Mysterious Dual GPU Graphic Card (Dual GT300?)

[walterman]

http://vr-zone.com/forums/509194/galaxy-working-on-mysterious-dual-gpu-graphic-card-dual-gt300-.html

http://www.donanimhaber.com/Galaxy_cift_GPUlu_yeni_bir_basyapit_uzerinde_calis iyor-16591.htm

According to Donanimhaber.com, Galaxy one of the most popular Nvidia's partners is working on mysterious dual GPU graphic card. There is no specific information about that card but site claims that it supports 512-bit memory interface. By the way, Galaxy calls that card as Masterpiece!

http://i35.tinypic.com/2ntybk2.jpg

Prolly it is a 2 x 256bit GDDR5 PCB.

I was hoping to see a 512bit GDDR5 GPU monster since i got my GT200b, but, i think that it is not going to happen with this new GT300 gen.

[-=DVS=-]

Wonder what are those chips in between GPU's , one is probably bridge of sorts second maybe fixed decoder for media , how about 3rd ?

[Heinz68]

Sure mysterious since Fermi doesn't have the final tapeout yet.

[jolle]

would be sorta cool if it was a dual GT200 board with onboard Hydra chip.
as a proof of concept atleast.

probably too soon for GT300, specially dual GPU boards.

[LydianKnight]

would be sorta cool if it was a dual GT200 board with onboard Hydra chip.
as a proof of concept atleast

Isn't that the concept behind ASUS MARS? If that's what the design is for, why would the Hydra chip be needed, then?

[Toss3]

Sure mysterious since Fermi doesn't have the final tapeout yet.

They could have gotten the blue-prints already though in order to prepare custom coolers etc. :)

[XMAN52373]

Sure mysterious since Fermi doesn't have the final tapeout yet.

Right! Please dont post, your anti-green pro red colored glasses are making it sick to even read your posts. Nvidia just demo'd working Fermi silicon 2 days ago, kinda hard to do that if it hasn't taped out yet.

:headexplode::retard:

[shadow001]

http://vr-zone.com/forums/509194/galaxy-working-on-mysterious-dual-gpu-graphic-card-dual-gt300-.html

http://www.donanimhaber.com/Galaxy_cift_GPUlu_yeni_bir_basyapit_uzerinde_calis iyor-16591.htm



http://i35.tinypic.com/2ntybk2.jpg

Prolly it is a 2 x 256bit GDDR5 PCB.

I was hoping to see a 512bit GDDR5 GPU monster since i got my GT200b, but, i think that it is not going to happen with this new GT300 gen.


Whatever the card is,it isn't following the PCI-e power specification,since if we look at the upper right side corner where the PCI-e power connectors would be located,we see that it has soldering points for a pair of 8 pin PCI-e power plugs(8 solder bumps each),which isn't part of the PCI-e spec(yet),and that means this card can draw up to 375 watts of power,as the slot provides 75 watts,and each of the 8 pin connectors,another 150 watts a piece.


Fermi is using a 384 bit bus,so that would be 12 memory chips per GPU,given the current GDDR 5 modules available on the market,and the card shows 8 memory modules locations per GPU,so unless there's more of them on the opposite side,that's a 256 bit bus per GPU right there.


Fermi still uses an external NVIO chip to output the final image to the DVI outputs,so that could be one of the middle locations,but the other 2 locations could be for cheaper G92 GPU's used as physics processors....Perhaps an updated GTX295 but with built in physics processing abilities(just a guess though,i really have no idea for sure).

[Heinz68]

Right! Please dont post, your anti-green pro red colored glasses are making it sick to even read your posts. Nvidia just demo'd working Fermi silicon 2 days ago, kinda hard to do that if it hasn't taped out yet.

:headexplode::retard:

What's so hard to understand about final tape out. You don't like my post than just ignore it. Just don't tell me what to post or where to post, like you already did before.

About the NVIDIA demo, they already did demo on the end off September. Maybe one day they will demo something what is at least ready for production.

Wish you didn't make me do it but since Jen-Hsun Huang always puts a smile on my face when I see it, why not.

"I have one sitting right here. This, ladies and gentlemen, this puppy right here, is Fermi." (http://www.youtube.com/watch?v=mJOv3VlkEjQ)
http://img145.imageshack.us/img145/6919/fakefermi.jpg (http://img145.imageshack.us/i/fakefermi.jpg/)

What do you think, was that puppy Fermi?

[XMAN52373]

What's so hard to understand about final tape out. You don't like my post than just ignore it. Just don't tell me what to post or where to post, like you already did before.

About the NVIDIA demo, they already did demo on the end off September. Maybe one day they will demo something what is at least ready for production.

Wish you didn't make me do it but since Jen-Hsun Huang always puts a smile on my face when I see it, why not.

"I have one sitting right here. This, ladies and gentlemen, this puppy right here, is Fermi." (http://www.youtube.com/watch?v=mJOv3VlkEjQ)
http://img145.imageshack.us/img145/6919/fakefermi.jpg (http://img145.imageshack.us/i/fakefermi.jpg/)

What do you think, was that puppy Fermi?

http://www.vizworld.com/2009/11/nvidias-fermi-sc09/

Missed that link a few days ago didn't on the front page of the site didn't ya? WORKING Fermi vs GT200 card using same Simulation. Given your red colored eyes, I'm not surprised you missed it. Eat that and shove it where the sundont shine. No final Tape out my ass. Kinda hard to demo a none tape out product.

[jolle]

Isn't that the concept behind ASUS MARS? If that's what the design is for, why would the Hydra chip be needed, then?
I thought that board still used SLI, or rather I just assumed, as any other NV based Dual GPU boards.
Hydra claims better scaling and more transparency towards the game and no driver profile dependancies and all that..

Im not saying I´d buy one, but It´d be a interesting thing to see.

[LydianKnight]

Galaxy Technology to Release Graphics Card to Rival Asus Mars – Rumour (http://www.xbitlabs.com/news/video/display/20091119193252_Galaxy_Technology_to_Release_Graphi cs_Card_to_Rival_Asus_Mars_Rumour.html)

More likele to be a dual GTX 285, although it's still a rumour.

[harl]

..Fermi still uses an external NVIO chip to output the final image to the DVI outputs,so that could be one of the middle locations,but the other 2 locations could be for cheaper G92 GPU's used as physics processors....Perhaps an updated GTX295 but with built in physics processing abilities(just a guess though,i really have no idea for sure).

http://www.islabit.com/wp-content/imagenes/geforce-gtx295-single-2.jpg

I think it's like the GTX295 the pcie bridge and two NVIO

[josiahsuarez]

http://i35.tinypic.com/2ntybk2.jpg

http://i36.photobucket.com/albums/e16/josiahsuarez/GTX295_card_3.jpg

compared to GTX295 it's very similar indeed, except for the 8+8 pin power instead of 8+6. maybe it's just a new overclocked model with more memory.

[Revs]

NVM

[Revs]

compared to GTX295 it's very similar indeed, except for the 8+8 pin power instead of 8+6. maybe it's just a new overclocked model with more memory.

I see 8+6

[Bman212121]

Right! Please dont post, your anti-green pro red colored glasses are making it sick to even read your posts. Nvidia just demo'd working Fermi silicon 2 days ago, kinda hard to do that if it hasn't taped out yet.

:headexplode::retard:

http://www.vizworld.com/2009/11/nvidias-fermi-sc09/

Missed that link a few days ago didn't on the front page of the site didn't ya? WORKING Fermi vs GT200 card using same Simulation. Given your red colored eyes, I'm not surprised you missed it. Eat that and shove it where the sundont shine. No final Tape out my ass. Kinda hard to demo a none tape out product.

Just because they had an engineering sample doesn't mean it's the final product. They could easily change several components before the card is ready for retail.

[XMAN52373]

Just because they had an engineering sample doesn't mean it's the final product. They could easily change several components before the card is ready for retail.

True, but the problem I take issue with him is he said, "Fermi hasn't even taped out yet" suggesting that what hey are showing/sneak peaking to everyone isn't Fermi. Its first in carnation has taped out, the A2s are in the shop now for, A3 more than likely will be the consumer chip. Heinz is so ANTI Nvidia, how he is allowed to post anything in this site is beyond me, hell, he is propably one of Charlie's chronies.

[josiahsuarez]

In fact, that lower-centre chip is the NF200 I think

http://www.3dnews.ru/_imgdata/img/2009/01/18/110488.jpg

Looks just the same ^



I see 8+6

the board diagram is from donanimhaber website, unknown card referred to as Masterpiece with 8+8 power. the photo I posted is a regular GTX295 with 8+6 power. they are not identical, but very similar.

the diagram also shows 8 memory chips per GPU instead of the 7 on the 295. maybe it has a full 2 gigs now.

[Revs]

Ah, I see :)

[Johnny C]

True, but the problem I take issue with him is he said, "Fermi hasn't even taped out yet" suggesting that what hey are showing/sneak peaking to everyone isn't Fermi. Its first in carnation has taped out, the A2s are in the shop now for, A3 more than likely will be the consumer chip. Heinz is so ANTI Nvidia, how he is allowed to post anything in this site is beyond me, hell, he is propably one of Charlie's chronies.

Wow haven't met Rollo, MTL, and Hell_of_Doom, have you?

[XMAN52373]

Wow haven't met Rollo, MTL, and Hell_of_Doom, have you?

in person, no. on the forums yes. but i'm not running around cap about the compition they they do and Hienz does. Just like they will post anything and everyhting negative about AMD, Hienz stops at nothing to try and do the same about Nvidia regaurdless if his posts are truthful or not. Same can be said for the others aswell.

[Viral]

I think the dual 8-pin wouldn't necessarily mean all the pins are used. The 5970 is the same, has the points on the PCB for dual 8 pin but only uses 8+6pin.

My guess is that it's dual GTX285. Don't know what they hope to charge for it considering the 5970 is out...

[Ninja Prime]

in person, no. on the forums yes. but i'm not running around cap about the compition they they do and Hienz does. Just like they will post anything and everyhting negative about AMD, Hienz stops at nothing to try and do the same about Nvidia regaurdless if his posts are truthful or not. Same can be said for the others aswell.

Heinz isn't 1/10th as bad as any of those 3.

[Heinz68]

True, but the problem I take issue with him is he said, "Fermi hasn't even taped out yet" suggesting that what hey are showing/sneak peaking to everyone isn't Fermi. Its first in carnation has taped out, the A2s are in the shop now for, A3 more than likely will be the consumer chip. Heinz is so ANTI Nvidia, how he is allowed to post anything in this site is beyond me, hell, he is propably one of Charlie's chronies.
You're just full of it. Go back (http://www.nvnews.net/vbulletin/showpost.php?p=2125051&postcount=3) and read what I said if you want to quote me. It was very short sentence and you still managed to twist what I said.

in person, no. on the forums yes. but i'm not running around cap about the compition they they do and Hienz does. Just like they will post anything and everyhting negative about AMD, Hienz stops at nothing to try and do the same about Nvidia regaurdless if his posts are truthful or not. Same can be said for the others aswell.
Same baloney you keep on posting. Show me ONE post, where I posted some lies about NVIDIA. I actually like NVIDIA products, always was happy when I had some and also hope Fermi is going to be some awesome gaming Graphic Card and most likely it will be. In the meantime lets enjoy what AMD/ATI has, they sure need it and deserved it.

As concerned and educated customer I just dislike some of the NVIDIA tactics they use lately same as many many other consumers do. If you don't like it or disagree with me, tough luck.

Just stop the on going personal attack on me, that is not what forums are about. If you want to discuss some other issue than personal you're more than welcome.

If not just ignore my posts and don't tell me what or where to post, how many times I need to repeat it before you get it.