Alleged pictures of NVIDIA's GeForce RTX 4090 Ti graphics card featuring the full-fat AD102 GPU have been leaked by Greymon55.
NVIDIA's Full-Fat GA102 GPU Powered GeForce RTX 4090 Ti Graphics Card Pictured, Massive Triple-Fan Cooler For 600W+ Powerhouse
The card pictured is alleged to feature NVIDIA's full-fat AD102 GPU based on the Ada Lovelace chip architecture. The full-fat chip is presumably going to feature a TBP of 600W (+) and as such, it will require some beefy cooling solutions such as the one pictured here.
As per the leaker, this seems to be an early engineering sample that is used to test the full limit of the monstrous AD102 GPU which will land in the consumer segment in a couple of months. The final variant for the Founders Edition design is expected to use a vastly different cooler but this seems to confirm that AD102 is going to be one hard chip for AIBs to cool down.
Most likely engineering sample, the reference version should not be designed so simple.
— Greymon55 (@greymon55) September 5, 2022
This isn't the first time that we have heard of a triple-fan cooler for the reference AD102 board. A while back, Kopite7kimi had pointed out the same that a triple-fan and a 3-slot cooler were made purely for AD102 GPUs and this leak seems to be a confirmation of that.
At least, they designed a triple-fan cooler for reference board of AD102. https://t.co/vQb2KnpE4y
— kopite7kimi (@kopite7kimi) June 9, 2022
As for the card itself, it appears to be very thick, both in terms of sheer size and width. The triple-fan solution is based on an axial-blade design and there's a large heatsink underneath the shroud. The shroud itself is very simple with a matte black texture and it looks like the card comes with a 3-slot form factor. NVIDIA has been relying on a 16-pin connector so if this is indeed a 600W variant, then a single Gen5 connector should be enough to boot it up unless the ES design has two of those connectors.
NVIDIA GeForce RTX 4090 Ti 'Expected' Specifications
The NVIDIA GeForce RTX 4090 Ti is going to be the full-fat configuration with all of the 144 SMs enabled for a total of 18432 CUDA cores. The GPU will come packed with 96 MB of L2 cache and a total of 384 ROPs which is simply insane. The clock speeds are not confirmed yet but considering that the TSMC 4N process is being used, we are expecting clocks between the 2.0-3.0 GHz range.
As for memory specs, the GeForce RTX 4090 Ti is expected to rock 24 GB GDDR6X capacities that might come at faster 24 Gbps speeds across a 384-bit bus interface. This will provide up to 1.152 TB/s of bandwidth. Now all these boosted specifications will result in higher power draw too and the flagship is expected to operate at a TBP of around 600W. Now for 600W, a single 16-pin Gen 5 connector should be enough but most of the custom variants will definitely end up utilizing dual Gen 5 connectors since AIBs don't necessarily stay within spec and even the slightest of factory overclocks will push the TBP above 600W which is the limit of a single Gen 5 power connector.
We have also seen an alleged NVIDIA GeForce RTX 4090 Ti heatsink and cooler shroud which hints at the use of a beefier cold plate that provides coverage for both the GPU and memory dies along with an overall larger structure. The previously leaked cooler seems more reminiscent of a final FE or Founders Edition design than the new leak which is more of an in-house testing design & judging by how big it looks, the AIB models will end up being vastly bigger and we may even end up with quad-slot designs from all partners.
NVIDIA GeForce RTX 4090 Ti & RTX 4090 'Preliminary' Specs:
|Graphics Card Name||NVIDIA GeForce RTX 4090 Ti||NVIDIA GeForce RTX 4090||NVIDIA GeForce RTX 3090 Ti||NVIDIA GeForce RTX 3090|
|GPU Name||Ada Lovelace AD102-350?||Ada Lovelace AD102-300?||Ampere GA102-350||Ampere GA102-300|
|Process Node||TSMC 4N||TSMC 4N||Samsung 8nm||Samsung 8nm|
|Transistors||TBD||TBD||28 Billion||28 Billion|
|TMUs / ROPs||TBD / 384||TBD / 384||336 / 112||328 / 112|
|Tensor / RT Cores||TBD / TBD||TBD / TBD||336 / 84||328 / 82|
|Base Clock||TBD||TBD||1560 MHz||1400 MHz|
|Boost Clock||~2800 MHz||~2600 MHz||1860 MHz||1700 MHz|
|FP32 Compute||~103 TFLOPs||~90 TFLOPs||40 TFLOPs||36 TFLOPs|
|RT TFLOPs||TBD||TBD||74 TFLOPs||69 TFLOPs|
|Tensor-TOPs||TBD||TBD||320 TOPs||285 TOPs|
|Memory Capacity||24 GB GDDR6X||24 GB GDDR6X||24 GB GDDR6X||24 GB GDDR6X|
|Memory Speed||24.0 Gbps||21.0 Gbps||21.0 Gbps||19.5 Gbps|
|Bandwidth||1152 GB/s||1008 GB/s||1008 GB/s||936 Gbps|
|Price (MSRP / FE)||$1999 US?||$1499 US?||$1999 US||$1499 US|
|Launch (Availability)||2023?||October 2022?||29th March 2022||24th September 2020|