Nvidia’s much-anticipated next-generation flagship graphics card, the RTX 5090, is said to use TSMC’s cutting-edge 3nm manufacturing process and will be launched by the end of next year, according to a report from Hardwaretimes.
Nvidia’s last top-of-the-range graphics card, the RTX 40, was nicknamed Ada Lovelace. However, according to Hardwaretimes, the upcoming RTX series is codenamed Blackwell. These new GPUs will be built on TSMC’s groundbreaking 3nm (N3) node technology, which has over 150 billion transistors and a density of almost 300 million transistors per square millimeter. In addition, the core clock speed is expected to exceed 3GHz with a bus density of 512 bits.
Also read: Project Helix: The New AI Solution for Enterprises by Nvidia and Dell
According to leaked information from international media outlet Club 386, the RTX 5090 will have 144 sets of SM units, resulting in 18,432 CUDA cores (assuming each set of SM still holds 128 CUDA cores). This is a 12.5% improvement over the RTX 4090. It should also have 96MB of L2 cache and support GDDR7 memory with a 384-bit bus width. The card is also expected to be PCIe 5.0 x16 compliant.
MSI demonstrated its revolutionary heat dissipation architecture for the upcoming Nvidia RTX flagship graphics card at the recent Computex convention in Taipei. The design uses dynamic bimetallic fins, six high-performance copper heat pipes and large-area aluminum fins with integrated copper sheets to improve heat dissipation. The memory portion of the card also features special copper plates to further improve cooling performance.
As the release date of the next-generation Nvidia RTX 5090 approaches, fans and gamers alike are looking forward to the performance and technical breakthroughs that this graphics card will bring to the market. The RTX 5090 has the ability to rewrite the standards for high-end gaming and professional graphics applications thanks to TSMC’s 3nm process and unique cooling technologies.