Intel GPU: Larrabee

[crux2005]

Intel Larrabee s 24 jadrami ?

Podľa týchto informácií má byť Larrabee predstavený v druhej polovici roku 2009 na IDF v meste San Francisco. Najprv má byť potrebným hardvérom zásobovaný segment serverov. Až o šesť mesiacov neskôr majú mať aj "obyčajní" zákazníci možnosť zakúpiť si grafické karty "made by Intel". Čo sa týka špecifikácií, tak sa Intel tradične zdržiava konkrétnych údajov. U TGDaily sa ale špekuluje, že Larrabee bude obsahovať až 24 jadier. Spojenie s GDDR 5 pamäťou má zabezpečovať 1024bit široká zbernica. Ako sa majú tieto špekulácie interpretovať nie je jasné, keďže sme 1024bit širokú internú zbernicu zaznamenali už s čipom R600 a 24 jadrový (10 clusterov * 24 SPs) čip bude aj high-end G200 od nVidie s očakávaným predstavením - 18. júna 2008.

Intel - novinky o Nehalem, Larrabee a Dunnington

Poslednou a zrejme najzaujímavejšou novinkou má byť produkt ktorý zabezpečí vstup Intel-u na trh s ne-integrovanými grafickými kartami - Larrabee.
Larrabee bude kombináciou viacerých x86 jadier s celkom novou architektúrou a taktiež VPU (vector processing unit). Gelsinger upresnil, že táto programovateľná architektúra umožní urýchľovať takmer všetko od HD videa a audia, fyziky, umelej inteligencie, ray tracing po global illumination. Larrabee má byť plne kompatibilné s DirectX a OpenGL, existujúce hry bude možné spustiť ihneď, nebude treba čakať na výrobcov hier pre odladenie na túto architektúru.



Prvé funkčné demá tejto architektúry môžme očakávať ku koncu roka, pričom reálne uvedenie produktov na nej založených sa zrejme objavia v priebehu rokov 2009 a 2010.

Larrabee's Rasterisation Focus Confirmed

I've been trying to keep quiet, but I need to get one thing very clear. Larrabee is going to render DirectX and OpenGL games through rasterisation, not through raytracing.

I'm not sure how the message got so muddled. I think in our quest to just keep our heads down and get on with it, we've possibly been a bit too quiet. So some comments about exciting new rendering tech got misinterpreted as our one and only plan. [...] That has been the goal for the Larrabee team from day one, and it continues to be the primary focus of the hardware and software teams. [...]

There's no doubt Larrabee is going to be the world's most awesome raytracer. It's going to be the world's most awesome chip at a lot of heavy computing tasks - that's the joy of total programmability combined with serious number-crunching power. But that is cool stuff for those that want to play with wacky tech. We're not assuming everybody in the world will do this, we're not forcing anyone to do so, and we certainly can't just do it behind their backs and expect things to work - that would be absurd.

Larrabee: Samples in Late 08, Products in 2H09/1H10

When Doug Freedman asked Paul Otellini about Larrabee during yesterday's conference call, we didn't think much would come out of it. But boy were we wrong: Otellini gave an incredibly to-the-point update on the project's timeframe. So rather than try to summarize, we'll just quote what Otellini had to say here.

1. Larrabee first silicon should be late this year in terms of samples and we’ll start playing with it and sampling it to developers and I still think we are on track for a product in late ’09, 2010 timeframe.

Computex 2007: Intel planning to enter discrete graphics card market in 2008

Sources from multiple graphics card vendors have confirmed that Intel has approached them to lay down plans to enter the discrete graphics card market for desktop PCs. The new products from Intel would be positioned as direct competitors to AMD's ATI Radeon and Nvidia's GeForce series of cards.

At this stage all plans are still rather vague, according to the sources, with Intel scheduled to deliver a more complete roadmap and specifications in the fourth quarter of this year. However, according to most of the sources, Intel will aim for the mid-range market in the initial stages with pricing targeted around US$300. In the current market, graphics cards at the high-end are priced at around US$800.
[...]

a ešte aj rýchlejšia ako konkurencia ?

Ďalšie informácie k Larrabee:

PCB ?




Zdroj

Technické údaje ?

1,7-2,5GHz pracovná frekvencia
dátum vydania: 2.Q 2009
počet jadier: 16-24
cache: 32KB L1, 256KB L2
1024bit RingBus
GDDR5 pamäte
> 200 GB/s priepustnosť
> 1TFLOPs/s aritmetický výkon

Roadmap ?
PCwatch

Larrabee to be made in Intel’s FABs
Larrabee to capture 20+ percent discrete in 2008
Larrablee, Intel's discrete graphic is challenging
Besätigung? @ IDF Fall 2007


Zdroj

[fobos]

no snad uz intel konecne schladi rozvasnenu dovolujucu si nVidiu , kedze AMD na to uz nema

[crux2005]

tak s Larrabeem v 1. gen pochybujem

v Inteli si najprv musia uvedomiť, že:
1.) vývoj GPU sa musí podriadiť potrebám vývojárov a nie opačne
2.) hardware sám o sebe nestačí, musí mať aj schopný software

[fobos]

chce to len cas, a naviac firma s takym rozpoctom, co minula neskutocne miliony dolarov za Itanium a nejak vyrazne sa to nej nepodpisalo ...

Myslim ze Intel na to ma dobyt pozicie nVidie

[Fullesh]

konecne nieco nove sa deje. Uvidime ked sa to uz vyjde von medzi spotrebitelov ...Mozno sa z husenice vyvyne krasny motyl v podobe Larrabee

[crux2005]

@fobos: však môžu, ale zatiaľ len skúšali v oblasti GPU

[dexman]

s prihliadnutím na úžasné intelácke igp by som to videl len ako alternatívu, ktorá výkonovo dlho neohrozí nvidiu.

[GABO]

Vzhľadom na to čo produkuje NVIDIA a aké ma blízke plány by som s Dexmanom súhlasil.

[lopo1]

system-on-a-chip Nvidia Tegra

[cyan]

http://www.fudzilla.com/index.php?optio ... 1&Itemid=1" onclick="window.open(this.href);return false;

lol 300W TDP....kde mam chladnicku?

[beardie]

ak to ma byt 16-24 jadier = bude to mat vacsiu plochu = lepsie a ucinnejsie chladenie

zober si to tak, ze teraz je vsetko sustredene v jednom malom bode (570mm^2 v pripade G200), ak by to bola uz 2-2,5x vacsia plocha, nie su take naroky

[crux2005]

ak to ma byt 16-24 jadier = bude to mat vacsiu plochu = lepsie a ucinnejsie chladenie

len aby nepotrebovali ešte viac jadier
--------------------------------------------------

Intel's Top Five Take Five Videos: Roger Chandler Part 1

[crux2005]

Pat Gelsinger, Intel senior vice president and co-general manager of the firm's Digital Enterprise Group, briefly talked about Larrabee and GPGPU languages such as CUDA during a Q&A session at Intel's 40 birthday party. Gelsinger claims programmers don't have the time to learn how to program for these new architectures and says CUDA will be nothing more than "an interesting footnote in the history of computing annals".
The problem that we’ve seen over and over and over again in the computing industry is that there’s a cool new idea, and it promises a 10x or 20x performance improvements, but you’ve just got to go through this little orifice called a new programming model,’ Gelsinger explained to Custom PC. Those orifices, says Gelsinger, have always been ‘insurmountable as long as the general purpose computing models evolve into the future.’

Gelsinger used the Cell architecture used in the PlayStation 3’s CPU as an example to prove his point. ‘It [Cell] promised to be this radical new computing architecture,’ said Gelsinger, ‘and basically years later the application programmers have barely been able to comprehend how to write applications for it.’

This, according to Gelsinger, is one of the major reasons why Intel’s forthcoming Larrabee graphics chip will be entirely based on IA (Intel Architecture) x86 cores. ‘Our approach to this has been to not force programmers to make a radical architectural shift,’ explained Gelsinger, ‘but to take what they already know – this IA-compatible architecture – and extend the programming model to comprehend new visual computing data-parallel-throughput workloads, and that’s the strategy that we’re taking with Larrabee.’

Larrabee, according to Gelsinger, will simply expand on a standard programming model. ‘It’s an IA-compatible core,’ explained Gelsinger, ‘and we’re extending it with a graphics vector visualisation instruction set that has full support for native programming models, full support for the graphics APIs like DX and OpenGL, and then this broad set of new programming models to go with it.’

http://www.custompc.co.uk/news/602868/i ... story.html" onclick="window.open(this.href);return false;

aká propaganda aj keď v "jadre" je niečo pravdy

[crux2005]

Prozessorgeflüster, Von Intels Zukunft und dem guten alten Pentium

Für den Bereich „Visual Computing“ und in einer späteren Version auch fürs High Performance Computing ist Larrabee gedacht, ein Prozessor, der bisher mit 16 bis 24 Kernen gehandelt wurde, wahrscheinlich aber gleich mit 32 Kernen im nächsten Jahr debütieren wird – und zwar wie inzwischen durchdrang zur allgemeinen Überraschung wohl mit genau den gut bekannten Pentium-Kernen: dem Pentium P54C. Es wird sich wohl um die modernere Version mit den größeren Caches namens Pentium MMX handeln, natürlich geshrinkt auf 45 nm. Rattner gab allerdings auf die Frage, ob der Larrabee-Kern denn 32- oder 64-bittig sei, die Antwort, er unterstütze die volle Intel-Architektur – und die umfasst heutzutage ja eigentlich 64 Bit. Ob der Kern wohl „verbreitert“ wurde? Die namensgebende MMX-Einheit selber wird er aber wohl nicht brauchen, denn für die mehrfach parallelen Berechnungen (SIMD) besitzt jeder Kern die neue Vektoreinheit mit mutmaßlich 512 Bit Breite. Das ergibt mit Multiply-Add-Befehlen maximal 32 Flop/Takt in einfacher Genauigkeit (SP), die fürs Visual Computing völlig ausreicht. Bei 32 Kernen würden dann schon 2 GHz Takt reichen, um die 2-Teraflops-Marke zu erreichen, womit er etwa doppelt so schnell wäre, wie die ATI/Nvidia-Konkurrenz jetzt. Vermutlich wird die Larrabee-Karte aber auch doppelt so viel verbrauchen wie jene, also rund 300 Watt. Das lässt sich den bisherigen Veröffentlichungen entnehmen, wo man auf einer Larrabee-Karte zwei zusätzliche Spannungsanschlüsse für 150 und 75 Watt ausmachen konnte, was sich zusammen mit dem Strom via PCIe-Konnektor auf 300 Watt summiert

[slawo206]

http://dsl.sk/article.php?article=6011&title=" onclick="window.open(this.href);return false;

neviem ci vam to nieco nove da, ale stoji to za povsimnutie...