That proof can actually be seen when the video zooms at the 4-second mark, where you can see a technical document with the title “Diagram for the computation by the Engine of the Number of Bernoulli” at the top of it. For all the non-engineering geeks out there, that’s a chart that was created by the English mathematician, Ada Lovelace.
— NVIDIA ANZ (@NvidiaANZ) September 14, 2022 There’s another hidden message inside the Lovelace teaser as well, although it actually took some sleuthing by some eagle-eyed viewers to find. Right before it cuts out and zooms into the chart, there is a sticky note that can be briefly seen, stuck at the bottom of the left display. That sticky note actually contains the numbers 208, 629, and 7538. While nothing is confirmed, the first number is speculated to be a reference to the AD102 GPU, which again is based on the Lovelace architecture and more precisely, its alleged performance improvement over last generation’s flagship GPU, the GA102. 2.08 times more, to be precise. The second number could be a hint towards the size of the GPU die; while there are no specific rumours about this, it is possible that the Lovelace GPU could measure in at 629mm², which just 1mm² larger than the GA102. So, more or less the same footprint, but just a lot more performance. The last number in the Lovelace teaser is believed to be the 75.38 billion transistors that the GPU will house. For context Ampere’s GA102 GPU had a maximum transistor capacity of 28.3 billion, so if that really is the number of transistors inside the GPU, then how NVIDIA managed to cram that many more transistors into the die is nothing short of impressive. NVIDIA is expected to announce its new Ada Lovelace architecture and GeForce RTX 40 Series next week, during its official GeForce Beyond broadcast. Additionally, the GPU maker should be announcing three GeForce GPU models: the RTX 4090, RTX 4080 16GB, and the RTX 4080 12GB. However, rumours have pointed out that only the first card will see a launch in October this year, while the other two will only be made available sometime at the beginning of next year. (Source: Videocardz, NVIDIA via Twitter)
