NVIDIA teases AD102 GPU performance and Ada Lovelace architecture name
NVIDIA continues to tease the launch of its next-generation graphics cards on social media, having teased a September 20 arrival earlier this week. Subsequently, the company appears to have confirmed the name of the architecture powering what are thought to be RTX 40 series graphics cards, as well as what VideoCardz believes are three specifications for a specific SKU.
As the screenshot below shows, NVIDIA briefly included a representation of Diagram for the computation by the Engine of the Numbers of Bernoulli, a chart that Ada Lovelace produced in 1843 for Charles Babbage's Analytical Engine. In other words, we suspect NVIDIA is teasing that its next-generation graphics cards will be based on its Lovelace architecture, confirming a long-running rumour. Currently, the Lovelace architecture is expected to power all RTX 40 series graphics cards, including those intended for laptops. Additionally, NVIDIA included a sticky note containing the numbers '(208) 629 - 7538' across three lines.
According to VideoCardz, all lines may refer to the AD102, likely the flagship GPU powering the RTX 40 series generation and the successor to the GA102. The website adds that '(208)' could be a nod to the AD102 offering 2.08x the performance of the GA102, with '629' being the GPU's die size of 629 mm², which would only be 1 mm² larger than the GA102. However, the AD102 should have a considerably denser die than the GA102. While the GA102 contains 28.3 billion transistors with 45.1 million transistors per 1 mm², VideoCardz asserts that '7538' is code for 75.38 billion transistors, which should equate to 119.8 million transistors. The RTX 40 series is rumoured to start with the RTX 4090 and two versions of the RTX 4080, although only the RTX 4080 may be available at first.
Source(s)
@NvidiaANZ via VideoCardz