Shulou(Shulou.com)11/24 Report--

IT House June 2 news, according to foreign media Hardwaretimes reported that the next generation of Nvidia flagship graphics card RTX 5090 will use TSMC's 3nm process, is expected to be launched by the end of next year.

Tu source Pexels Nvidia launched last year's RTX 40 series graphics card code name Ada Lovelace, while foreign media Hardwaretimes said that the next generation of Nvidia RTX graphics card code name Blackwell, and that these GPU will be manufactured in TSMC's 3nm (N3) node, the number of transistors will exceed 15 billion, the density will be close to 300 million / mm ², the core clock will exceed 3 Ghz, and the bus density will reach 512 bits.

According to earlier leaked information from foreign media Club 5090, GB102-based RTX 5090 contains 144sets of SM units, that is, 18432 CUDA (assuming each group of SM is still 128MB), which is 12.5% more than RTX 4090 with 96MB second-level cache, matches GDDR7 video memory (384bit bit width), and supports PCIe 5.0x16.

Recently, at the Computex computer Show in Taipei, MSI also showed the heat dissipation design of the next generation of Nvidia's flagship RTX graphics card. According to reports, microstars use dynamic bimetallic fins (Dynamic Bimetallic Fin), six penetrating pure copper heat pipes and large area aluminum fins are also embedded in copper sheets to further enhance heat dissipation, while there are corresponding copper sheets in the apparent memory area.

▲ image source Club 386

▲ image source Club 386CTOnews.com believes that combined with the current exposure of information, combined with such heat dissipation conditions, the original power consumption and heating of RTX 5090 should be impressive.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.