VoodooAlert - let´s get loud

heh here's something more about that the Fermi showed is a fake heh
http://www.semiaccurate.com/2009/10/01/n…rmi-boards-gtc/

Anyways it's a rumor to me, the pics are not highres enough to prove this It seems more like an anti-Nvidia post than one with facts.

Anyways such news should be left aside and taken as a pile of salt

Well it does seem to have been a mockup after all:
http://www.fudzilla.com/content/view/15798/34/

But the physx demo's were ran by aworking Fermi Engineering Sample only it was packed by wires aka hand done reworks and it looked like an octopus...hmmm reminds me of the Voodoo5 6000 AGP 128MB Rev.A1 1500 from osckhar
Yups an Octopus allrighty

Even so it may look like an Octopus as them previews go about it to me it's more like a thing of beauty and rarity that V6K or even that GT300 heh

Die gezeigte Tesla war gar nicht lauffähig




bilder: pcgh

Boah wenn ich das hier schon wieder lese...und dann wieder dieses unsägliche Fanboy getue von einer gewissen Person

Wartet doch einfach ab! Wen juckts ob die Karten denn nun gefaket sind oder nicht? AMD war schneller und wird wohl bis mindestens Ende des Jahres davon profitieren.

Ganz genau - wurscht egal ob das ne Fake Karte ist, hauptsache Fermi wird wie geplant am Start sein. Ist ja auch nicht so, dass so ne Vorstellungsshow von heute auf morgen geplant und organisiert ist, das braucht ja auch ne gewisse Vorlaufzeit. Und wenn dann die ersten Testsamples noch irgendwo in den Laboren zum debuggen wesentlich dringender gebraucht werden, ists doch halb so wild, dass el Cheffe bei der Show ein aufgemotztes stück Schrott präsentiert

what the point even 3dfx had a fake card, which makes me wonder many seem to forget things


So much for it's missing traces eh?



None the less that Comdex V6K is probably the most beautiful fake card ever made

Even ATi made fake cards over time, normally fake cards are used to keep the people busy at the dog and pony shows aka confrences so it's nothing special to see a fake card people, as the fanboyism goes its all over the place but then again is being a fanboy of Nvidia 3dfx, Sis, XGI, matrox ATi and other brands illegal? I don't think so, I'm like accept all things then the hassle about it will be milder.

The pattern will remain the same ATi comes with something nice and fast and NVIDIA always managed to beat them with their new product one massive super chip that will crush anything an really the GPU's that did this were alot of them NV10, NV15, NV 20, NV25, NV38, NV 40, G70, G80, G92-450 and GT200a & GT200b.

Even before the GeForce days, the Nvidia TNT series always proved to be the fastest when it came to 32 bit rendering and such, but back then Glide had the upperhand because most games used the Glide API, thus these games were programmed for 3dfx VGA's like we see alot today that most games are programmed for NVIDIA VGA's and with the DX11 era the same is happening for ATi, though ATi will only succeed if all game companies NVIDIA owns also move to them, which seems to be the only thing which is holding ATi back at the moment they may have the tech and speed and such but they also require the software companies to support their new tech, but this has seemed to have gone quite well since the HD 5000 seires launched.

Ati also has another problem they seem to have a livery shortage a big one too of the 5800 series products, this also can be a problem at times such as these heh, you can have the tech and the VGA power to win but if you can't deliver them, then you'd still face a big problem.

NVIDIA still holds the most game vendors and titles and as long as this still grows they wills till have the upperhand as VGA support goes, ATi's major weakness is driver stabillity and performance, drivers from NVIDIA always had the upper hand over ATi's drivers.

NVIDIA's driver are like that of 3dfx almost bugfree.

Here something I read @ Fudzilla about how the TWIMTBP capmagn goes to work and why it's being used:

"ATI certainly won't apologize for getting DirectX 11 in Dirt 2, STALKER and Alien vs Predator, as they do have DirectX 11 hardware and Nvidia doesn’t. Nvidia will have it, but not in the next few weeks. We simply believe that ATI has to confront Nvidia and its TWIMTBP as these guys make a difference that hurts ATI in the long run. The more games TWIMTBP touches and more PhysX they get inside of the important titles, will generally make ATI's hardware less competitive as it can't support it."

Source:
http://www.fudzilla.com/content/view/15793/1/

more here:
http://www.fudzilla.com/content/view/15794/1/

"So if you want DirectX 10, 11, Anti-aliasing, Shader model 4 or 5 effects, hardware tessellation, PhysX effects, the work of at least 50 engineers at Nvidia makes things look better. You should not criticized them and we should actually thank them for it, and if Nvidia would not do this, the PC games could become an endangered species."

So for some it may read somewhat odd but really I have to admit, Thanks to NVIDIA PC gaming is thriving and yes also to ATi for taking part to a small sector of it DX11 is useless because there is no game that uses it it's more a marketing issue than something useful today and even if DX11 hits the cards will be too slow to render it properly heh.

@Obi, ich sehe keinen zusammenhang des Threades mit 3dfx?

Ansonsten abwarten und tee trinken...

glaub er meint die fake VSA100 karten
aber soweit ich weiss hat 3dfx gesagt das die nicht laufen und es auch noch ne weile dauern wird

Yes I meant them fake 3dfx VSA-100 cards as GAT also mentioned earlier in this thread

Quoted from "GrandAdmiralThrawn"

The fan holes look pretty normal, that's standard, although the PCB seems a bit sawn off indeed. Although I don't actually recognize the PCB from any existing cards.

Well, even if you're right, so what. I remember 3dfx showing off a v5 6000 Fake. They still built a real one in the end, right?

ATi also made fake cards, this is a normal thing VGA vendors do to keep people busy at the dog and pony shows and yes they try to convine people that it works, this also explains why people are not aloud to see what's in a computer case during theser times during demo runs, so basically it's a common thing and nothing special

All they are doing is trying to overconvince the crowd that they are working on it and have a product to overwealm them

OB 384bit oder 512bit, das macht keinen grossen Unterschied. Wenn man allerdinsd von 128 Streamprozessoren auf 512 aufsteigt, DAS ist ein Unterschied. Da macht das Speicherinterface nicht viel, da die CPU eh zu schwach ist fuer die Grafikkarte. Der Durchsatz wird irgendwann doch eh einfach das Limit erreichen, wenn die CPU limitiert. nVidia macht das schon richtig, somit kann man auch die Kosten etwas druecken.
Ich freu mich eher aufs C++ interface fuer die Karte, denn das wird mal ordentlich reinhauen

Also der GT200 lechzt ja gerade zu nach Speicherbandbreite, hab ich mal mit meiner GTX280 getestet. Wenn man nur den GPU Takt erhöht fällt die Mehrleitung recht bescheiden aus, eine Erhöhung des Speichertakts von 1100 auf 1300MHz dagegen bringt dagegen gute 10-15%, soweit ich mich erinnere
Nicht zu vernachlässigen ist allerdings auch der neu integrierte L2 Cache bei Fermi. Insgesamt hat Nvidia die Architektur doch beachtlich umgekrempelt, was eine Leistungseinschätzung umso schwerer macht. Nimmt man alles zusammen, Anzahl Shader, neuer Cache und erheblich höhere Speicherbandbreite (384bit GDDR5 entspräche ja in etwa 768bit GDDR3) könnte man sogar davon ausgehen, dass der GT200 um 100% +x überflügelt wird. Ich glaube auch nicht, dass man bei AMD damit gerechnet hat, dass Nvidia so ein Monster in die freie Wildbahn entlässt - HD5870 und co werden sich auf nen harten Winter einstellen dürfen denk ich mal.

Do keep in mind, even that the bus of the ram is only 384 Bit GDDR5, this does have the same bandwidth as 768Bit GDDR3 GDDR5 doubs the performance of GDDR3, this is why ATi cards used 256Bit GDDR5 instead of 512Bit GDDR3 with their 4800 series

Thats exactly the point, additionally Fermi has a new L2 Cache which can also boost Performance as well
That's why I guess Fermi could have double performance of an GT200, or even more

Quoted from "666psycho"

Thats exactly the point, additionally Fermi has a new L2 Cache which can also boost Performance as well
That's why I guess Fermi could have double performance of an GT200, or even more

Yes that's it indeed man I can't wait you haha I wonder how NV will will don with the GX2 variants, since the GTX 285 had 512Bit GDDR3 and their GTX 295 aka GX2 295 had 448 Bit GDDR3 per GPU , maybe that the GTX 395 may get 320Bit GDDR5 per GPU? just a wild guess as where GTX 360 has 320Bit and GTX 380 would get 384Bit hmm. I do wonder this, I also thought about 2x 256Bit GDDR5 for the GX2 variant not sure though.

with the 8 series it was 384 bot for the 8800 Ultra and 8800 GTX and 320 bit for the 8800 GTS 320 & 640 so this made me think about the GX2 variant since the GTX 295 had the same ram bus as the GTX 260 C192, C216 and GTX 275 as hereby my thoughts for the GTX 395 2x 320Bit or 2x 256bit? hmm I wonder

Quoted from "GrandAdmiralThrawn"

Maybe I'm wrong, but I don't think that nVidia will do a GX2 on a 40nm process. Remember G80? GX2 came out only after the shrink which resulted in the less hot G92 core. Fermi looks like another monstrosity, so I doubt they'll release a GX2 before serious die shrinking. As for the memory interface it depends. Since Fermi uses 64-Bit wide memory controllers both 320 Bits as well as 256 Bits are an option. I guess it mainly depends on what it takes to beat a 5870X2, since a GX2 card will most likely be released for gamers only (no Tesla, Quadro..).

As for the Cache hierarchy novum, it remains to be seen what the hierarchical cache structure can do for regular game engine related stuff like pixel-, vertex- or geometry shaders. Some suspect it's a unified set associative cache system very comparable to what is being used by x86 CPUs right now. We'll see...

For graphical performance only, not taking CPU or platform bottlenecks into account at all, this should mean at least a +100% increase in raw power. Doesn't mean we're gonna see +100% anywhere.. I just hope we get some numbers before new years eve.

well I did find out that the 7900 GX2 like the 7950 GX2 and 9800 GX2 were also produced as Quadro variants:

The 7900 GX2 is the Quadro FX 4500 X2
http://www.nvidia.com/object/quadro_fx_4500_x2.html




compared to the 7900 GX2, it only has 2 DVI-I's




The 9800 GX2 is the Quadro FX 4700 X2
http://www.nvidia.com/object/product_qua…4700_x2_us.html




compared to the 9800 GX2, it only has 2 DVI-I's





The 7950 GX2 and the GTX 295 are the only GX2 variants that were not made as Quadro FX variants, yet

heh Pro ATi websites such as this one are really taking iit all too serious heh
http://www.semiaccurate.com/2009/10/06/n…igh-end-market/

With the cancellation of the GTX285, GTX275, and GTX260, possibly the GTX295 too, Nvidia is abandoning the entire high end and mid-range grahics market. Expect a reprise in January on the low end. The company is badly mismanaged and hated by the very partners they need to throw them a life preserver.

Anyways I don't think it's that bad, Everyone that is against NVIDIA or that likes to make BS stories of some wishfull myth they all hope for, hereby they are making the problems there are; as bigger ones to make them look devistating for NVIDIA it's self, anyways I take such complete nonsense as a massive pile of salt nothing more and nothing less.

If nvidia dies there will be new players in the game!

Quoted from "GrandAdmiralThrawn"

Unlikely. The extreme investments and work necessary to develop a competitive product in this market will most likely not allow anyone to compete anymore. The architectures built by ATi and nVidia have incrementally been developed since over 10 years now. 10 years of optimization, of making given GPU units more efficient and powerful while also adding on to them. 10 years of learning to do, what they do best.

Whoever wants to compete will need comparable levels of experience, know-how and money. Several such startups or restarts have been attempted. Even Matrox failed to compete, and of all competitors they had the best chances of doing it. SiS/Xabre couldn't do it, neither could s3 or the Bitboys, who are now building mobile phone GPUs.. with mediocre success at best.

So, no, there won't be any new players in the game. If nVidia dies, ATi will simply dominate the whole market, with all the consequences that any given monopoly will bring.

The only other player that could be a serious one is Intel. And you know why? Cause no other chip company on earth is as advanced in both development and manufacturing. Plus, they have all the money they want to pump into the project. If the first chip fails, they just build a second one. If the second one fails, they just build a third one. If they are determined, they can simply do anything they want. Buy anybody they want. Build anything they want. But that's Intel, that's one isolated case with no real other match to them.

So, Intel set aside, conclusion is, that there will be no others. Startups are either bought out or destroyed, and there are no other big players left. Period.

Shit, I didn´t know that the situation became so hopeless...