GPU [Rumor] Leaked pictures of RTX 3080

[ParallaxStax]

I'll start with the pictures:

Thoughts and implications for the SFF world:

• WTF is up with the fan placement?
• The fan in line with the PCB doesn't seem like a good idea with sandwich cases, maybe better for temps in Ncase type cases?
• End mounted power connectors adds additional length
• 280 mm long? Guessing based on pcie slot length
• Short PCB adds to hope for ITX variants
• "RTX 3080" text is upside down in most sandwich cases

 

[scatterforce]

@ParallaxStax Some console style cases could work well with venting on both sides.

The letters won't be visible in a sandwich case; they are on the backplate.

I really hope we get an itx variant; we missed out on the 2080/2080S/2080ti. If the PCB can be shortened that much on the founders card, then OEM's like Zotac should be able to come through for the community.

 

[AlexTSG]

I suspect this design is under consideration for Nvidia’s Founders Editions, and we can expect to see more traditional designs from the usual manufacturers.

 

[VegetableStu]

a design like that would mean the PCB would stop short of the rear fan

... more ITX cards, anyone? o_o

 

[max31092]

The design could produce better airflow in bigger cases. So not the worst idea.

But the placement of the power adapters seems quite unlogical to me, so I'd guess it's probably a fake.

 

[scatterforce]

The design could produce better airflow in bigger cases. So not the worst idea.

But the placement of the power adapters seems quite unlogical to me, so I'd guess it's probably a fake.

The founders edition of the 2060/2060S had power connectors at the end just like this one. Looking at the source and the details in the leak, I think this is legit. I really like the look from a design and performance perspective.... I just hope we get some better ITX options this generation.

 

[max31092]

The 2060 has this weird design as well? I also think the other pictures look too real do be fake. But I actually couldn't believe someone would make such a placement for the power connectors.
But if the 2060 is the same I stay corrected.

 

[Phuncz]

While I like the design in a bubble, I'm having a hard time thinking of a good reason in standard ATX cases (where this is designed for most likely) that this would be a good idea. Considering CPU heatsinks will probably introduce more heat or atleast give issues with airflow with the fan "on the back". Unless the consideration is that there is an AIO in almost all people's builds which isn't a terrible idea in 2020, but still I'm wondering what problem they're trying to solve with this design.

The passthrough rear cooling is interesting, although it has been done before. The 90° bended heatsink fins are looking very nice. I secretly hope this design is the new "Founders Edition".

 

[VegetableStu]

The passthrough rear cooling is interesting, although it has been done before.

i want board partners to do that for the 30 series soooooooooo badly ¡-¡

EDIT: actually going back to the rear fan: i'd imagine that'll be a nightmare for the PSU cables in cases like the Dan A4 o_o

 

[Valantar]

The passthrough rear cooling is interesting, although it has been done before. The 90° bended heatsink fins are looking very nice. I secretly hope this design is the new "Founders Edition".

Note that both fans are intakes, so this has a very odd flow path - "front" fan over the GPU and VRAM seems to exhaust a bit out the back + through the fins in the middle, while the "rear" fan blows straight through, taking in air from the RAM/CPU socket area in a standard ATX layout and blowing it towards the PSU shroud. While the design is interesting, it does look horribly inefficient from a thermal standpoint. Note that the rear fan blocks any vapor chamber from extending to the fins on the inner ⅓ of the card, so either they are only for VRM cooling and have their own heatpipes for that, are connected to the vapor chamber through heatpipes, or the design has regressed to a heatpipe-based design rather than Nvidia's recent vapor chamber based reference coolers. Combine that with the highly dubious flow path for the front fan (an L-turn through a minimal fin stack on the I/O side and a U-turn out through the exposed fins on the card on the opposite side, where most of the air is likely to escape out before passing over much fin area) and there is potential for trouble. The design is interesting, but I'm skeptical.

 

[tinyitx]

An unofficial rendering:-

https://uploads.disquscdn.com/images/50c12b7a42983e17389ebdb6ae5e8b169988a3ec2d56912c27e1550c0b248fe3.gif

Some (in Techpowerup) made a front rendering. Based on the fins orientation, I suppose the airflow will be like this.

The other fan on the back of the card will probably do a similar airflow.
Looks like the the flow path of the 2 fans will not cross path with each other.
And, it looks like airflow management will be a little challenge.

 

[VegetableStu]

Some (in Techpowerup) made a front rendering.

oh he's Ace Failure in the SFFn discord o_o

 

[AlexTSG]

The other fan on the back of the card will probably do a similar airflow.
Looks like the the flow path of the 2 fans will not cross path with each other.
And, it looks like airflow management will be a little challenge.

I saw some speculation that the fan on the back may pull air through the heat sink. This would make airflow management easier if it is the case.

 

[Valantar]

An unofficial rendering:-

https://uploads.disquscdn.com/images/50c12b7a42983e17389ebdb6ae5e8b169988a3ec2d56912c27e1550c0b248fe3.gif

Some (in Techpowerup) made a front rendering. Based on the fins orientation, I suppose the airflow will be like this.

The other fan on the back of the card will probably do a similar airflow.
Looks like the the flow path of the 2 fans will not cross path with each other.
And, it looks like airflow management will be a little challenge.

That neatly illustrates the main issue with this design: even if the lack of a shroud makes for a thicker than usual fin stack, that fin area is tiny compared to most open-air coolers. Also, the lack of a shroud means nothing directing flow, meaning the air passing through those two \ / fin stacks in the middle of the card will be negligible - the majority of it will take the path of least resistance and get out of the fin stack ASAP (blue lines) rather than following the fins all the way out. As such the red arrows there are highly unrealistic. Also, part of that fin stack will receive pretty much no air at all (orange fields) - their only access to air from the fans would be a tiny channel in the middle, underneath the silver bar. Overall, for this to be even remotely effective there needs to be heat pipes running from that part of the cooler and into the other. If that is indeed the case, this could be good, though it would lead to a lot of recirculation unless there are other fans around to ensure a steady supply of fresh air.

 

[tinyitx]

That neatly illustrates the main issue with this design: even if the lack of a shroud makes for a thicker than usual fin stack, that fin area is tiny compared to most open-air coolers. Also, the lack of a shroud means nothing directing flow, meaning the air passing through those two \ / fin stacks in the middle of the card will be negligible - the majority of it will take the path of least resistance and get out of the fin stack ASAP (blue lines) rather than following the fins all the way out. As such the red arrows there are highly unrealistic. Also, part of that fin stack will receive pretty much no air at all (orange fields) - their only access to air from the fans would be a tiny channel in the middle, underneath the silver bar. Overall, for this to be even remotely effective there needs to be heat pipes running from that part of the cooler and into the other. If that is indeed the case, this could be good, though it would lead to a lot of recirculation unless there are other fans around to ensure a steady supply of fresh air.

This kind of design is more or less a copy from the CPU's.
Like this one:-

https://images-na.ssl-images-amazon.com/images/I/41h-TP9eu7L._AC_.jpg

They might even modify the shroud and put a piece of transparent cover around the fan to prevent the air flowing right back out of the fins rather than flowing through the fins (ie the brown shaded area).
Regardless, if Nvidia does use a new reference design, it will be both very interesting and challenging.

 

[adesignaddict]

Mod edit: joined similar threads

Have you guys seen this?

Seems a bit problematic for a lot of SFF cases out there. (Note that both the top and bottom of the card are pictured here, rotated by 180 degrees in two directions)

Edit: Ah my bad. Thanks for moving it to the right thread

 

[Valantar]

This kind of design is more or less a copy from the CPU's.
Like this one:-

https://images-na.ssl-images-amazon.com/images/I/41h-TP9eu7L._AC_.jpg

They might even modify the shroud and put a piece of transparent cover around the fan to prevent the air flowing right back out of the fins rather than flowing through the fins (ie the brown shaded area).
Regardless, if Nvidia does use a new reference design, it will be both very interesting and challenging.

Sure, coolers like that both exist and work okay, but they generally aren't meant to cool 250W GPUs, nor do their fin stacks extend as far away from the fan as this design does. Even with a plastic cover over the fins surrounding the fan, pretty much no air will reach the ends of the diagonal fins as the majority of air will leave the fin stack as quickly as possible, and the areas marked in orange in my sketch will still receive next to no airflow at all due to fin orientations and flow paths. On the other hand the orange field sits pretty much directly on top of where the VRMs are likely to be, so it might be intended as a semi-passive VRM fin stack. That still leaves the question of how to get heat from the die to the inner fin stack/fan area though. If this is indeed real it will be very interesting to see it torn down.

 

[tinyitx]

. If this is indeed real it will be very interesting to see it torn down.

I hope it is real. If it is, then it must work. Nvidia is not stupid enough to release a non working thermal solution.
So, this new design must be, somehow, superior to what Nvidia's previous designs.
I love to see new designs.
Personally speaking, this design does not seem to pose a problem to my upcoming built in a Deng F15.
But, I bet all AIBs are going to use their own traditional thermal solution.

 

[tinyitx]

More 'exposed' details of this cooling design.
Take a look:-

RTX 3080绝密开发中：功耗高达320W、散热器成本超千元

这几天，NVIDIA下一代安培架构高端新卡RTX 3080的工程样卡被曝光，别出心裁的前后双面风扇错位设计引发强烈关注。现在，更多曝料来了，还有一些规格甚至是成本方面的消息。据称，NVIDIA正在

news.mydrivers.com

2 side notes:-
a. there is a 3090 (=Titan)
b. the high tbp might require a larger cooler size of any AIB ITX version.


Edit:- another 'leaked' pic of the cooler.

https://static.chiphell.com/forum/202006/10/180235lki7096h3wphz97w.png

 

[VegetableStu]

yeah there won't be airflow (or good airflow at least) from the primary fan directly above the GPU. the channels are flushed off