rendezvous
Regular
Well, he may have been in Singapore, but that's hardly evidence that he was in a store room and saw these new Nvidia grapics card boxes.
Last edited by a moderator:
NVIDIA GT200 Rumours & Speculation Thread
- Thread starter Arun
- Start date
- Status
Well, he may have been in Singapore, but that's hardly evidence that he was in a store room and saw these new Nvidia grapics card boxes.
And finally, there is a leak, we waited for:
Mechanical drawing of NV upcoming D10U P651
If picture will removed:
http://img216.imageshack.us/img216/1631/newcardoi3fp6.png
- monster-die
- 16 chips -> 512-Bit
- 6+8Pin -> up to 300W
- NVIO?
Mechanical drawing of NV upcoming D10U P651
If picture will removed:
http://img216.imageshack.us/img216/1631/newcardoi3fp6.png
- monster-die
- 16 chips -> 512-Bit
- 6+8Pin -> up to 300W
- NVIO?
Please tell me those are supposed to be two different SKUs, because I see no reason to have a dual PCB graphics card with dual PCI-e connections... What the hell kind of mobo would that plug into anyway?
edit: upon closer inspection they definitely are. One card uses 2x6 pin power connectors and the other is 2x8 pin.
edit: upon closer inspection they definitely are. One card uses 2x6 pin power connectors and the other is 2x8 pin.
Honestly I wouldn't give it too much credence. I could probably throw something together pretty quickly that would look very similar (I use Solidworks on a daily basis). It's just a simple model in a viewer. edrawings most likely, or maybe Solidworks viewer. You can tell by the green e...
If that is to be true... it seems like NV-IO chip makes it come back again... but this round there are 2 type of the chip if we look at the connector-end of the board in that picture
Edit: Just look carefully, one of which has got 2 SLi connector for one type of that end chip.. it would be a GTX card and the other one has got 1 SLi connector which may be a GT card... Another point to be noticed is that why they split power connector for one on the far right and one for the far left of the board???
I migh be wrong through....
Last edited by a moderator:
satein, I'm not quite sure I understand what you think is different b/w the two cards rendered there, but both boards seem to be identical. Some of the components of the left card are showing through the right card (e.g., poking through the NVIO, and making it look like the left card has an extra six-pin power connector next to the eight-pin one at the end of the board).
And, yeah, the six-pin power connector right next to the outputs is veird!
And, yeah, the six-pin power connector right next to the outputs is veird!
Sorry my bad english
Just look closely and you will see that both chips those susposed to be NV-IO... they are not the same. They are different
One for 2 SLi connector and One for single SLi connector That was what I want to say.And for the Power connector on both end of the card... it, yes, looks really weird.
Don't apologize if you're speaking a second language better than me!
I thought you meant the two NVIOs, b/c I noticed them first, too. They might look different at first, but I think a closer look will show the difference to be the viewer application's rendering errors: components of the card on the left showing through components of the card on the right. Compare the left card's NVIO and the right's card's components under its 8-pin power adapter, then look at the right card's NVIO. You can see that most differences b/w it and the left card's can be explained by imagining elements of the left card showing through the right card. As to why the extra green perimeter of the right card's NVIO is incomplete, it lines up with the right edge of the left card, so it's an error of insufficient Z precision or something.
(I'll refrain from mentioning the horrible moire and edge aliasing in this 3D rendering of a 3D renderer. Whoops, too late.)
I thought you meant the two NVIOs, b/c I noticed them first, too. They might look different at first, but I think a closer look will show the difference to be the viewer application's rendering errors: components of the card on the left showing through components of the card on the right. Compare the left card's NVIO and the right's card's components under its 8-pin power adapter, then look at the right card's NVIO. You can see that most differences b/w it and the left card's can be explained by imagining elements of the left card showing through the right card. As to why the extra green perimeter of the right card's NVIO is incomplete, it lines up with the right edge of the left card, so it's an error of insufficient Z precision or something.
(I'll refrain from mentioning the horrible moire and edge aliasing in this 3D rendering of a 3D renderer. Whoops, too late.
)Thank you for kindly reply :smile:Don't apologize if you're speaking a second language better than me!
I thought you meant the two NVIOs, b/c I noticed them first, too. They might look different at first, but I think a closer look will show the difference to be the viewer application's rendering errors: components of the card on the left showing through components of the card on the right. Compare the left card's NVIO and the right's card's components under its 8-pin power adapter, then look at the right card's NVIO. You can see that most differences b/w it and the left card's can be explained by imagining elements of the left card showing through the right card. As to why the extra green perimeter of the right card's NVIO is incomplete, it lines up with the right edge of the left card, so it's an error of insufficient Z precision or something.
(I'll refrain from mentioning the horrible moire and edge aliasing in this 3D rendering of a 3D renderer. Whoops, too late.
I think that doggy rendering on the drawing is due to the drawing file was not opened in its application, but the 3rd party viewer
That means so many incompatibility can occur due to different type of object draw on different 3D drawing software. Probably that drawing was exported as the IGES format someting (standard solid modelling for porting between CAD Software) from the original drawing and put on the viewer... and finally got broken on that viewer.Still crurious about those power connector anyway
Is memory channel width orthogonal to the RAM chip, or, if we're expecting GDDR3 and 1GB of it, do we extrapolate from G80's 12 chips and 768MB?
satein, the OP of those pics has supplied a "STEP file." Dunno what it's native to, or if it was a file native to another program and ported to STEP.
satein, the OP of those pics has supplied a "STEP file." Dunno what it's native to, or if it was a file native to another program and ported to STEP.
some better renders of the step file:
http://www.xtremesystems.org/forums/...1&d=1210392484
http://www.xtremesystems.org/forums/...1&d=1210392484
http://www.xtremesystems.org/forums/...1&d=1210392484
http://www.xtremesystems.org/forums/...1&d=1210392484
For 512-Bit 16 chips (GDDR is only available in 32-Bit) are needed.
It rather seems like a rendering bug.
- Status
