Throughout this deep dive, we’ll refer to prior-generation hardware from Nvidia, and it’s important to understand what we’re talking about. However, we’re only going back as far as the first RTX cards, which launched in 2018—we’ll leave out the earlier GTX and other generations. Here’s the short rundown, from oldest to newest.
TURING (2018): The RTX 20-series arrived in 2018 and marked a big change in Nvidia’s graphics hardware. It came with new branding (RTX instead of GTX), and included two major new hardware innovations: hardware ray tracing functionality, and AI-focused tensor cores. Each Streaming Multiprocessor (SM) contained 64 CUDA cores capable of two 32-bit FMA (fused multiply accumulate) operations per cycle, with eight tensor cores capable of 128 16-bit floating point (FP16) FMA operations per cycle. It was manufactured…
You're reading a preview of
Maximum PC (Digital) - 1 Issue, April 2025