and quirky 'feel good' reasons (like Mario,Zelda,etc fans), but what surprised the heck out of me was sales of the Wii as people's only console.

The only negative if you had to look for one on the Wii seems to be the low game purchase rate on Wiis where people buy a few specific titles, but nowhere near as many per console as the X360 & PS3.

I agree though, next time around the tables have totally turned (except a little less so for M$ who's still in the Middle IMO), Nintendo has this massive hit that they can be proud about despite lower specs, and never need to justify their spec choice again, nor do they need to worry about selling on specs either. M$ is once again in the position of returning competant #2 player with solid specs and a growing solid cred in the market. Sony this time takes on the role of trying to bring out a console to restore their name in gaming (selling it as a cheap BluRay player doesn't convince people you're serious about gaming), Sony is in Nintendo's role from last time, where they underwhelmed, have loyal fans, but really need to recapture those old fans and also make something compelling to get people interested in their product and games again. Overall I'd say this puts all 3 in a very balanced situation for the next console battle, and while the X360/PS3 speculation was incredible last time, it's likely to be worse this time since the rules and roles have changed considerably.
August 7, 2008 11:30:50 PM

TheGreatGrapeApe said:

Right now I'd say it's something talking about gen 2 Larabee solution and a console product way off into the future.

Agreed. Buying something that is not mature and not tested does not seem smart. I've a feeling though that Larrabee will implement DirectX with a "simple" driver and with that all intel needs to do is show MS their cards. I don't know how likely that is, since Larrabee is their "secret" weapon right now. It could make the difference though and i could actually see intel selling them for cheap just to get a greater foothold into the graphics market. That's all speculative though.
August 8, 2008 5:37:09 AM

Crucially, 360 gamers buy a far higher nummber of games on average, something like 7 per console.

PS3 and Wii average around only 3. I have a massive 3 for my own PS3.
August 8, 2008 9:14:56 PM

I dont think there are more than 3 games for the PS3.
a b U Graphics card
August 8, 2008 9:23:21 PM

spathotan said:
I dont think there are more than 3 games for the PS3.


SURE THERE ARE.... but only if you have the PS3 with the hardware compatability chip for PS-One/PS2 games. [:mousemonkey:1]
August 9, 2008 8:16:56 PM

Let's see Larrabee first. :) 
Nowadays, I'm now trying to get the second HD 2600XT on my computer to work on Ray Tracking! :D  It does, indeed. Given the scene isn't very complicated and the resolution isn't more than 1024x768. I know it's a little low but hey, my card is also low! :D 

PS. I got the CTM, stream processing SDK for ATi/AMD! CUDA is sleeker. ;)  Let's see Larrabee, when it comes.
a c 130 U Graphics card
August 9, 2008 8:29:29 PM


From things i have read and i dont quite understand it fully so please put me right where needed. But as i said i have read in the news section and on other sites that CUDA and Larrabbe wouldnt have a problem running more than one API.
Intel say it will be fast even with out optimisation. Also it depends who you talk to but each camp thinks the other will have problems with x86.
So what i was wondering was why cant you run multiple drivers same as you do Operating Systems then there wouldnt be a compatability problem ? Is it really so hard to have differant drivers loaded and the card use whatever one the game is coded for ?
Mactronix
August 9, 2008 8:32:16 PM

Uhm. mactronix, have you ever tried putting an nVidia and an ATi card on the *same* computer?
a c 130 U Graphics card
August 9, 2008 8:40:01 PM


Sorry i didnt mean both cards i meant for example putting a larrabbe card in a machine and having drivers for Raytracing and open gl and DX so it could do what ever ? I would say a driver that can use each but that seems harder some how.

Mactronix
August 9, 2008 8:52:48 PM

Ah! Yes. It's possible.
Now we have both DX and OGL drivers in our computers. But I'm still sceptical about the "proposed" performance of Larrabee. That's why I'm on a "wait-and-see" stance.
BTW, please don't call it "raytracing". What's implemented on computers is "raytracking" - same result but totally different fundamentals.
a c 130 U Graphics card
August 9, 2008 9:09:43 PM

duzcizgi said:
Ah! Yes. It's possible.
Now we have both DX and OGL drivers in our computers. But I'm still sceptical about the "proposed" performance of Larrabee. That's why I'm on a "wait-and-see" stance.
BTW, please don't call it "raytracing". What's implemented on computers is "raytracking" - same result but totally different fundamentals.


Fair enough as i said i don't fully understand it, can you link me to something that explains it/the differences. I vaguely remember an article that mentioned it being tracking and not tracing but that was a while ago. I just did a quick google but got nowhere with it. So if you have a link or some such i would be grateful.

Mactronix
August 10, 2008 10:06:02 AM

You can check wikipedia.
http://en.wikipedia.org/wiki/Ray_tracing_(graphics)
It has some mistakes, but you can also check the references part of the article.
There's also a good book on rendering "2D projection of 3D geometries" it has lots of good algorithms on the topic, if you're interested. :) 
August 10, 2008 1:48:33 PM

larrabee ‘s power consume maybe burn 720 into the hell haha
a b U Graphics card
August 10, 2008 3:01:07 PM

Meh.... It's from Fudzilla.....
August 11, 2008 12:55:46 PM

TheGreatGrapeApe said:
Larrabee is slated to be more power consumptive than a GTX or HD4KX2, but that's no good for a console. They can't have them running 500W and hot like a PC, they've learned that from the X360 / PS3 problems.

I'm a bit surprised that no one else but you mentioned this, and even you only mentioned this in passing, the main reason I remain highly skeptical of Larrabee...

From what we know, not only are the P4's designers working on the chip, they also seem to be following the same strategy of starting with a mainstream chip that will run at absurdly high clock speeds while consuming copious amounts of power. As you mention, that's useless for a console; the Xbox 360 with the original 90nm parts was infamous for heat-induced problems at only around 180w, which is apparently less than what the first-generation Larrabee is designed to put out... And while Larrabee makes Intel look like they're trying to do everything, it still doesn't incorporate general system I/O like input and networking, as well as lacking a hard drive or optical drive.

And really, given the immense power consumption of the GPU and its focus on being huge and hot, I can't really see it having a place in a Nintendo console; the Wii supposedly burns only a mere 18 watts of power, (which I suspect may be a rating that doesn't include the 8-12x DVD drive spinning at full speed) and at any rate, has no stability issues with a power supply designed to deliver a mere 54 watts maximum.

Plus, I suspect that Nintendo may lock into ATi as their graphics provider, in order to ensure legacy support for Wii and GameCube games, though they may rule against it this time; the Wii's legacy support for the GameCube may likely have been included as incentive for GC owners to quickly replace theirs with a new machine. At any rate, I wouldn't be surprised to see the Wii's successor hold fairly close to the architecture that they've already had for the GPU in the GC and Wii. Now if only I could find out just what they'd actually changed when they went to the Wii; the thing appears to have at least a good 200% more transistors, so it's gotta have added SOMETHING. :p 

TheGreatGrapeApe said:
The most surprising of course is the incredible success of the Wii, which has outdone both competitors in both areas.

I'm not entirely surprised, though I'm slightly surprised that the Wii's North American sales appear to surpass its competitors so quickly. I knew that Japan was effectively lost to anyone outside of Nintendo, and they represent a good portion of the overall console market; 25-30% right there. With the PS3's design and underwhelming library, it was pretty clear that Nintendo would take the vast majority of that market, giving them the bulk of that 25-30% of the entire global market right off.

Also, remember that when Nintendo spoke of seeking a broader market than the PS3 and Xbox 360, that didn't necessarily mean abandoning mainstream or hardcore gamers. It seems that for a lot of their "new" market, they're drawing in people who'd previously been PC-only gamers. They already HAD rigs that left the PS3 and Xbox 360 in the dust, had been gaming online with TeamSpeak, and enjoying High Definition visuals for years, so it was impossible to sell them a console on those sort of things, which were the Xbox 360 and PS3's main selling points. However, the Wii tries for something DIFFERENT than what a PC could crush it easily at. Plus, it's cheap.

spathotan said:
I dont think there are more than 3 games for the PS3.

Just two: Talladegga Nights, and Metal Gear Solid 4's half-hour installation. :kaola: 
a c 130 U Graphics card
August 11, 2008 5:27:39 PM


You make some very good points there nottheking. Console hardware isnt relay an area i have done much research in and i am finding this thread very interesting and educational.
I do agree that it wouldn't make much sense for the wii to adopt a different GPU solution as it seems to be doing very well as is and as you point out its the interface that has been the thing with the wii. Its fun and a good laugh after a few beers. :lol: 
Im not up to date on what add ons or other abilities it has though. I know there are various driving wheels and the wii fit etc but does it have any HD capabilities or plans for it ?
If so then it would seem to be the perfect solution for almost all needs, Fun for the kids, party entertainment for the grown ups and if you can add to that a half decent Blue ray player like the PS3 then it would end up in just about every house that could afford a console.
Larrabee ? no dont think so not this time around.

Mactronix
a b U Graphics card
August 11, 2008 10:01:38 PM

nottheking said:
I'm a bit surprised that no one else but you mentioned this, and even you only mentioned this in passing, the main reason I remain highly skeptical of Larrabee...
...
And really, given the immense power consumption of the GPU and its focus on being huge and hot, I can't really see it having a place in a Nintendo console; the Wii supposedly burns only a mere 18 watts of power, (which I suspect may be a rating that doesn't include the 8-12x DVD drive spinning at full speed) and at any rate, has no stability issues with a power supply designed to deliver a mere 54 watts maximum.


Yeah, to me it would be an adapted versioin of the launch model, IMO something that has 4 units ( beside a small efficient multi-core CPU on the same package, and potentially even the same die.
But yeah it would need to be a mature and efficient application, especially to be a Wii like product. Howdver, another approach could be for Nintendo to grow the legacy Wii and to split it's market further with another product. They are not averse or strangers to doing this in the handheld department, with the DS and the GBA still chugging along.

Quote:
Plus, I suspect that Nintendo may lock into ATi as their graphics provider, in order to ensure legacy support for Wii and GameCube games, though they may rule against it this time; the Wii's legacy support for the GameCube may likely have been included as incentive for GC owners to quickly replace theirs with a new machine. At any rate, I wouldn't be surprised to see the Wii's successor hold fairly close to the architecture that they've already had for the GPU in the GC and Wii. Now if only I could find out just what they'd actually changed when they went to the Wii; the thing appears to have at least a good 200% more transistors, so it's gotta have added SOMETHING. :p 


Well there is more on die memory and the addition of some new I/O and DSP stuff, but yeah the information is very scant. Heck, there's not very much information on even just the Gamecube's Flipper chip, let alone the Vegas portion of Hollywood. I'm not sure that Nintendo would need to keep ATi, however it would likely be a wise choice to keep the IP 'in-house'. However you could probably do the TEV functions on the larrabee ALUs.

Quote:
I'm not entirely surprised, though I'm slightly surprised that the Wii's North American sales appear to surpass its competitors so quickly.


I'm not surprised that they did far better than the nay/doom-sayers were predicting, but I am totally surprised that they ended up #1 overall and not #2 as I mentioned I thought they'd do the '2nd console role' and then there'd be the obvious loser of this round. I think without the BluRay Sony definitely would've owned that title, now they're simply stuggling but till selling (no dreamcast this round).
August 12, 2008 4:37:06 PM

TheGreatGrapeApe said:
Well there is more on die memory and the addition of some new I/O and DSP stuff, but yeah the information is very scant. Heck, there's not very much information on even just the Gamecube's Flipper chip, let alone the Vegas portion of Hollywood. I'm not sure that Nintendo would need to keep ATi, however it would likely be a wise choice to keep the IP 'in-house'. However you could probably do the TEV functions on the larrabee ALUs.

Yeah, what little I can determine about Hollywood is very, very little. As far as I can gather, I've mostly decided that all I can perhaps say with any degree of accuracy about Vegas (provided I even have the names attached to the right dies!) is that it's primarily made up of 24MB of EDRAM that functions, in many ways, similar to how the main memory does toward Flipper. Also, I've thought that perhaps, like the Xenos "daughter die," the ROP structure is located there. Given that the resolution isn't increased much, and the Xbox 360 and PS3 get by fine with the equivalent of 8 ROPs, that there are only 4 present for the Wii, the same number as with the PS2, Game Cube, and Xbox. I suggest that the ROPs would likely be located there since on the GC, it was that bank of memory where it appeared to write the first buffer, and then flipped that to its on-die buffer, hence being double-buffered; it would make sense to just put them on the same chip.

What really gets me is the non-memory part of it... All I've learned is that, by size and process, is closest to RV530; not much else I can say. There is, of course, the mention of the "TEVs," Though I don't know how flexible they are, and what, if any, programmable pixel shader capacity is present. I know that at the very least it has some degree of flexibility, as evidenced by the appearance of shader effects like normal-mapping, specular-mapping, and bloom in games like Super Mario Galaxy, Need for Speed: Pro Street, and The Force Unleashed.

TheGreatGrapeApe said:
I'm not surprised that they did far better than the nay/doom-sayers were predicting, but I am totally surprised that they ended up #1 overall and not #2 as I mentioned I thought they'd do the '2nd console role' and then there'd be the obvious loser of this round. I think without the BluRay Sony definitely would've owned that title, now they're simply stuggling but till selling (no dreamcast this round).

At least to me, I had predicted that because this generation of consoles saw a harsher economy than the previous two generations, that price would weigh in more heavily than before, hence making the Wii's relative price a bigger selling point than it would've been in years past. 1995-1996 were very good times, economically, so the then-$300US launch tag of the Playstation didn't really hurt it, especially since at the time CD-players were not as widespread as they were around 2000, or MP3 players today.

In 2000, though there were some economic troubles there, they did pass, and the PS2's sales price didn't sting so much because it was also a DVD movie player. Unlike Blu-Ray, DVD offered a slew of absolutely compelling reasons to replace the old format; while everyone's quick to think about how crisp 1080p with 7.1 surround is compared to what DVD can do, one can just as quickly realize that this is practically irrelevant compared to the size and cost savings of DVDs compared to VHS, coupled with the ability for random-access, eliminating the need to rewind them, and by being digital, allowing for options like subtitles in multiple languages on the same disc, as well as generally being more reliable, lacking the fatal vulnerability to magnets, fingerprints, etc., while being, again, much smaller, too.

Yeah, again, I predicted a landslide in favor of the Wii in Japan, though I'd actually predicted the Xbox 360 would hold at least a small lead in North America for most, if not all, of its lifespan. Overall, I predicted a win for the Wii, though I honestly admit that I didn't think it'd be quite this high; I was originally estimating that the market split would be along the lines of 37.5/32.5/30.0% or so, while here we're seeing closer to 50/25/25.
!