GPU companies want you to buy more, and artificial segmentation is a neat way to force you to do just that. To me, it feels obvious that a higher-end SKU deserves to have more VRAM than lower-end counterparts, but Nvidia doesn't agree. While its RTX 5060 Ti has a 16GB variant, the 30% faster RTX 5070 is left to struggle with just 12GB of VRAM. Sure, the RTX 5070 Ti has 16GB of VRAM, but so does the top-tier RTX 5080, as inexplicable as that is. Even the last-gen RTX 4080 and RTX 4080 Super have 16GB of memory; why does the successor not have more? The simple answer is that Nvidia is the market
5 PC hardware downgrades the industry disguised as "progress"… Not quite — the answer is superior power efficiency and performance-per-watt. Apple Silicon's efficiency is one of its most celebrated traits, and Intel Firefly's appeal is rooted in making those benefits available outside the Apple ecosystem. …
… Related 6 GPUs that are officially too old in 2025 These GPUs can no longer bring it in modern titles AI is being used for the Linux kernel Vibe-assisted coding isn't anything new Linux kernel contributions can be AI-assisted, though official guidelines are in place to ensure quality remains high. …
… The differences matter, especially in front-end-sensitive workloads like games, but they're not large enough to explain the whole Apple Silicon efficiency story by themselves. …
… They are calling AI-generated frames the future of gaming instead of admitting that hardware-powered gains have hit a wall. …
… There's only so much you can extract from it before efficiency collapses. For years, the answer had been simple enough. Just shrink more nodes, add more cores, push higher clocks, and increase memory bandwidth. Now, however, every incremental gain comes at a disproportionately higher cost. …
… "When you want performance you go to the GPU, but when you want efficiency you go to the NPU," Tikoo said. …