More Pictures of NVIDIA's Cinder Block-sized RTX 4090 Ti

erek

erek

[H]F Junkie
Joined
Dec 19, 2005
Messages
10,906
It’s too big

“The card's PCB isn't conventional—not perpendicular to the plane of the motherboard like any other add-in card—but is rather along the plane of the motherboard, with additional breakaway daughter cards interfacing with the sole 12VHPWR power connector, and the PCIe slot. This slender, ruler-shaped PCB spans the entire length of the card, without coming in the way of its heat dissipation surfaces. The length is used for the large AD102 ASIC that's probably maxed out (with all its 144 SM enabled), twelve GDDR6X (possibly faster 23 Gbps), and a mammoth VRM that nearly maxes out the 600 W continuous power delivery design limit of the 12VHPWR.”

1687787146948.png

Source: https://www.techpowerup.com/310516/...cinder-block-sized-rtx-4090-ti-cooler-surface
 
I can see how the unconventional design would be more efficient for cooling, but that is in no way practical for a consumer or even a prosumer card. The current designs already push PC cases and motherboards beyond the limit, and this would make it even worse. Why not go for a factory water cooling option? Modern cases are almost all designed with water cooling in mind, having diverse mounting options for integrated systems or multiple AIO solutions. In my opinion it's the only sensible solution moving forward if we're going to keep pushing power consumption like this. And nobody should be trying to replicate ASUS's ridiculous plexiglass encasing shown off recently.
 
erek said:
It’s too big
Click to expand...
Armenius said:
I can see how the unconventional design would be more efficient for cooling, but that is in no way practical for a consumer or even a prosumer card. The current designs already push PC cases and motherboards beyond the limit, and this would make it even worse. Why not go for a factory water cooling option? Modern cases are almost all designed with water cooling in mind, having diverse mounting options for integrated systems or multiple AIO solutions. In my opinion it's the only sensible solution moving forward if we're going to keep pushing power consumption like this. And nobody should be trying to replicate ASUS's ridiculous plexiglass encasing shown off recently.
Click to expand...
No one really cares.

It could take 6000 watts to power and people would still buy it. So long as it’s the most performant.

The GPU could be the size of a full tower case needing a cooling system the size of a refrigerator and people would still buy it.

If anything they’ll label it “retro” and talk about how it looks like a mainframe. “I can’t move my PC without a forklift. How retro!”
 
UnknownSouljer said:
No one really cares.

It could take 6000 watts to power and people would still buy it. So long as it’s the most performant.

The GPU could be the size of a full tower case needing a cooling system the size of a refrigerator and people would still buy it.

If anything they’ll label it “retro” and talk about how it looks like a mainframe. “I can’t move my PC without a forklift. How retro!”
Click to expand...
you into performance, or what?
 
UnknownSouljer said:
No one really cares.

It could take 6000 watts to power and people would still buy it. So long as it’s the most performant.

The GPU could be the size of a full tower case needing a cooling system the size of a refrigerator and people would still buy it.

If anything they’ll label it “retro” and talk about how it looks like a mainframe. “I can’t move my PC without a forklift. How retro!”
Click to expand...
I've purchased every "early gen" top-tier consumer card from NVIDIA since the GTX Titan X. It's the form factor that is the issue, not the power consumption. If I did not already have a 4090, I would still not buy a 4090 Ti if this is what it will be in its final consumer form. People will still buy it, of course, but I feel the form factor is a serious issue considering the complaints on the size of the standard 4090. I'm sure it will turn a good portion of people away, especially since the unconventional design means you can't easily tear it apart to watercool it yourself.
 
erek said:
you into performance, or what?
Click to expand...
Not in absolute performance.
Armenius said:
I've purchased every "early gen" top-tier consumer card from NVIDIA since the GTX Titan X. It's the form factor that is the issue, not the power consumption. If I did not already have a 4090, I would still not buy a 4090 Ti if this is what it will be in its final consumer form. People will still buy it, of course, but I feel the form factor is a serious issue considering the complaints on the size of the standard 4090. I'm sure it will turn a good portion of people away, especially since the unconventional design means you can't easily tear it apart to watercool it yourself.
Click to expand...
I think those affected will be minimal. A literal refrigerator sized cooler is still fine. For everyone else, there likely will be a board partner that will make a blocked card. At a $500+ premium I’m sure.
 
I may have to skip this just to save my onboard sound card. That extra slot is going to smother it. I was blessed to snag the 4090 FE and keep the size reasonable. This might be too much. Honestly the 4090 is such a beast at 4k, I don't even need a Ti. Would be the first top end GPU release I've skipped in years, it's just that good.
 
Last edited:
Armenius said:
I've purchased every "early gen" top-tier consumer card from NVIDIA since the GTX Titan X. It's the form factor that is the issue, not the power consumption. If I did not already have a 4090, I would still not buy a 4090 Ti if this is what it will be in its final consumer form. People will still buy it, of course, but I feel the form factor is a serious issue considering the complaints on the size of the standard 4090. I'm sure it will turn a good portion of people away, especially since the unconventional design means you can't easily tear it apart to watercool it yourself.
Click to expand...

I suspect most people buying a 600W video card are probably going to slap a water block on it anyway....

Which makes me wonder how this daughter-board mashup will work with water-blocks, and what the volume will be of these. If it will even be worth it for the block makers to spend the money coming up with custom designs for it only to sell a handful.

I'm guessing this thing will probably be priced higher than my first car, which means that while I could probably afford it, I think I would be disinclined to actually do so. The 4090 as it is already tackles everything I can throw at it at 4k without a problem. At this moment I don't see a reason to spend the money and go through the effort making this thing fit for added performance I don't need.

I suspect the 4090 will be just fine until next gen, and thus while the "It's the fastest" still does have an appeal to me, I'll probably just pass on it.
 
It seems odd to change the pcb so drastically for a tier refresh GPU. And the current 4090 heatsink/pcb design is shorter than a 3090 but has 600W capability.

On the one hand, the size indicates that they are still aiming for the 600W power target, maybe more? 4 slots geez, but that would be the only way to horizontally mount the GPU+vRam.

Does this improve the cooling of the gddr6x somehow, maybe leading to significantly higher clocks on the vRam? It appears the chips still directly surround the GPU, and that there is even less space...

Maybe it's an experimental design and nothing more. Research and all that.

If the FE went this route, I don't see why AIB's would necessarily follow suit. More information would be needed as to the reasoning before we would know for sure the route the board partners could choose.
 
GoodBoy said:
It seems odd to change the pcb so drastically for a tier refresh GPU. And the current 4090 heatsink/pcb design is shorter than a 3090 but has 600W capability.

On the one hand, the size indicates that they are still aiming for the 600W power target, maybe more? 4 slots geez, but that would be the only way to horizontally mount the GPU+vRam.

Does this improve the cooling of the gddr6x somehow, maybe leading to significantly higher clocks on the vRam? It appears the chips still directly surround the GPU, and that there is even less space...

Maybe it's an experimental design and nothing more. Research and all that.

If the FE went this route, I don't see why AIB's would necessarily follow suit. More information would be needed as to the reasoning before we would know for sure the route the board partners could choose.
Click to expand...

AIB versions are gonna be 24ft long with 5 cooling fans
 
UnknownSouljer said:
No one really cares.

It could take 6000 watts to power and people would still buy it. So long as it’s the most performant.

The GPU could be the size of a full tower case needing a cooling system the size of a refrigerator and people would still buy it.

If anything they’ll label it “retro” and talk about how it looks like a mainframe. “I can’t move my PC without a forklift. How retro!”
Click to expand...
This.

I couldn't care less about how large a card is. As long as I can fit it into my case or if it has some sort of external interface that doesn't negatively impact performance, I really don't care.
 
a like the idea that it is a lot of them testing things overbuilt, moving the gpu normal window a bit with this, more than a necessity because the 4090 was probably already built for 600w continuous.

But if we are honest, wouldn't most potential buyer really like that it look like this and be a bit of the point ?

That does scream halo product even more, visible right away in their built video-picture that it is and the kind of people that considered if not went SLI quite late in the game
 
Last edited:
  • Like
Reactions: erek
like this
At this point, I'm surprised GPUs having their own cases/docks outside the case isn't more mainstream.
 
  • Like
Reactions: erek
like this
GoodBoy said:
It seems odd to change the pcb so drastically for a tier refresh GPU. And the current 4090 heatsink/pcb design is shorter than a 3090 but has 600W capability.

On the one hand, the size indicates that they are still aiming for the 600W power target, maybe more? 4 slots geez, but that would be the only way to horizontally mount the GPU+vRam.

Does this improve the cooling of the gddr6x somehow, maybe leading to significantly higher clocks on the vRam? It appears the chips still directly surround the GPU, and that there is even less space...

Maybe it's an experimental design and nothing more. Research and all that.

If the FE went this route, I don't see why AIB's would necessarily follow suit. More information would be needed as to the reasoning before we would know for sure the route the board partners could choose.
Click to expand...

it's not changing it per say.. they're likely just using the enterprise PCB. that being said if AIB's are forced to use that PCB there's probably not going to be a whole lot of them in existence since the current high end coolers are already designed to dissipate 600w, i can't see them waste money on this crap trying to re-design them.
 
sirmonkey1985 said:
it's not changing it per say.. they're likely just using the enterprise PCB.
Click to expand...
That sounds different to me:
The card's PCB isn't conventional—not perpendicular to the plane of the motherboard like any other add-in card—but is rather along the plane of the motherboard, with additional breakaway daughter cards interfacing with the sole 12VHPWR power connector, and the PCIe slot. This slender, ruler-shaped PCB spans the entire length of the card, without coming in the way of its heat dissipation surfaces.

I would imagine enterprise will tend more to go for made to have multiple of them racked next to each other design.
 
  • Like
Reactions: erek
like this
LukeTbk said:
That sounds different to me:
The card's PCB isn't conventional—not perpendicular to the plane of the motherboard like any other add-in card—but is rather along the plane of the motherboard, with additional breakaway daughter cards interfacing with the sole 12VHPWR power connector, and the PCIe slot. This slender, ruler-shaped PCB spans the entire length of the card, without coming in the way of its heat dissipation surfaces.

I would imagine enterprise will tend more to go for made to have multiple of them racked next to each other design.
Click to expand...

You can see where the PCB is supposed to go/what shape it would take (just like a ruler as they said) here:

1687803722631.png
 
HeadRusch said:
I salute all of you who are going to buy this, and hope you all have YouTube channels to expense these things around.
Click to expand...
Do you think it’ll be worth upgrading from a 4090 FE to a 4090 Ti FE?
 
You must log in or register to reply here.
Back
Top