Nvidia Pascal

[Morais-GT]

Boa tarde,

Mas quando raio é que a Nvidia cospe cá para fora isto? Carago, farto de esperar. A GTX 750 2GB tem de ir dar uma volta, tenho de passar para os 4GB de VRAM de preferência HBM 2.0 já, se não que seja GDDR5X.

Cumprimentos.

Então não vais trocar nesta geração

Quase de certeza que as 10XX vão usar GDDR5 ou GDDR5X

[Enzo]

wait for next-next-gen then

[Zavtekirou]

Boa tarde,

Eu disse lá, se não for HBM 2.0 que seja GDDR5X. Já é superior ao que tenho (GDDR5) por isso não perco nada. Mas se me derem a escolher, claro que vou para HBM 2.0.

Cumprimentos.

[jotinha17]

Superior ao que tens ? Neste momento praticamente qualquer gráfica é superior ao que tu já tens.

[Zavtekirou]

Boa tarde,

Estava a falar exclusivamente do GDDR5, GDDR5X e HBM 2.0. Em termos da gráfica em si eu sei que há muita coisa superior (quando a comprei ela não era nenhuma gráfica top).

Mas acho que não vale a pena estar a comprar da geração atual com a nova a chegar aí daqui a pouco tempo. Pelo menos penso eu que vá demorar pouco, já passou tanto tempo desde a última geração.

Cumprimentos.

[jotinha17]

Já tens HBM no mercado e não saiu à muito tempo, se um gajo vai estar sempre à espera que saia algo novo para comprar, nunca vai comprar nada porque daqui a X meses sai algo novo.

Digo isto, porque já te vejo dizer que vais comprar as novidades desde que estás aqui na tasca

[Zavtekirou]

Boa tarde,

Não será no imediato porque ainda nem saíram (e dinheiro também anda curto ainda por cima com o carro a dar despesas de pneus e jantes e eu sem emprego já me vejo e desejo para manter o carro quanto mais para gastar no PC) mas espero conseguir endireitar as coisas até as gráficas Pascal saírem (até me dá jeito ainda nem terem saído).

Mas será a próxima coisa a mudar no PC. Já faltou mais. Será o passo que falta para tirar este PC da categoria de low-end para mid-end. Para já, safa-se bem como está, mas a gráfica nova vai mudar bastante a performance (relativamente a gaming). Não é que não consiga correr algo que queira, simplesmente vou passar de settings medianos para settings mais elevados. O i3 4330 tem sido uma surpresa bastante positiva, bastante mesmo. A GTX 750 nunca desiludiu, simplesmente chegou ao limite das suas capacidades para as exigências que se adivinham (jogos nativos em 64 bits consomem mais VRAM e 2GB fica curto rapidamente, se quero começar a correr jogos em high e very high ela também não vai lá, fica em medium, etc).

Cumprimentos.

[Jorge-Vieira]

Nvidia GeForce X80 and X80 Ti Pascal Specs?

In our forums somebody posted the spec list shown after the break. Indicating the Pascal graphics cards would be called GeForce X80 and X80Ti, first a warning as the validity seems very doubtful.
Now, the sheet is nothing in the style from any Nvidia sheet I have ever seen, the specs are based on GDDR5, the cards called GeForce X80 and X80 Ti etc etc so the likely-hood that the table has been fabricated is far bigger being false than true.
Um yeah I had to think a couple of times over to even post this as heck it was spotted on a vague imguru link with no other info. Lets be honest here, the chart takes 5 minutes to make in Excel. Still for brevity's sake, I'll post them as they are bound to show up on other websites somehow, discuss here. And yes again, take it with a HUGE grain of salt (and then some more) please.
Okay, that's enough disclaimers right ? here we go:

Noticia:
http://www.guru3d.com/news-story/nvi...cal-specs.html


É apenas uma imagem, não está confirmado que é real.
Analizando a imagem, vemos para já um aumento considerável de nucleos CUDA face ás gerações anteriores, vemos também o regresso da nvidia ao BUS de 512 bits (X80TI), e vemos a inclusão de HBM.
A ser verdade, acho que vamos ter finalmente um salto considerável em performance como já não se via há algumas gerações.

[Jorge-Vieira]

Nvidia "leak" suggests rather dull Pascal GPU naming scheme

We disagree, offer some alternatives

On Wednesday, someone thought it would be clever to post an unsourced Nvidia Pascal GPU naming scheme spreadsheet to popular image hosting website Imgur.com. As a result, GPU enthusiasts from all around have frequented hardware forums to discuss the matter of Nvidia’s yet to be announced 2016 GPU lineup based on 14-nanometer FinFET technology.

The image leak, complete with the words “NVIDIA Confidential” at the bottom, appears as if anyone could have typed it up in Excel and posted it to the popular photo sharing site. Nevertheless, it suggests that Nvidia is preparing to depart from the “Geforce GTX” naming scheme that it adopted for Tesla, Fermi, Kepler and Maxwell-based GPUs, and will instead use a much simpler “Geforce X80” nomenclature followed by the usual “Ti” prefix for any higher-clocked card variants.


Source: Imgur.com
While we do not agree with the post and suggest it be taken with two or three grains of salt, the author suggests that the midrange GP104 Pascal GPU with 4096 shaders will be named “Geforce X80,” while the enthusiast and flagship variants based on GP100 will be named “Geforce X80 Ti” and “Geforce X80 Titan.”
If this were true in any respect, it would certainly indicate a backwards naming step for such a powerful architecture, especially coming from “GTX 980,” “GTX 780,” “GTX 680,” etc. The other issue is that many people may be inclined to believe that a GTX 980 is more powerful simply because it uses more letters in the prefix and digits in the numbering format, and simply because it sounds like a product with a heaver amount of horsepower.
History of notable Nvidia GPU naming schemes
In October 2001, Nvidia adopted the “Geforce 3” naming scheme for its NV20 architecture, giving it names including “Geforce 3,” “Geforce 3 Ti 200” and “Geforce 3 Ti 500.” Fast forward to February 2002, and the Geforce 4 series included names such as “Geforce Ti 4400” based on NV25 and “Geforce 4 Ti 4800” based on NV28 Ultra. Over a year later between March and December 2003, the company switched to “Geforce FX” names for its NV30 and NV35-based lineups with models including “Geforce FX 5800,” “Geforce FX 5900XT” and “Geforce FX 5950 Ultra,” to name a few.
The Geforce FX line then moved to PCI-Express in early 2004 and adopted naming including “Geforce PCX 5300,” “Geforce PCX 5750,” “Geforce PCX 5900” and “Geforce PCX 5950.” Over the next year between 2004 and 2005, these models were eventually replaced with the Geforce 6 Series, marking Nvidia’s transition into the “modern era” of GPUs with support for features including SLI technology, Shader Model 3.0, Direct3D 9.0c and OpenGL 2.1. This number-first naming scheme thrived for the second half of the decade and was followed up with the Geforce 7 Series in June 2005, Geforce 8 in November 2006 and Geforce 9 in February 2008.
During the timeframe between Geforce 6 and Geforce 8, Nvidia used a variety of suffixes for its number-first naming scheme, including “GE,” “GS,” “GSO,” “GT,” “GTX,” “GTX+” and “GX2” for dual-GPUs. Finally, the company decided it was running out of number-first options after the Geforce 9800 GX2 launched, and it proceeded to switch back to a prefix-first, rather than number-first, naming scheme as it had used with the Geforce 4 and 5 series.
At this stage beginning in June 2008 with Tesla-based GT200 architecture, the company went back to prefixes first with “Geforce GT 230,” “Geforce GTS 240,” “Geforce GTX 260,” and “Geforce GTX 260 Core 216,” etc. This naming scheme was followed all throughout the first half of the next decade from 2009 to 2015, which saw the releases of the Fermi-based Geforce 400 and 500 Series in 2010, the Kepler-based Geforce 600 Series in 2012 and Geforce 700 Series in 2013, and the Maxwell-based Geforce 800M Series and Geforce 9 Series in 2014.
What’s next for naming GeForce GPUs based on Pascal, Volta and beyond?
In February 2013, Nvidia did something fundamentally unique with its Geforce lineup by introducing a flagship GPU series without numbers, the Geforce GTX Titan series. This initial card was followed up with the Geforce GTX Titan Black in February 2014 and the Geforce GTX Titan Z dual-GPU in March 2014, all of which featured no numbers but were nevertheless still indicative of top-performance choices. We have reason to believe that Nvidia is definitely prepared to do this again for Pascal and enjoys its choice to market top-level enthusiast products without number designators. At this point in the game, however, there is simply no more room for the “GTX” prefix.
Suggestion: "GeForce PX" for Pascal, "GeForce VX" for Volta
We have a suggestion that may be more fitting for Pascal, Volta and successive generation Geforce products. Perhaps it’s time for Nvidia to follow what it did with the recently-announced Drive PX 2 automotive supercomputer, switch out the aging "GT" and "GTX" prefixes and replace them with something more indicative of the architecture on which a product is based. With its automotive lineup, the company has already indicated it will use “Drive” to denote the automotive segment, “PX” to denote a Pascal-based product, and a number at the end to signify the number of GPUs.
For its consumer market segment, we propose staying in line with what it introduced in the automotive segment – maintaining its highly-recognized “Geforce” brand at the front, followed by “PX” to denote a Pascal-based product, followed by a number (either arbitrary or based on shaders) to denote level of performance.
If the ending SKU number indicated the number of GPU shaders, the product lineup would look as follows:
GeForce PX 4096
GeForce PX 5120 Ti
GeForce PX Titan
GeForce PX Titan M
(Note: Hypothetical names)

Eventually, the shader count would get too high, albeit in the 10,000s by next decade or with a dual-GPU variant, making a shader-based naming scheme less plausible over the long term.
Alternatively, if the ending SKU performance number were based on historical values for performance segments, the product lineup would look like this:
GeForce PX 4000
GeForce PX 5500 Ti
GeForce PX 6500
GeForce PX 7500
GeForce PX 8500 Ti
GeForce PX Titan
GeForce PX Titan M


(Note: Hypothetical names)
The company could even do the same with Volta in 2018: GeForce VX, GeForce VX Ti, and GeForce VX Titan, etc.
The only wildcard this time around is that Nvidia will use both GDDR5X memory modules for its mainstream and midrange segment and High Bandwidth Memory (HBM) 2 for its high-end segment. A third alternative would be to differentiate number choices based on the use of GDDR5X or HBM2. Given the fact that AMD has chosen to do this with its Radeon “Rx” 300 Series (all using GDDR5) and Radeon R9 Fury, Radeon R9 Nano, Radeon R9 Fury X and Radeon Pro Duo (all using HBM), this is another possibility not out of question for Nvidia’s Pascal, Volta, and whatever may come next.

Noticia:
http://www.fudzilla.com/news/graphic...-naming-scheme


Mais uma noticia sobre o leak de ontem, que ainda está por confirmar e que mostra uma suposta nVidia 1080 TI mais ou menos umas 2,5X mais rápida que a actual GTX 980 TI, esta noticia também faz algumas previsões sobre os nomes que as proximas placas da nVidia podem vir a ter.

[Jorge-Vieira]

Nvidia Pascal 3DMark 11 Performance Entries Spotted – Potential GTX 1080, GTX 1070 & GTX 1060 Benchmarks

Multiple potential Nvidia Pascal 3DMark 11 GPU entries have been spotted. The entries are varied and show GPUs with performance figures that range from very high – GTX 980 Ti territory – to entries that show performance closer to the GTX 970.

Before we get into the performance figures I’d like to give a shout out to videocardz who have spotted two of the four entries we’re sharing with you today. Also as a reminder, please keep your salt shaker handy at all times during this. The entries you’re about to see certainly don’t match any current or previously launched Nvidia graphics cards but there’s no surefire way of confirming that these are indeed Pascal chips.
Multiple Potential Nvidia Pascal 3DMark 11 Entries Spotted

The first entry we have is of an unidentified Nvidia GeForce graphics card with 7,680 MB of graphics memory, 512MB short of 8GB, very likely a test chip. More interestingly is that this GPU’s VRAM is clocked at 2000Mhz – 8000Mhz effective. It’s a good time to point out that Nvidia has no GPUs on the market configured with 8GB of 8000Mhz GDDR5 memory. And certainly no 8GB GPUs with performance that’s nearly identical to the GTX 980 Ti. The only 8GB 900 series video cards Nvidia currently has in the market are its GM204 based mobility chips, which can come configured with 4GB or 8GB of 7000Mhz GDDR5 memory.


However as mentioned earlier, neither does the GTX 980M, or GTX 980 notebook match the performance figures you see above. Which are inline with the GTX 980 Ti. So we might very well be looking at the GTX 1070 ,GTX 1080 or the mobility version of either. Also worthy of note is the 545mhz clock speed recorded above. After we’ve done a bit of digging around we found that in most cases when 3DMark does not fully identify the graphics card in question it will report a ~540mhz clock speed, despite whatever the actual clock speed of the card is.
So in this case and indeed in all the following entries you will see 540mhz erroneously reported as the GPU frequency because 3DMark failed to read the actual clock speed of the card, possibly as the result of a BIOS misread. This can be confirmed by looking at some other misidentified GTX 970 and GTX 980 entries with a recorded clock speed of 540mhz. Which also happen to show performance figures that are on par with what’s normally expected of these cards at their boost frequencies. Really no where near what they would score at ~540mhz.
Now let’s move on the next two entries. Again we have two cards with 8GB of 8000Mhz GDDR5 memory, showing performance that’s very similar this time around to factory overclocked GTX 970 graphics cards. So we could potentially be looking at early mobile variants of Pascal.
Advertisements




The last interesting entry is of an unidentified Nvidia 3GB 2500mhz – 10,000Mhz effective, graphics card. So we could potentially be looking at the very first GDDR5X pascal graphics card here. 3DMark will on occasion also report erroneous memory frequency figures doubling the actual speed. However there are no 3GB 1250Mhz Nvidia graphics cards. The GTX 780 features 1500Mhz – 6000Mhz effective – GDDR5 memory and the GTX 780 Ti features 7000Mhz effective GDDR5 memory.


Interestingly this card is also showing performance very similar to factory overclocked GTX 970 graphics cards. Keeping in mind that GDDR5X memory modules come in 512MB, 768MB, 1GB, 1.5GB and 2GB configurations this card could plausibly have a 128bit or 192bit bus for a total of 160GB/s or 240GB/s of bandwidth. So we might be looking at a ~ GTX 1060 class graphics card here. The fact that GDDR5X memory chips come in a variety of capacities also means that Nvidia could conceivably launch 4GB or even 6GB versions of this card.

Yesterday we discussed the rumored GTX 1080, GTX 1080 Ti and nextgen Titan specs that have made the rounds on the web. Today we got to look at potential performance figures more for Nvidia’s mid-range offerings.
Nvidia’s GTX 1080 and GTX 1070 graphics cards are set to launch this summer and we can’t wait to see them in action.

Nolticia:
http://wccftech.com/nvidia-pascal-3dmark-11-entries-spotted/#ixzz43I8c1LXg


Uma primeira amostra do Pascal, deveremos estar aqui a ver possivelmente as sucessoras da GTX 970 e 980TI.
Curisiodade também para uma placa que aparenta ter o novo tipo de memória GDDR 5X.
Estou com muita expectativa para ver isto (e também o Polaris da AMD) em ação.

[Zavtekirou]

Boa noite,

Venha ela venha ela...
Cumprimentos.

[Enzo]

Isto este ano vai ser mesmo hoooooott!!!

[Jorge-Vieira]

Rumor: Nvidia’s Pascal Architecture Is In Trouble With Asynchronous Compute

According to a report published by Bitsandchips.it, Nvidia’s upcoming Pascal architecture will not be significantly better at Async Compute than its predecessor (Maxwell). Needless to say, this one is 100% a rumor and should be taken with a grain of salt. Pascal architecture will be landing sometime later this year and improves upon Maxwell with far better FP64 support amongst a plethora of other things. The report further mentions that Nvidia is hoping to win this round by raw performance numbers alone.

Pascal architecture allegedly facing difficulty with asynchronous compute

Asynchronous compute has been a deal sweetener for Radeon buyers ever since the DirectX 12 API hit the stage. AMD is currently leading in all DirectX 12 benchmarks available on the market right now and this is solely due to the fact that its GPUs can process ASync concurrently. Interestingly, Nvidia GPUs perform much better without ASync turned on. This is probably due to the fact that Nvidia has disabled ASync from their driver suite due to the fact that its GPUs cannot process ASync concurrently on the hardware level, rather they need context switching which is expensive in terms of frame rate. Geforce GPUs instead rely on a technique called pre-emption. Which is why you see frame rates actually being unpredictable when ASync is forced.

On the other hand, Maxwell remains the only architecture on the discrete GPU market right now which supports DirectX 12 Feature Level 12_1 (Radeon cards only extend up to Feature Level 12_0). This will allow existing Geforce cards to use advanced rendering techniques made available by Direct3D 12 that will not be available to AMD users. VXGI/VXAO and Hybrid Ray Traced Shadows are examples of such a feature. So on both sides of the fence, whether you are in the red camp or in the green camp, there is an upside and a down side. Pascal and Polaris however, were supposed to bridge this gap and unify full compatibility of DirectX 12. Before we let that train of thought run away however, there is one thing we must keep in mind, the entire point of features like ASync is to maximize the use of a GPU’s resources to allow the maximum possible performance.
Advertisements


One reason that we think that this rumor might be true is the fact that chip design isn’t something that happens overnight. In fact, it takes an architecture many years to go from the drawing board to hit the shelves. Asynchronous Compute was hyped and became a major point of interest in the last year, which is most definitely not enough time for Nvidia to do anything about it. If Asynchronous Compute wasn’t something that was a focus when the Pascal chips were initially designed – then there is nothing Nvidia can do about it at this late in the game. The report further mentions that Nvidia also recently pushed its entire GameWorks SDK into the public domain via GitHub which could be seen as a move to make sure that all games that utilize its technology are fully optimized to leverage Nvidai GPU capabilities (No ASync + Bad Game Optimization = Bad Combo).

What we do know for a fact is that Nvidia has been focusing on its FP64 or double precision performance – which was completely culled in Maxwell to offer superior value to gamers (which are mostly single precision). So whether or not lack of ASync support in Pascal translates into a real world loss depends on how the raw performance of Pascal’s Graphic Cards compares to ASync enabled Pascal graphic cards. Even if this report is true Nvidia could still win the next round by focusing on raw gaming performance of its Pascal GPUs (Techniques like ASync can only help utilize the maximum potential of all its resource, not actually add more power out of nowhere.)

Noticia:
http://wccftech.com/nvidia-pascal-trouble-asynchronous-compute/#ixzz43vZhEWwu


A ser verdade, são más noticias para a nVidia, esperava-se que a nvidia implementasse os Async Shaders no Pascal para ficar igual (ou superior) à oferta da AMD na sua arquitetura GCN.
Espera-se umas guerras interessantes entre ambos se os Asyn Shaders começarem a ter relevância nos jogos à medida que forem implementados.

[Zavtekirou]

Boa tarde,

Microsoft a criar problemas para a Nvidia com o DX12 e a sincronização de gráficas de ambos os fabricantes xD Devem estar a chamar uns quantos nomes ao W10 e ao DX12 lá na Nvidia LOL.

Cumprimentos.

[Enzo]

Fermi is back?
Não foi nada que eu e o Jorge já não tivessemos falado aqui.