Monday, August 20th 2018

NVIDIA GeForce RTX 2080 ASIC is "TU104," Smiles for the Camera

Here are some of the first pictures of the "TU104," the chip at at heart of NVIDIA GeForce RTX 2080, in its top-spec TU104-400A-A1 variant. While the package itself isn't bigger than that of its predecessors, the GP104 and GM204, the die of the TU104 is huge! If it's built on existing silicon fabrication processes such as 14 nm or 12 nm, the TU104 could end up with transistor-counts higher than those of even the GP102.

You'll recall that we posted bare-PCB pictures of this board recently. In its production (or close-to-production) variant, the board draws power from a combination of 6-pin and 8-pin PCIe connectors, conditioning it with an 8+2 phase VRM. The GPU is surrounded by eight GDDR6 memory chips. Display connectivity includes three DisplayPorts, an HDMI, and a USB type-C (VirtuaLink).
Add your own comment

10 Comments on NVIDIA GeForce RTX 2080 ASIC is "TU104," Smiles for the Camera

#2
comtek
USB C on a GPU? What's it for? Why not thunderbolt instead?
Posted on Reply
#3
Vayra86
Vya Domus said:
Every time I see the name TU104 I can only think of this : https://en.wikipedia.org/wiki/Tupolev_Tu-104

And that definitely looks like a blower style cooler, guess that reference design render probably is fake.
If these GPUs are as reliable as the plane... I'll pass :)
Posted on Reply
#4
noel_fs
comtek said:
USB C on a GPU? What's it for? Why not thunderbolt instead?
VR
Posted on Reply
#5
jabbadap
comtek said:
USB C on a GPU? What's it for? Why not thunderbolt instead?
Like said it's for VR headsets:
What Is VirtualLink™?

VirtualLink is an open industry standard developed to meet the connectivity requirements of current and next-generation virtual reality (VR) headsets. VirtualLink is a consortium defined Alternate Mode of USB Type-C™ designed to deliver the power, display, and data required to power VR headsets through a single USB Type-C connector.
Posted on Reply
#6
ppn
This double decker thing again in place of DVI.
This GPU is meant for 7nm, 500 mm2 gets shrinked to 300mm2 like normal. And the clocks at 7nm are 1.40x higher. So why release this GPU now.
Posted on Reply
#7
Midland Dog
ppn said:
This double decker thing again in place of DVI.
This GPU is meant for 7nm, 500 mm2 gets shrinked to 300mm2 like normal. And the clocks at 7nm are 1.40x higher. So why release this GPU now.
Because its been 2 years since GTX1080 and the nvidia market has refreshed every 2 years since i have been into PC gaming (around the 780 is when i started getting into PC gaming rather than playing my PS1 lmao)
Posted on Reply
#8
jabbadap
ppn said:
This double decker thing again in place of DVI.
This GPU is meant for 7nm, 500 mm2 gets shrinked to 300mm2 like normal. And the clocks at 7nm are 1.40x higher. So why release this GPU now.
TSMC made 12nm FFN process line for especially for nvidia. It might be part of the deal, that they will make more chips on it than just GV100.
Posted on Reply
#9
renz496
ppn said:
This double decker thing again in place of DVI.
This GPU is meant for 7nm, 500 mm2 gets shrinked to 300mm2 like normal. And the clocks at 7nm are 1.40x higher. So why release this GPU now.
because they need to release it right now. simple as that.

jabbadap said:
TSMC made 12nm FFN process line for especially for nvidia. It might be part of the deal, that they will make more chips on it than just GV100.
not only that, 12nmFFN was build specifically for nvidia architecture. with right process optimization it could end up being more efficient than 7nm itself. but even if 7nm end up being better there is no way nvidia going to use the "new" 7nm process while it is still very expensive with yield that probably still not at optimal level to meet the demand by non pro consumer. worse nvidia will have to fight for fab capacity with big player such as Qualcomm/Apple. and we know which customer TSMC will always prefer. nvidia already learnt their lesson when they have to skip 20nm before.
Posted on Reply
#10
coonbro
looking at the pic's one thing for sure is they did not skimp on paste or did it all smush off the chip and off on to the sides ? sloppy .
Posted on Reply