NVIDIA's new RTX 40-series GPUs are insanely powerful, but also wildly expensive. That's my big takeaway after reviewing the RTX 4090 and RTX 4080 — sure, they're fast, but who can justify spending over $1,000 on a video card? With the RTX 4070 Ti, which debuted at CES 2023, NVIDIA is offering a slightly more reasonable alternative. Starting at $799, it's still fairly pricey, but at least it's under $1,000. And best of all, it's in many ways better than last year's 3090 Ti, which initially cost a whopping $2,000.
After announcing two RTX 4080 cards a few months ago, NVIDIA surprised us all when it "unlaunched" the $899 12GB model. Given its much lower specs, there were plenty of complaints that it seemed a bit too expensive to be called a 4080. So now we've got the 4070 Ti at $100 less, with the same 7,680 CUDA cores and 12GB of GDDR6X memory that the 4080 was supposed to get. Sometimes, yelling at companies online gets results.
If you've got a small case, the 4070 Ti may also be the first RTX 40-series GPU you can actually use. Both the 4080 and 4090 Ti were triple-slot behemoths — they took up a significant chunk of my fairly roomy mid-tower case – whereas the 4070 Ti just needs two. It also requires far less energy than either of those cards, since it can run with a 700-watt PSU and has a maximum power draw of 285W. (The 4080 requires a 750W PSU, while the demands an 850W unit.) NVIDIA says the 4070 Ti uses around 49 percent less power on average than the 3090 Ti.
Given where it sits alongside the RTX 4080, the 4070 Ti performed exactly as I expected. It clocked in around 20 percent slower in 3DMark's TimeSpy Extreme Benchmark, as well as the Geekbench 4 Compute test. It was also a full 30 fps slower while playing Halo
Read more on engadget.com