Full specifications for the NVIDIA AD102, AD103 and AD104 GPUs
September 23, 2022
0
The green giant has confirmed the full specifications of its three next-generation graphics cores, the NVIDIA AD102, the high-end chip that will be used in the GeForce RTX
The green giant has confirmed the full specifications of its three next-generation graphics cores, the NVIDIA AD102, the high-end chip that will be used in the GeForce RTX 4090; and AD103, which will be used in the GeForce RTX 4080 16 GB; and AD104, which will be used in the GeForce RTX 4080 12 GB. Please note that these GPUs may also be used in other graphics cards coming later, such as the hypothetical (but likely) GeForce RTX 4090 Ti, GeForce RTX 4080 Ti, and GeForce RTX 4070.
With the Ada Lovelace architecture, NVIDIA introduced important changes at the technological level that made the GeForce RTX 40 the most advanced graphics cards that exist today, but we also find deep modifications in each of these graphics cores that go beyond switching processes and increasing the number of shaders. NVIDIA has significantly increased the number of transistors and made it clear as well The TSMC 4N node is not 4nm, but actually 5nm. We have updated all our recently published articles to reflect this.
The jump to the 5nm node and these changes at the architecture level enabled the green giant increase the maximum number of shaders to nothing more and nothing less than 18,432 units in the NVIDIA AD102 core. In AD103 we have a maximum number of 10,240 shaders and in AD104 we find 7,680 shaders. These numbers allow us to confirm that only the GeForce RTX 4080 12 GB uses the full graphics core, and therefore the NVIDIA AD102 and AD103 can be used to create versions of Ti from the presented models, i.e. GeForce RTX 4090 Ti and GeForce RTX 4080 Ti.
NVIDIA AD102 Graphics Core Specifications
Ada Lovelace architecture, manufactured at TSMC’s 5nm node.
76.3 billion transistors.
Transistor density: 125.5 million.
144 SM units.
18,432 shaders.
576 texturing units.
576 tensor kernels.
144 RT cores.
192 raster units.
96 MB of L2 cache.
NVIDIA AD103 Graphics Core Specifications
Ada Lovelace architecture, manufactured at TSMC’s 5nm node.
45.9 billion transistors.
Transistor density: 121.1 million.
80 SM units.
10,240 shaders.
320 texturing units.
320 tensor kernels.
80 RT cores.
112 raster units.
64 MB of L2 cache.
NVIDIA AD104 Graphics Core Specifications
Ada Lovelace architecture, manufactured at TSMC’s 5nm node.
35.8 billion transistors.
Transistor density: 121.1 million.
60 SM units.
7,680 shaders.
240 texturing units.
240 tensor cores.
60 RT cores.
80 raster units.
48 MB of L2 cache.
We can see that in addition to an increase in transistor density, an increase in shaders, and a change in the production process, Ada Lovelace has seen a significant increase in the total number of texturing units and the number of tensors. and RT cores, something completely understandable since we have more SM units per GPU. The number of raster units has also increased considerably compared to the Ampere, as the NVIDIA GA102 and GA103 cores had 112 and 96 raster units, respectively.
The loading L2 cache It is another of the most important changes that NVIDIA has presented in Ada Lovelace, although we must not forget the other details that we have already told you recently when breaking down the keys to the GeForce RTX 40, such as the introduction of the fourth tensor cores and the 3rd generation RT core.
With all this information in hand we can get an idea of what awaits us in the NVIDIA catalog. The GeForce RTX 4090 Ti should be based on a full NVIDIA AD102 core, and the GeForce RTX 4080 Ti will almost certainly use a full AD103 chip. Finally, the AD104 will be used in a cut version to shape the GeForce RTX 4070.
Alice Smith is a seasoned journalist and writer for Div Bracket. She has a keen sense of what’s important and is always on top of the latest trends. Alice provides in-depth coverage of the most talked-about news stories, delivering insightful and thought-provoking articles that keep her readers informed and engaged.