NVIDIA novinky

[Haste]

otazkou zostava, ako moc bude R580 uvedena papierovo .... a ako moc realne :\

Podle ATi bude dostupna hned v den vydani(a nizsi takty jadra i pameti tomu nasvedcuji, ze dostupnost bude podstatne lepsi nez u X1k), takze bych u nas cekal neco jako X1300, kterous mohl koupit cca 2-5 dni po oficialnim vypusteni ...

[no-X]

Tak růžově bych to neviděl, ale myslím, že za týden až deset dnů (od vydání) bude snadno dostupná.

[Masster]

Next generation NVIDIA GPU product details revealed

It would seem that a few key NVIDIA partners have already been sampled with GeForce 7900, and they're currently working towards a hard launch planned for CeBIT 2006 in Hannover, Germany (9th through to 15th March), and if so, our money's would be on watching for announcements from NVIDIA, ASUSTeK, eVGA, XFX and possibly MSI, between the 9th and 11th March.
Unlike the launch of its GeForce 7800 GTX, the current plan seems that NVIDIA will launch both the faster GTX and GT variants at the same time.
The flagship single-card product GeForce 7900 GTX is claimed to be a 32 pixel processor part with a core GPU frequency of 700/750 MHz, and should ship with 1.1ns graphics RAM clocked at 800/900MHz.

the money-shot is that a 700MHz NVIDIA GeForce 7900 is currently coming in at circa 13,000 3DM2K5's on an AMD Athlon 64 FX 60 system.

http://www.hexus.net/content/item.php?item=4343

[Haste]

No bylo by to pekne, kdyby se jim to podarilo, ale:
1.. 750MHz dohromady s 32pixel pipelines zas tak moc neverim
2.. 1.1ns pameti budou asi slusne drahe

Kazdopadne, nejaka 650-700MHz karta s 32 pipelines, tomu bych uz veril a rekl bych, ze nebudu tak daleko od pravdy. Jestli k ni pridaji 1.1ns pameti, tak si zbytecne dost snizi vyteznost + zvysi cenu, ale zas na druhou stranu NV z duvodu pouheho pridavani pipelines (a tim neunosneho zatezovani pametoveho subsytemu) nema na vyber nez po 1.1ns sahnout.

[THX]

Tu su nejake dalsie vysledky 3dmarkov, je tam x1800, x1900, 7900, samozrejma aj crossfire a aj SLI ->

http://www.hexus.net/content/item.php?item=4346

[Haste]

Tak to je snuzka dohadu a co je horsi, tak i nesmyslu.

Uz jenom:
ATi RADEON X1900 XT 512MB (625MHz VPU / 1450MHz GDDR-3) (AMD A64 FX-57 2.8GHz) ~10,000 3DMarks for $549 (£311.04)

Kdyz dle vr-zone:
"More interestingly, we heard faintly that a early sample of X1900 XTX clocked at 500MHz scored 10.2K in 3DMark05 ", tedy teoreticky by XTcko (625MHz) melo dat az kolem 12k...

Nehlede na fakt, ze 7900 (ani samply) jeste nema nikdo v ruce...

[Masster]

NVIDIA G80

We heard that G80 will be in time for launch in June during Computex and the process technology is likely to be 80nm at TSMC. In the recent statement, NVIDIA has said that they will be backing the 80nm "half-node" process by TSMC where it allows reduction of die size by 19%. We have previously mentioned that G80 is likely to take on the Unified Shader approach and supports Shader Model 4.0. G80 is likely to be paired up with the Samsung GDDR4 memories reaching a speed of 2.5Gbps.

According to Xpentor, NVIDIA G80 will make ATI stumble on April. Quad SLI itself can be implemented on a single card with two chips solution because it will carry the first dual core GPU ever with the support of DirectX10 and Shader Model 4.0. The development of G80 is also mentioned as being running very intensive since NVIDIA's acquisition over ULi. As for the upcoming G71, there will be 32pipes, increase in ROPs and a little speed bump over the core clock.

http://www.vr-zone.com/?i=3177

[Comby]

Som zvedavy na velkost dvojjadra v mm², ked teraz ma G70 (jedno jadro) velkost 334 mm². Uvidime ako sa na velkosti jadra prejavi novy vyrobny proces a kolko bude obsahovat tranzistorov.

[.:::iGor:::.]

...no v podstate od pondelka sa po inete zacali sirit rozne "unconfirmed rumours" k G80/R600, najskor sa to objavilo na xpentor.com a ostatne weby sa toho hned chytili, citovali, pridavali si dalsie svoje detaily atd, proste tak jak to vzdy chodi

...na druhu stranu sa hovori: "na kazdom sprochu, pravdy trochu"

[no-X]

Ovšem některý informace jsou taky postavený na hlavu Je to pokaždý stejný - má přijít nová grafika, někdo si na nějakým fóru vymyslí, jaký by mohla mít parametry, další člověk to někde ocituje bez kontextu, přečte si to redaktor "bulvárního" serveru (Fuad), kterej to hodí no novinek jako specifikaci novinky a kachna je na světě

[.:::iGor:::.]

Ovšem některý informace jsou taky postavený na hlavu Je to pokaždý stejný - má přijít nová grafika, někdo si na nějakým fóru vymyslí, jaký by mohla mít parametry, další člověk to někde ocituje bez kontextu, přečte si to redaktor "bulvárního" serveru (Fuad), kterej to hodí no novinek jako specifikaci novinky a kachna je na světě

...jj presne tak, a Fuad je presne na toto "specialista"

[Masster]

Jenze tohle prave neni Fuad Zvesti o dual-core GPU od nV uz tu jsou od nV 40/45, nekteri predvidali dual-core GPU u G70. Je to jen predzvest quad SLI. V tomto pripade je dual-core jedina moznost, protoze by si tezko nekdo koupil 4 stejne graficke karty v hodnote 400 dolaru a vice.

[Eagle]

Jenze tohle prave neni Fuad Zvesti o dual-core GPU od nV uz tu jsou od nV 40/45, nekteri predvidali dual-core GPU u G70. Je to jen predzvest quad SLI. V tomto pripade je dual-core jedina moznost, protoze by si tezko nekdo koupil 4 stejne graficke karty v hodnote 400 dolaru a vice.

Sice grafikám tak moc nerozumím, ale jaký je princip "dual-core" u grafik? Není to tak trošku blbost?

[Masster]

No v podstate ano, ale z marketingoveho hlediska to zni dobre a jeste to tu nebylo

[richie08]

Blbost to neni z hledeska vyteznosti chipu coz vzhledem k jejich narustacim rozmerum je dost podstatne. Navic se daji parovat, ktere se povedly stejne a nenastane situace, kdy se musi cele jadro vyhodit protoze se neco v pametovem radici nepovedlo jak melo.

Ovsem ma to i nevyhody, ktere dneska prevazuji. To neznamena, ze na tom nemuzou zapracovat tak, ze se to v blizke budoucnosti vyplati.

[Eagle]

Blbost to neni z hledeska vyteznosti chipu coz vzhledem k jejich narustacim rozmerum je dost podstatne. Navic se daji parovat, ktere se povedly stejne a nenastane situace, kdy se musi cele jadro vyhodit protoze se neco v pametovem radici nepovedlo jak melo.

Ovsem ma to i nevyhody, ktere dneska prevazuji. To neznamena, ze na tom nemuzou zapracovat tak, ze se to v blizke budoucnosti vyplati.

Výtěžnosti? Heh? A jak bys to chtěl provést? Víc čipů na jedné destičce? A jak je spojíš, aby to bylo dostatečně rychlé? Jako současné SLI to být nemůže, protože tam má každý čip vlastní řadič pamětí. A v případě, kdy by to bylo víc čipů s nějakou propojkou a lá HyperTransport a centrální řadič pamětí, půjde výkon do úplně do háje. Takže IMO to je sakra velká blbost.

[no-X]

Eagle: Samo o sobě to blbost je. Tím myslím, že je to nesmysl v případě, že by oba čipy pracovaly v AFR, SFR atp. Smysl by dávalo, kdyby měly nějaké efektivnější propojení - např. že by se celek choval jako jeden čip s více quady, případně kdyby oba "čipy" byly asymetrické. Ale párovat dva čipy přes AFR a ty ještě kombinovat s další dvojicí přes AFR by byl marketingový ojeb, protože každé zvýšení FPS by bylo na úkor zvýšení grafického lagu a snížení reakční doby pro hráče (jednočipové řešení při 60FPS ~ 17ms; čtyřčipové řešení při 60FPS ~ 67ms). Plně doufám v první (lepší) možnost

[THX]

Tie karty s dvoma chipmi na doske su dobra finta, umoznuju quadSLI (teda, nie vsetky, treba na to aj dosku ale hlavne driver), teda ak by zrazu ATi mala nejaku velmi dobru kartu, tak nvidia stale moze urobit quadSLI a povedat ze Ati je technologicky zaostale atd. atd. Ja to vidim na cisty marketing, pripadne na profesionalne pouzitie. Tie dualcore karty by mali potom quadSLI ? Bude vobec quadSLI (ofic.) pred vydanim tychto "dualcore" nexgen chipov? Ja tipujem ze nie.

Kazdopadne som zvedavy ze co to bude, az to bude

[no-X]

Sice nevím, jestli se to týká přímo G80 (nebo až nějakého nástupce), ale je to hodně zajímavé

I thought I'd post a thread here (after being told off!) about a patent I posted in the console tech a while ago,

NVIDIA Patent: G70/G80/RSX?



Programmable graphics processor for generalized texturing

It describes a unified shader unit with a coupled texture unit. The texture unit has sub-units that are multi-threaded. These specialised sub-units could be used in combination for a desired result and therefore increase untilisation. Effectively the unified shader unit is a multi-threaded processor. They also discuss removing ROPs completely so that the unified shader unit could act as a 'programmable' ROP...



Here's a pic of the unit... thoughts?

I think it's worth posting one of the '240s' as well, just for some context. Whether G80 ends up related to this patent or not, I think the fact that this patent itself describes a unified shader architecture is fairly evident.



With some select quotes from the patent:

In a typical implementation Programmable Graphics Processing Pipeline 150 performs geometry computations, rasterization, and fragment computations. Therefore Programmable Graphics Processing Pipeline 150 is programmed to operate on surface, primitive, vertex, fragment, pixel, sample or any other data. For simplicity, the remainder of this description will use the term "samples" to refer to graphics data such as surfaces, primitives, vertices, pixels, fragments, or the like.

FIG. 3 is an illustration of an exemplary embodiment of Execution Pipeline 240 containing at least one Multithreaded Processing Unit 300 in accordance with one or more aspects of the present invention. An Execution Pipeline 240 can contain a plurality of Multithreaded Processing Units 300, each Multithreaded Processing Unit 300 containing an Execution Unit 370. Each Execution unit 370 includes at least one PCU 375. PCUs 375 are configured using program instructions read by a Thread Control Unit 320 via a dedicated Read Interface 205. In an alternate embodiment Read Interface 205 is shared between two or more Multithreaded Processing Units 300. Thread Control Unit 320 gathers source data specified by the program instructions and dispatches the source data and program instructions to at least one PCU 375. PCUs 375 perform computations specified by the program instructions and outputs data to at least one destination, e.g., Pixel Output Buffer 270, Vertex Output Buffer 260 or Register File 350.

I read this patent when it first appeared in console and just re-read it again. One thing that stands out to me is that the threading and arbitration logic is implemented at the pipeline level and there isn't a master scheduler like in Xenos.

The other thing is the extra level of nesting of processing elements.

Pipeline (Shader) -> Multi-threaded processing unit(s) MPU(s) -> Execution Unit -> PCU(s).

A PCU looks to be just what we call an ALU today. What I don't get is why you would need multiple MPU's in a pipeline ? ILP is handled through more PCU's. TLP can be handled by a single MPU. What benefit is there to having multiple MPU's per shader as opposed to more shaders with a single MPU each?

[no-X]

And one more, a very important thing. G71 at 700+ MHz can and will beat R580, Radeon X1900 even the XTX variation. The other issue is, when will Nvidia manage to ship those cards in reasonable quantities? A middle March launch press launch and early April for the real availability sounds like the best shot to us. Remember Nvidia taped out in first weeks of January not even a month from now and it takes roughly three months from tape out till the production. We can still be wrong but will keep our eyes wide open.

http://www.theinquirer.net/?article=29516

[THX]

Na xbitlabs sa objavili nove informacie o G80 -->
http://www.xbitlabs.com/news/video/d...220100915.html

Nvidia’s First DirectX 10 Chip to Be “Hybrid” Design.
.......
Nvidia’s code-named G80 graphics processing unit (GPU) will incorporate 48 pixel shader processors and an unknown number of vertex shader processors, some unofficial sources said. The chip is still expected to support feature-set of DirectX 10 along with Shader Model 4.0, even though it will not take advantage of the unified processors that can compute both pixel and vertex shaders.
.......

[no-X]

48 pixel shader processors? To má přeci už G70...? Takhle je to docela nicneříkající číslo; pokud tam není alespoň náznak jejich architektury, tak je to prakticky nulová informace

Pro rejpaly preventivně: ano, už NV40 měla dva pixel shader procesory na pipeline (tzn. 32 celkem), prezentovala to tak i nVidia:



takže G70 má 48 pixel shader jednotek

[no-X]

Tohle je zajímavý:

When I met with Kirk in March 2004 during the build up to dropping the NV40 bomb I asked him about design lead times and how difficult it must be to choose the right feature set in such a fast moving market place when laying down the parameters of a chip that wouldn't be on sale for a couple of years minimum. And he intimated that at that time NV50 was essentially done and dusted. Not taped out or anything like that, but certainly complete in terms of high level design. In other words, you would infer from his comments that any decision regarding unified architectures and supporting AA with HDR would have already been taken back then. Of course, plans and products can and do change, but as we head towards DX10 it's kinda interesting to reflect on, I would think.

[Masster]

GeForce 7600 GT and 7600 GS to promote H.264

WE are happy to inform you that Nvidia plans to get aggressive on H.264 acceleration and HDTV support. A month ago we wrote about Nvidia's driver that is supposed to make H.264 much faster and we are still waiting for it. The driver should re-surface around Cebit time. If you remember, Nvidia usually have only four drivers each year, one per quarter.

http://www.theinquirer.net/?article=29852

[richie08]

Výtěžnosti? Heh? A jak bys to chtěl provést? Víc čipů na jedné destičce? A jak je spojíš, aby to bylo dostatečně rychlé? Jako současné SLI to být nemůže, protože tam má každý čip vlastní řadič pamětí. A v případě, kdy by to bylo víc čipů s nějakou propojkou a lá HyperTransport a centrální řadič pamětí, půjde výkon do úplně do háje. Takže IMO to je sakra velká blbost.

Kdyz das dva cipy na jednu desticku, tak vzhledem ke vzdalenosti muzou byt propojeny mnohem rychlejsi sbernici nez je SLI nebo HT dohromady. Kazdopadne pokud by to bylo jen takove modifikovane SLI, tak by jiste narust vykonu nebyl zanedbatelny. Vzhledem k tomu, ze nyni se zacinaji objevovat quad SLI tak bych se ani nedivil, kdyby nVidia nebo Ati na podobnych resenich jiz davno pracovaly. A u ctyrcipoveho reseni jiste nezustane.

Navic se da uvazovat o ruznych jinych moznostech. Treba kazdy cip muze mit vlastni pametovy radic s pridelenou casti pameti a pamet sdilet po rychle sbernici. Nebo klidne i ten centralni radic, ktery by diky tomu mohl obsahovat velkou vyrovnavaci pamet. Moznosti je spousta a jiste na spouste z nich se na vyvoji pracuje. Jenom se to nikdy takovi pisalci na foru jako jsme my nedovime. Vyzkum muze skoncit v kosi, supliku nebo se to dovime az tehdy je-li produkt na trhu nebo chce-li vyrobce navnadit zakazniky a pousti info predem kvuli marketingu.