NVIDIA GeForce GTX 1180 Bare PCB Pictured

[FordGT90Concept]

You're right, probably isn't DVI-D but something is on there:

 

[jabbadap]

You're right, probably isn't DVI-D but something is on there:

Full size USB3.1 Gen2 maybe?

Edit: Buildzoid's analyse:

 

[FordGT90Concept]

USB-A has less bandwidth than USB-C so I think it's fairly safe to rule out USB-A entirely.

 

[TheLostSwede]

You're right, probably isn't DVI-D but something is on there:

It actually looks a bit like an Ethernet port... Maybe for the Quadro boards?
Thanks for the highlight.

 

[FordGT90Concept]

Ethernet is only 8 contacts. It has 10+10. It is also 20% larger than DisplayPort but that could just be because of the riser structure.

DisplayPort has 20 pins
HDMI has 19 pins
USB-C has 24 pins

DisplayPort is most likely but I've never seen a DisplayPort on a riser like that.

 

[Captain_Tom]

Guys. It's a 400mm^2 die that is clocked slightly higher than the 1080. It will use more energy than the 1080 lol.

This is more-or-less a refresh of the 1080 Ti on 12nm.

 

[Fluffmeister]

Guys. It's a 400mm^2 die that is clocked slightly higher than the 1080. It will use more energy than the 1080 lol.

This is more-or-less a refresh of the 1080 Ti on 12nm.

Soo faster than a GTX 1080 Ti whilst using just a bit more energy that a GTX 1080? NICE!

If only vega was any good, shame.

 

[Captain_Tom]

Soo faster than a GTX 1080 Ti whilst using just a bit more energy that a GTX 1080? NICE!

If only vega was any good, shame.

After 2 years..... 10% more performance for the same price. How impressive /s

Why do so many people have to make this a fanboy thing? I didn't even bring up Vega.


P.S. I do love that my Vega cards have made me thousands of dollars while you drool over a 10% performance increase though lol

 

[Fluffmeister]

After 2 years..... 10% more performance for the same price. How impressive /s

Why do so many people have to make this a fanboy thing? I didn't even bring up Vega.


P.S. I do love that my Vega cards have made me thousands of dollars while you drool over a 10% performance increase though lol

Hmm, I wish I cold borrow your crystal ball. It's a fanboy thing becuase your a big fat fanboy.

This card for one is replacing the 1080, not the 1080 Ti which has no competition either.

You bought Vega? Oh you surprise me.

 

[Captain_Tom]

Hmm, I wish I cold borrow your crystal ball. It's a fanboy thing becuase your a big fat fanboy.

This card for one is replacing the 1080, not the 1080 Ti which has no competition either.

You bought Vega? Oh you surprise me.

"This card replaces the 1080"

It's a significantly bigger card lol, it may have the same last two numbers in it's name - but that's it. It's configuration is almost exactly the same as the 1080 Ti and it will sell for a higher price!

If they called it the GTX 2050 and sold it for $1500, would you claim we are only allowed to compare it to the GTX 1050?! Do some people around here think before they talk? Or.... Think at all for that matter?


Oh, and it shouldn't surprise you anyone bought Vega - it's selling very well. In fact it is taking marketshare from Volta (its actual competitor).

 

[R-T-B]

It's a significantly bigger card lol, it may have the same last two numbers in it's name - but that's it. It's configuration is almost exactly the same as the 1080 Ti and it will sell for a higher price!

What part of the dies "104" nomemclature suggests one bit of that? Didn't we see "GA104" shipping manifests just a post over?

I mean it's all rumor, granted.

 

[Captain_Tom]

What part of the dies "104" nomemclature suggests one bit of that? Didn't we see "GA104" shipping manifests just a post over?

I mean it's all rumor, granted.

Dude, they can call it whatever they want - it's 400mm^2. Keep in mind Nvidia paid TSMC to make significantly bigger dies with their 12nm process, and so it is costing Nvidia extra to make these. Although yes, Nvidia has massive profit margins already...

 

[R-T-B]

Dude, they can call it whatever they want - it's 400mm^2.

But their nonmemclature still suggests at minimum one larger die in this series.

 

[Captain_Tom]

But their nonmemclature still suggests at minimum one larger die in this series.

Yeah, bigger 12nm cards exist already: GV100 is an 815mm^2 die. Most likely the stripped down gaming version will be ~550mm^2 if I had to guess.

Wish they would just go for a 512-bit 800mm^2 monster though, and I would consider paying $1500+ for that.... But AMD won't force them to do that till 2020/21 most likely :'(.

 

[bibob94]

Honestly, if AMD didn't give a good competition, Nvidia will sell their new flagship gpu around $500 or more, they are being a dick recently, and also the availability of the gpu it self, I don't have any good feeling about this new release

 

[efikkan]

That new SLI port is interesting. It looks remarkably like an inverse NVLink, and I know there was some talks of NVLink coming to GeForce a couple of years ago…
It's no coincidence that the GV102 prototype board featured the same thing.

Guys. It's a 400mm^2 die that is clocked slightly higher than the 1080. It will use more energy than the 1080 lol.

This is more-or-less a refresh of the 1080 Ti on 12nm.

No, GTX 1080 Ti (GP102) have a different memory controller and core infrastructure. The GV-series have a different SM-layout than Pascal.

GV104 is a scaled down GV100, no relation to GP102.

Why do so many people have to make this a fanboy thing? I didn't even bring up Vega.

Because you keep complaining when Nvidia is pushing the industry forward, without arguments and reasoning to justify your complaints.

Wish they would just go for a 512-bit 800mm^2 monster though, and I would consider paying $1500+ for that.... But AMD won't force them to do that till 2020/21 most likely :'(.

Nvidia gave up 512-bit memory controllers years ago since they were too complex and costly. None of their upper models since then have been bottlenecked by memory bandwidth. Nvidia have managed to squeeze out more performance than their counterpart, even though they sometimes have less memory bandwidth.

 

[TheLostSwede]

Ethernet is only 8 contacts. It has 10+10. It is also 20% larger than DisplayPort but that could just be because of the riser structure.

DisplayPort has 20 pins
HDMI has 19 pins
USB-C has 24 pins

DisplayPort is most likely but I've never seen a DisplayPort on a riser like that.

But if you compare the solder points for the other DP connectors, it doesn't look right either. It would also block two DP ports, so it's in lieu of two other ports.
I didn't say it was Ethernet, just that it looks somewhat like it.
I really have no idea what it's for and I've seen more connector types than most people...
Looks like we're going to have to wait and see...

 

[yeeeeman]

Given GTX 1080Ti has 11GB, it would be weird to release its successor with WAY less. 11GB was weird number by itself. Odd numbers are rare and are very weird. 3GB HD7900 cards, 5GB GTX 1050Ti's, 11GB GTX 1080Ti, 3,5GB GTX 970 etc These are very rare actually.

GTX1180 is not the successor of GTX1080Ti. Yeah, it just happens to be on the same level, maybe even better, but the price class is not the same.

 

[RejZoR]

It is of GTX 1080. Ti's are always separate level and would be replaced with Ti version of whatever comes next.

 

[jabbadap]

But if you compare the solder points for the other DP connectors, it doesn't look right either. It would also block two DP ports, so it's in lieu of two other ports.
I didn't say it was Ethernet, just that it looks somewhat like it.
I really have no idea what it's for and I've seen more connector types than most people...
Looks like we're going to have to wait and see...

Two slotter and raised hdmi port? Well it is what it is, it might not be even used at all.

 

[Vayra86]

It has model PG180 written on it twice and 180 appears in the model extended model number.


GTX 1080 has PCB number 180-1G413-DAAA-A00 where this PCB is 180-1G180-DAAA-A02. Looks like an iteration on the previous design.

Haha we might still get a Pascal Refresh with minor jumps at same TDP and lower price like we did back with 7xx.

I'll be laughing at all those hype trains saying 'will go faster than 1080ti' Which I already am doing ... in fact.

 

[efikkan]

Some of the (possible) features to look forward to:
- Increased core count:
GV102 will feature up to 5376 cores, 84 SMs, 336 TMUs (+40% vs. full GP102)
GV104 will feature up to 3584 cores, 56 SMs, 224 TMUs (+40% vs. GP104)
- Marginal node improvement
- New SLI bridge
- RTX raytracing
- VirtualLink support? (speculation)
- HDMI 2.1 support? (speculation)
- Full fp16 support? (or will Nvidia keep saving this for a rainy day?)

Some features not to look forward to:
- More absurd boosting?
- The 2-3 usual gimmicks.

 

[Vayra86]

Some of the (possible) features to look forward to:
- Increased core count:
GV102 will feature up to 5376 cores, 84 SMs, 336 TMUs (+40% vs. full GP102)
GV104 will feature up to 3584 cores, 56 SMs, 224 TMUs (+40% vs. GP104)
- Marginal node improvement
- New SLI bridge
- RTX raytracing
- VirtualLink support? (speculation)
- HDMI 2.1 support? (speculation)
- Full fp16 support? (or will Nvidia keep saving this for a rainy day?)

Some features not to look forward to:
- More absurd boosting?
- The 2-3 usual gimmicks.

I think for this generation you can put RTX under the gimmick category. Highly doubt we will see much content with that tech actually bringing something better than what we have. So far we've seen a tech demo and a weak Metro demo.

 

[FordGT90Concept]

But if you compare the solder points for the other DP connectors, it doesn't look right either. It would also block two DP ports, so it's in lieu of two other ports.
I didn't say it was Ethernet, just that it looks somewhat like it.
I really have no idea what it's for and I've seen more connector types than most people...
Looks like we're going to have to wait and see...

When they use a riser, they need pins through the PCB to give it more rigidity. Just contact solder points are too frail. The ones that re right next to he board have two ground contacts through the board to take care of that, the one on the riser has two ground contacts and 10 pins through the board. It may even have that hole between them as a screw point to further strengthen the connector. Perhaps it can drive five displays simultaneously without using DisplayPort hubs.

Some of the (possible) features to look forward to:
- Increased core count:
GV102 will feature up to 5376 cores, 84 SMs, 336 TMUs (+40% vs. full GP102)
GV104 will feature up to 3584 cores, 56 SMs, 224 TMUs (+40% vs. GP104)
- Marginal node improvement
- New SLI bridge
- RTX raytracing
- VirtualLink support? (speculation)
- HDMI 2.1 support? (speculation)
- Full fp16 support? (or will Nvidia keep saving this for a rainy day?)

Some features not to look forward to:
- More absurd boosting?
- The 2-3 usual gimmicks.

Or Volta is never coming to consumers and is relegated to the Tesla, Quadro, and Titan. There's a significant amount of silicon on Volta that is completely useless for gaming. NVIDIA invented ATAA to use that RSX silicon on Volta in display tasks but the only people that would consider using that feature will likely already have the Volta Titan already (and anybody that needs it will buy the Volta Titan).

Ampere or Turing is more likely without RTX.

Haha we might still get a Pascal Refresh with minor jumps at same TDP and lower price like we did back with 7xx.

I'll be laughing at all those hype trains saying 'will go faster than 1080ti' Which I already am doing ... in fact.

It sort of makes sense. Pascal was mostly a minor refresh of Maxwell and it really only took off in performance because of the 16nm node versus 28nm node. Ampere or Turing could easily be a 12nm refresh of Pascal.


Back on topic, I randomly ran into this picture of Volta on PCIE:

The DVI connector on there is positioned exactly where the connector is on the PCB shot. DVI-D single link only requires 19 pins (maybe 18 pins). It may be a DVI plug on a riser which makes complete sense (very common).

 

[efikkan]

I think for this generation you can put RTX under the gimmick category. Highly doubt we will see much content with that tech actually bringing something better than what we have. So far we've seen a tech demo and a weak Metro demo.

It will certainly take some time before more than 2-3 games utilizes it, but the bigger question is if this is more a tech demo (like with tessellation) or if it's mature enough to be standardized.

Secondly, the usefulness in games will probably be limited in this generation. Accelerated raytracing will certainly be much faster than ever before, but I'm still not convinced that a GV102 will be fast enough for most top games to embrace it.

I wouldn't rush to buy one because of this feature, but still it's good that someone pushes the industry forward. And raytracing as a technology is not a gimmick, but we still don't know what the result of this attempt will be.

Or Volta is never coming to consumers and is relegated to the Tesla, Quadro, and Titan. There's a significant amount of silicon on Volta that is completely useless for gaming. NVIDIA invented ATAA to use that RSX silicon on Volta in display tasks but the only people that would consider using that feature will likely already have the Volta Titan already (and anybody that needs it will buy the Volta Titan).

They already have Tesla and Quadro well covered with GV100. Do you seriously think they made GV102 and GV104 just for them?