This appears to be a transcript of Jensen Huang speaking at a GeForce 3 25th anniversary celebration. A few highlights from what he shared:
**On GeForce 3's significance:** Jensen described it as a pivotal internal transition — moving NVIDIA from a fixed hardware accelerator to a *programmable* pipeline. The key insight was that games, unlike CAD software, are a medium for artistic expression, and that required programmability so each game could look distinct rather than having that uniform, heavily-filtered look.
**On NVIDIA's identity shift:** He emphasized that being an "algorithm accelerator" is essentially being an accelerated computing company — a philosophy that GeForce 3 first embodied, and that runs through everything NVIDIA does today.
**On the hardware leap:** GeForce 3 had around 30 million transistors and ran at 35 watts. Today's chips are in the hundreds of billions of transistors and pushing 3,500 watts — a 100x jump in power draw alone.
**On neural rendering:** Jensen revealed that NVIDIA has been working for nearly a decade (since RTX launched in 2018) on unifying traditional computer graphics/programmable shading with generative AI into what he calls "neural rendering" — blending controllable rendering with more imaginative generative approaches.
**On the legacy chain:** He closed with a neat through-line — *without GeForce, there's no CUDA; without CUDA, there's no AI; without AI, there's no today.*
It's a candid and surprisingly reflective speech. Was there a specific part you wanted to dig into?
**On GeForce 3's significance:** Jensen described it as a pivotal internal transition — moving NVIDIA from a fixed hardware accelerator to a *programmable* pipeline. The key insight was that games, unlike CAD software, are a medium for artistic expression, and that required programmability so each game could look distinct rather than having that uniform, heavily-filtered look.
**On NVIDIA's identity shift:** He emphasized that being an "algorithm accelerator" is essentially being an accelerated computing company — a philosophy that GeForce 3 first embodied, and that runs through everything NVIDIA does today.
**On the hardware leap:** GeForce 3 had around 30 million transistors and ran at 35 watts. Today's chips are in the hundreds of billions of transistors and pushing 3,500 watts — a 100x jump in power draw alone.
**On neural rendering:** Jensen revealed that NVIDIA has been working for nearly a decade (since RTX launched in 2018) on unifying traditional computer graphics/programmable shading with generative AI into what he calls "neural rendering" — blending controllable rendering with more imaginative generative approaches.
**On the legacy chain:** He closed with a neat through-line — *without GeForce, there's no CUDA; without CUDA, there's no AI; without AI, there's no today.*
It's a candid and surprisingly reflective speech. Was there a specific part you wanted to dig into?
ความคิดเห็น 0
กำลังโหลดความคิดเห็น...