Larrabee = Fail?

so its a failure since it was delayed to 2011? it didnt even have a set release date to begin with.

these things take time. intel doesnt magically come up with an archi in one day then release then next.

base failure on performance numbers, not on dates.
 
Larabee is interesting for sure. We will have to see how fast it is when it comes out...
 
Intel is throwing out a lot of propaganda about how much better it will be than regular GPUs, and the GPU makers are throwing out a lot of FUD saying Larrabee will never catch up to the GPUs.

Intel is going down a completely different road than the GPU makers are. It could be disruptive technology that catches nVidia and ATi with their pants down, or it could be a huge fiasco and money pit for Intel. Only time will tell.
 
Menelmarar said:
Seriously? Why are you regurgitated that speculation BS from Fuad? It's not from Intel.

Intel hasn't set a date.

Be patient. Nothing will be known until third party sites have review samples.
Click to expand...

This.
 
enzolt said:
base failure on performance numbers, not on dates.
Click to expand...

Well the reason to suspect it'll be a failure is Intel's information they have released seems to indicate it competes well with current graphics cards. Ok... Well that won't cut it in 2011. We have new graphics cards on the way here in the coming months from both companies that promise to be a good bit above what is current state of the art. They also aren't going to sit idle in 2010, and by the beginning of 2011 we'll have seen at least a minor improvement in performance.

So unless Intel is changing things around, or they have been vastly understating their performance claims, a 2011 launch time isn't looking good. It could work, but only if their cards are quite cheaply priced and thus competing with more midrange/budget products.

You can't get in to the graphics market with a card that is both expensive and underperforming. While ATi or nVidia might be able to sell some like that on past reputation, if you are the new kid you've got to prove yourself.

This was precisely the problem that companies like S3 and even Matrox had. Their products didn't compete with the big two's high end performance, and didn't compete with their midrange prices.

As with anything, we'll have to wait and see, and this is hardly an official news source. However overall I'm worried about Larabee. The schedule of it has been slipping and that isn't a good thing in hardware. It isn't as though you can just redesign something to be faster in a month, hardware design is a lengthy process. So if their performance targets were 2009 GPUs and they don't get it out then, they are going to be behind.
 
At this point Larrabee deserves the title of 'vaporware'. So far there is little to suggest that it'll ever be launched, let alone be successful.

Looking at Intel's current offerings, including the X3100 and X4500, one can't help but feel thoroughly unimpressed. Its IGPs get half the framerate compared to offerings from nVidia and AMD, and many operations still seem to be implemented in software rather than hardware, completely killing performance in certain games.

Intel's best GPU was the i740, and it wasn't even developed by Intel. I must admit that Intel's IGPs have seen a marked improvement since the days of GMA 950 (yuck), but they're still light years from where its competitors are. Larrabee is also an unproven concept (many in-order x86 cores employing an extended x86 instruction set), and we all know how well its last concept (Itanium) ended up.

Intel is best at improving upon existing technology, like they did with Core (2) and i7. They do get caught out of left field quite often, though, such as with x86-64. In that sense Intel is like Microsoft, good at copying and integrating ideas, but as agile as a sack of bricks hurtling down towards the ground. Let's just hope Larrabee doesn't become the next Microsoft Bob :p
 
i have read on it, and can only say Intel is trying a different approach than what ATI and nVidia is/will. Intel boasts and nVidia/ATI will rebut with their spiel.

the thing to me is, the two main GPU heavies are also chipset and platform adversaries (not catagorically but the overlap of silicon is complex). does anyone really think Intel is going to release a product that devastates the market share of either nVidia or ATI because of its rendering/3D? when Intel entered into this on previous occasions, they had different mix of motivations, and i am curious as to how much grey matter and cash is being pumped into Larrabee (whats the team, and what are they shooting for in terms of performance). BUT...there is also politics of business involved and everything until production silicon means nothing, if it is a competitive product, then all the better for us. more competition is a good thing...now hurry up and wait.
 
Not only are you regurgitating bullshit from fuad, but you aren't even reading it properly. This bullshit is about a Larrabee IGP. He is saying that the ONBOARD INTEGRATED version of Larrabee won't come out until 2011.
 
a lot can change at any rate. Intel has a LOT of resources to throw at this issue. to say they are out of the running is pretty premature. All the same a working product even a generation behind would let them get in the market
 
Hmm.. Larrabee IGP. Me wonders if such a product could boot without a discrete CPU. Why not make the north bridge a system on chip? Maybe integrate an Atom core. By 2011 todays current single core Atom should be a trivial part of the transistor budget.

Think of all the overclockers who could use such a feature to get on the internet and order a replacement CPU for the one they just fried.
 
2011 isnt really that bad. Think about it, its already almost 2010. You think of intels tick-tock plan with processors. We have to get a tock on the i7, and then a tick will be 2011 probably? I havent looked at the roadmaps in a while so dont quote me and link me to a Q4 2010 tick roadmap and call me a tard. Maybe theyre waiting for new procs, new mobos, chipsets and so on.
 
I hear Larrabee will be bundled with a free game, Duke Nukem Forever.
 
Cloud15x said:
2011 isnt really that bad. Think about it, its already almost 2010. You think of intels tick-tock plan with processors. We have to get a tock on the i7, and then a tick will be 2011 probably? I havent looked at the roadmaps in a while so dont quote me and link me to a Q4 2010 tick roadmap and call me a tard. Maybe theyre waiting for new procs, new mobos, chipsets and so on.
Click to expand...

Other way around... i7 is a tick (new arch), 2011 will be the tock (die shrink). Even then, however, I don't see Larabee being a complete failure. Even if it isn't as good as the competing offerings, that hasn't stopped them before, and any improvement to the GMA X4500 is a good thing!
 
Just some stuff read from a chinese forums so take it for what its worth:
According to some people claiming to be deving for Larrabee equivalent to CUDA/Brook+ called Ct (meant for multi core in general since the gpu is a bunch of x86). Right now the devs don't have working hardware to test on. They compile the code to run on the GPU, send it to Intel who then run and return the result. Full dev kits with working hardware for the devs aren't coming until December, with public release around mid 2010.
 
Sycraft said:
Well the reason to suspect it'll be a failure is Intel's information they have released seems to indicate it competes well with current graphics cards. Ok... Well that won't cut it in 2011. We have new graphics cards on the way here in the coming months from both companies that promise to be a good bit above what is current state of the art. They also aren't going to sit idle in 2010, and by the beginning of 2011 we'll have seen at least a minor improvement in performance.

So unless Intel is changing things around, or they have been vastly understating their performance claims, a 2011 launch time isn't looking good. It could work, but only if their cards are quite cheaply priced and thus competing with more midrange/budget products.

You can't get in to the graphics market with a card that is both expensive and underperforming. While ATi or nVidia might be able to sell some like that on past reputation, if you are the new kid you've got to prove yourself.

This was precisely the problem that companies like S3 and even Matrox had. Their products didn't compete with the big two's high end performance, and didn't compete with their midrange prices.

As with anything, we'll have to wait and see, and this is hardly an official news source. However overall I'm worried about Larabee. The schedule of it has been slipping and that isn't a good thing in hardware. It isn't as though you can just redesign something to be faster in a month, hardware design is a lengthy process. So if their performance targets were 2009 GPUs and they don't get it out then, they are going to be behind.
Click to expand...

well theres adaptation. they are just testing waters with what their architecture could do. im pretty it was made to scale up if needed. just a few tweaks and changes here and there.
 
Omegas said:
I hear Larrabee will be bundled with a free game, Duke Nukem Forever.
Click to expand...

yeah you're funny :rolleyes:

they are actually using larabee with Project Offset, obviously.

sometimes being an EE college student has its perks, including peers who have interned for Intel ;)
 
enzolt said:
yeah you're funny :rolleyes:

they are actually using larabee with Project Offset, obviously.

sometimes being an EE college student has its perks, including peers who have interned for Intel ;)
Click to expand...

except I think everybody knew that
 
While 2011 seems far away to us, in reality it isn't that far off. Certainly performance at the time it is released will be compared to current GPUs of the time. Intel could say they are releasing it in 2025 and that'd be fine, we'd still compare it with what is out at the time. Perhaps Intel is thinking it will compete quite well with 2011 hardware, we'll know when it is time. Intel has not set any dates yet.

I will say this though, the darn thing is being hyped so much that impressions of it are already forming in peoples minds, and this isn't necessarily a good thing. Let's just relax and see what happens when its released, and base performance and price compared to what is there at the time.
 
im more hyped about ati's sept 10 and nvidias gtx 300 demos in late sept!!!!
 
I think you misunderstood. The way I read it was that the IGP incarnation of Larabee will be available in 2011. The discreet Larabee product should be earlier than that.
 
Brent_Justice said:
While 2011 seems far away to us, in reality it isn't that far off. Certainly performance at the time it is released will be compared to current GPUs of the time. Intel could say they are releasing it in 2025 and that'd be fine, we'd still compare it with what is out at the time. Perhaps Intel is thinking it will compete quite well with 2011 hardware, we'll know when it is time. Intel has not set any dates yet.

I will say this though, the darn thing is being hyped so much that impressions of it are already forming in peoples minds, and this isn't necessarily a good thing. Let's just relax and see what happens when its released, and base performance and price compared to what is there at the time.
Click to expand...

Intel did most of the hyping, claiming how it'd be a new revolution in GPU/IGPs. This compared to their GMA IGPs which they have never said anything about other than 'it's an IGP', basically. I had to look up what the current GMA IGPs are a few weeks ago because I didn't have a clue what Intel had released the past few years.

In general hyping sets a product up for failure or a quiet life in the margins if it can't live up to most expectations. I fear that it's already too late for Larrabee at this point, but we'll see. If it ever is released :)
 
Elledan said:
Intel did most of the hyping, claiming how it'd be a new revolution in GPU/IGPs. This compared to their GMA IGPs which they have never said anything about other than 'it's an IGP', basically. I had to look up what the current GMA IGPs are a few weeks ago because I didn't have a clue what Intel had released the past few years.

In general hyping sets a product up for failure or a quiet life in the margins if it can't live up to most expectations. I fear that it's already too late for Larrabee at this point, but we'll see. If it ever is released :)
Click to expand...

I am thinking that Intel is looking at this very long term, even if the first couple of rounds are losers in the gpu war they might not be in up and coming programs. Open CL is going to take off I think, or at least software based rendering and the like. I really would not write this off no matter how bad it might look first generation.
 
Do you think Intel would bother to get into the discrete GPU market if they couldn't make one as fast as their competitors? I think not.
 
Blackstone said:
Do you think Intel would bother to get into the discrete GPU market if they couldn't make one as fast as their competitors? I think not.
Click to expand...

Intel thought they could reinvent the desktop and conquer the server space with Itanium as well, then AMD came up with x86-64 and Intel took a nosedive straight into a thick layer of mud. Intel once thought they stood a chance in the GPU market with i740 as well (after buying the design from another company), yet quickly came back on that, choosing to focus on the no-expectations GMA900/950 instead.

I wouldn't assume too much.
 
Intel's SSDs sure are doing well though. Though I would put my money on Larrrabee being Lawlabee at this point.
 
You must log in or register to reply here.
Back
Top