NVIDIA Fermi: Architecture discussion

neliz said:
They got pics man..
Click to expand...
20 more days, one more card? At that pace, they'll be ready for launch in no time.
trinibwoy said:
Nice, they got a Corsair H520 in there :D
Click to expand...
You can tell? If so, that's amazing news as far as Fermi power consumption goes...

[strike]There's something strange going on with that DVI connector, though. Somehow, it doesn't seem properly aligned and the second card doesn't have one.[/strike]
 
Last edited by a moderator:
A chip with 3.2 billion transistors has 7.2 billion vias, a number "which exceeds the world population." He called on the IEDM audience, and Nvidia's main foundry vendor Taiwan Semiconductor Manufacturing Co. Ltd. (TSMC, Hsinchu, Taiwan), to deliver one defect per part per billion (1 DPPB). "We have to make all the vias work; it has to be defect-free."
Click to expand...
More...
 
"We've seen the same problem on the upper metal levels on the ATI chips we've studied. It creates a reliability failure mode."
Click to expand...
Nice. He doesn't explain why we are more immune to the issues though... :cool:
 
Hmmm, that's a big improvement over GT200 density if those numbers are accurate. Probably just due to the bigger caches though.
 
trinibwoy said:
Hmmm, that's a big improvement over GT200 density if those numbers are accurate. Probably just due to the bigger caches though.
Click to expand...
I doubt -- Cypress is still well ahead regarding the register file size compared to Fermi, and that counts too as a dense structure category, along the SRAM arrays for the various caches and buffers.
 
True, and a 3.2 billion tranny Fermi at 400mm^2 would skip ahead of Cypress by a fair margin on the density front. But that 400mm^2 could easily be closer to 450mm^2 depending on how loose they were with the dimensions.
 
Chen said power became a big issue, particularly at the 90 nm node "when power consumption went up so fast." Although strained silicon, power rails, sleep modes and multiple threshold voltages have kept Nvidia's 20 × 20 mm die within a ~130 W power envelope,
Click to expand...
= G80
 
G80 is 480mm^2. The paragraph starts with a reference to a 40nm 3.2b transistor chip. Actually re-reading it they could be referring back to G80 with that sentence.
 
trinibwoy said:
Nice, they got a Corsair H520 in there :D What does his t-shirt say? :???:
Click to expand...

Looks similar to my computer. Not surprising though,. Mine was setup specifically to support 3 Way Fermi. No there aren't Fermi's in these pictures. :p

Code: 
Intel I7 860
Asus P7P55 WS Supercomputer
3 Way SLI capable + small PhysX PPU
8 Gigs DDR 3 1333 ((unimpressive memory types :P))
Silverstone 1.5 Kilowatt PSU
Windows 7 64 Bit Ultimate

[IMG]http://chrisray.soterial.com/chrisraymisc/maingear1.jpg[/IMG]

[IMG]http://chrisray.soterial.com/chrisraymisc/maingear2.jpg[/IMG]

[IMG]http://chrisray.soterial.com/chrisraymisc/maingear3.jpg[/IMG]
 
ChrisRay said:
Looks similar to my computer. Not surprising though,. Mine was setup specifically to support 3 Way Fermi
Click to expand...

Niiiice! So there's an upside to the grief that you have to suffer eh? :D

I installed an H50 and Q9550 a few weeks ago to upgrade my aging P35 setup. Wish I knew that P35 sucked at overclocking quads before that little experiment. Glad to see the H50 going mainstream though - it really is a nice bit of kit.

What do you mean it was setup specifically to support 3-way Fermi? What's different between that and say a 3-way GTX285 setup?
 
You must log in or register to reply here.

Similar threads

hoom
AMD: Zen 4, Speculation, Rumours and Discussion
8 9 10
Replies
199
Views
39K
MfA
M
R
nVidia Denver discussion
2 3
Replies
56
Views
31K
A1xLLcqAgt0qc2RyMz0y
A
J
GPGPU running on FPGA
Replies
16
Views
10K
silent_guy
S
W
Xbox One Architecture Panel *Main Console HW Points* Transcript
3 4 5
Replies
89
Views
34K
Deleted member 11852
D
T
Is Quadro's SP count more Marketing than performance?
Replies
5
Views
5K
trinibwoy
T
Back
Top