Nvidia GeForce RTX 4080 12GB Lose More Than Memory

Lots of interest surrounding Nvidia GeForce RTX 40 Series Launched Centered around the flagship RTX 4090. Really; The high-end GPU is an absolute beast that claims to be four times faster than the previous generation RTX 3090 Ti.

The numbers look equally staggering and terrifying – 76 billion transistors on a 4nm process – yet there is more intrigue down the ladder as Nvidia surprised onlookers with not one but two GeForce RTX 4080 SKUs.

A month after it arrives in November, avid gamers will have a choice of either the GeForce RTX 4080 16GB at £1,269/$1,199, or the GeForce RTX 4080 12GB at £949/$899. On paper, the latter seems like a meaningful alternative to the RTX 3080 12GB that arrived unexpectedly earlier in the year at $799, but don’t assume that tweaking the memory configuration is all that differentiates the two next-gen cards.

The Class III RTX 40 Series is actually a one-of-a-kind offering that uses the purpose-built AD104 die. What does that mean for players who just want high tires? We’re primarily looking for cuts in key areas of both GPU and memory specifications, as detailed in the table below.

GeForce RTX 4090 RTX 4080 (16 GB) RTX 4080 (12 GB)
death M 102 M 103 M 104
SM . units 128 76 60
nuclei 16384 9728 7680
Teraflops FP32 83 49 40
RT teraflops 191 113 82
tensor teraflops 1,321 780 641
increase the clock 2.52 GHz 2.51 GHz 2.61 GHz
base clock 2.23GHz 2.21 GHz 2.31 GHz
memory configuration 24 GB GDDR6X 16 GB GDDR6X 12 GB GDDR6X
Memory bus 384 bit 256 bit 192 bit
Graphics power 450 watts 320 watts 285W
MSRP $1,599 1199 USD $899

A 25 percent drop in onboard RTX 4080 memory is reflected in most other areas. The AD104 die featured in the Nvidia 12GB card had 21% fewer active SMs, dropping the core count from 9728 to 7680. Less than half of the RTX 4090 is all singing and dancing.

Fewer SMs mean a reduction in RT and Tensor cores as well, with teraflop throughput reduced by up to 27 percent. Nvidia is raising the boost clock by a fraction of 2.51 GHz to 2.61 GHz to help mitigate the losses, but when dealing with a smaller set of 12 GB of memory and a narrower 192-bit memory bus, it becomes clear that they are two very different types of GPUs.

Big cuts, but Team Green’s interest is clearly shifting away from pure drip and toward forward-looking technologies like AI and DLSS. Using the power of the Ada Lovelace architecture, Nvidia believes the RTX 4080 12GB is capable of matching the previous generation, the RTX 3090 Ti, in most titles, and advances significantly when developers are interested in using the latest generation of DLSS 3 acceleration.

To add more intrigue, Nvidia has confirmed that the RTX 4080 12GB will be exclusive to Additional Partners and won’t be available in a Founders Edition skin. Nvidia’s internal design has been greatly enhanced for this generation, with improved 23-phase power delivery (20 for the GPU plus three for memory), a single PCIe Gen 5 power connector, and an integrated PCB shaped to maximize heat flow in the design.

Next Generation Founders Edition

Nvidia tells us that it went through more than 50 unique fan applications before settling on the final product. The larger monocoque design features 20% more airflow with a special focus on consistent power management and lower memory temperatures. Although the cooler is primarily intended for the RTX 4090, the same build features as the RTX 4080 16GB and is said to provide eerily quiet operation during regular use.

We suspect partners will be hard-pressed to offer similar engineering work on the 12GB RTX 4080, and there will be little wiggle room for premium cooling configurations. That’s above $899 and the RTX 4080 16GB seems like a much better buy.

The question arises, why not release the RTX 4080 12GB like the RTX 4070? This is a discussion that we think will continue for some time.