Lerrabee vs 5870 vs 295 vs g300

by the time larrabee exists there will a gpu from either sides will be atleast 4x the fasts version of larrabee.

Links?

This whole thread started with speculation and you're expecting links?  

It was the 'confirmed' 2012 release that interested me.  

if the gpu evolves at vurrent speed doubling its trnasistors and sps every generation, in 2010 we have a single gpu which is 5870x2 speed and 2011 and 5870x4 and 2012 5870x8 and a dual version of that would be 5870x16 and correct me if im wrong. the eight core larrabee will be 4 x slower than a 5870x16 or the so called 8870x2. These are my predictions. if larrabee is predicted, i can predict the future cards ati 2, i guess.

http://www.n4g.com/tech/News-329717.aspx

from wikipedia



it is expecting to be release at the same time of g300(or maybe earlier?) 24core under single gpu die in 45nm fab and quad gpu in same gpu package then 4x gpu card that made of total 384 core. but however it seems likely they move fabrication from 45nm to 32nm which cause delay. which 45nm fab might be cancelled but it will be stupid if they do that. or they have to do it quick before 2010 or release of g300 to gain the market share. 16gpu make of 384core and 1024bit of memory bus width with ddr5 will murder5870 and g200/300 in same time.

and it won't take it to release til 2012.....unlike g300 had been rumor since 2007-8 and still nothing come out.



unless they think they can stuff 128 rops and 320 tmu and ridiculous number of shader unit and squeeze into a single die!? i bet not even 22nm fab can make that happen. nvidia had make that mistake when they done with their g200 and ati wants to do that too?

Holy FAQ Batman !!

That link is months old, and even that said 2010 not 2012, so as he said, where's the supporting info?

You're like 3 announcements behind the time bub! And Wiki !?!  


2012 seems pessimistic and 2010 Optimistic. More likely based on the multitude of statements and delays, late 2010 if everythin goes smoothly, early-mid 2011 if it goes as planned. 2012 is if everything goes off the rails, and by that time intel will be in big BIG trouble as they showed their design intentions over 2 years ago with their basic block diagram during the Terascale presentations (Larrabee is not a Terascale processor BTW). They've so over-exposed their intentions, that further delays would just leave things wide open for AMD & nV (well more so for AMD who could add the X86 component to the equation).




You're such a naive n00b, Larrabee was rumoured long before it's 2007 soft launch, and still it's about a year plus away. There's little doubt G300 will come before larrabee, the question is whether or not LRB can be here before both AMD ad nV refresh, and if it's the 2012 timeframe you originally said, then that's a given, if it's a more realistic late 2010 early 2011, then it could be tight with AMD said to launch the Heck-of-a-Core solution then, and likely nV improving on any shortcoming of the G300 (that they can outside of X86) as they shrink and refresh.

If you think either AMD or nV are just going to stick with what they currently have, you've Heck-a-Naive.  

I think OP meant q1 2010 ....as this is what he states in another thread!

Yeah, but he's either overly optimistic with his January 2010, or overly pessimistic with the 2012 he led with.

AS I mentioned, I would suspect that LRB hits the deck around the time that nV refreshes Fermi/G300, and nearby ATi launches that poorly Greek named Heca-core-majigg, mid-late 2010, if all goes well, not early 2010 other than a tech paper launch for interest & pre-orders.

I'm going for larrabee not seeing the light of day personally   If it does, I'm thinking 2011 earliest now.

And no way will ATI keep that stupid greek nonsense as a codename either.  

Intel could probably launch Larrabee sooner than 2011, hell its silicon is probably in the better shape than Fermi   That said, software side wont catch up NV/ATI for years. IMO what Intel should do - win next gen consoles contracts from both Sony and Microsoft (Intel could give them almost for free cpu + gpu), provide easy to use dev tools, easy way to port to PC as well. In this case all gaming studios will be forced to program for Larrabee strong sides, and NV/ATI would be forced to adjust their next gen cards to Intels tech (like complete programmability, etc), and not vice versa. Given Intels huge influence and cash supply, I can see them becoming very strong player within 5 years. If Nvidia can strong-arm devs as it is, imagine what Intel could do?.. Its on a whole new level.

intel dont need a flagship single gpu to "compete" with amd/nv. they just dont need to play fair game with both of them. using monstrous die shrinking and fabrication processing all they can do is just add more core and processing unit in the die(each gpu has up to 32 processing unit and they can fuse up to 4 gpu in the same die. quad gpu ). and overdose 2x/4x card like gtx 295/5870x2 with relatively low price. that was what amd done with r700 in last years. they tried to be like lerrabee but limited by fabrication made they look like loser. 16rops, 256bit memory were design to be a dual core gpu(yes on die!!)! not a creeple version of "x2"

sure fermi/5900 will beat the crap out of some line of lerrabee but the price will just kill them both. they simply dont have fund like intel do as when intel go with 4x quadcore/6core graphic card, amd will just have to pack up their stuff and go home and nvidia's employee will all go commit suicide.

they dont plan to take market share, but more, they plan to destroy amd and nvidia in the dirty way.

Question - how to ignore trolls? All forums have it, but cant find it on THG  

You have to understand, this isnt a normal troll outburst, its a cpu/Intel troll outburst.
Theyre different over there talking cpus and Intel, I guess thats what happens when 1 company dominates for so long.
Although I have alot of respect for Intel and its products, this market is UNLIKE any that Intels previously encountered.
They dont own anything in it, besides their igp solutions, unlike the SSD market, and obviously the cpu market, where they have all the experience .
No, LRB may be halfways decent, but I'll explain here as I did in another thread, with HKMG coming to gpus, immersion as well, and the process is no great advantage, as the 40nm came out as the highest advanced process used in high commercial availability ahead of the 32nm process, with the 28nm coming not far aways, so no, LRB has no great advantages like Intel has in its other markets, but has huge disadvantages, with a highly competitive market its entering, and the ONLY thing holding the gpu market back is power, and like I said, with immersion and HKMG coming, that impact hasnt been felt yet in gpus