The RTX 4-series from Nvidia, which includes the $1,599 flagship RTX 4090 and the slightly more inexpensive 4080, is the company's newest line of graphics cards. PC users who have been deprived of the GPU market by cryptocurrency miners for years now have an interesting choice between their long-awaited older cards and the sparkling (and pricey) new ones thanks to the upgrade.

The 4-series is (as you might guess) a major improvement over the 3-series, which dominated the crypto mining sector and a few other computation-intensive applications and was consequently frequently purchased in large quantities by bots and other dubious enterprises. This had the consequence of effectively eliminating availability and driving up the cost of the potent cards.

It was a chance that Nvidia's steadfast rival AMD seized, highlighting pricing, efficiency (more power for your dollar), and the availability of their GPUs. Over the past two years, perhaps more individuals have converted to AMD out of anger with the market than out of a desire to save a little power and money. (I'll admit it; I was one of them.)

However, Nvidia is back and going after the big fish with its next-gen GPUs, which are currently without a doubt by far the most potent ones available. Although AMD will soon make its next-generation debut, Nvidia often retains the title of "fastest card, but at what cost?"

Nvidia is promoting its new "Ada Lovelace" architecture in addition to boosting the amount of cores, processors, shaders, and the necessary boost in speed observed in today's games. The ray tracing for lighting and Deep Learning Super Sampling, a complex frame generation method that boosts frame rates and visual fidelity, are two examples of specific operations that are accelerated as a result and enhance image quality.

They make extensive boasts about their powers here. The legendary mind-bending first-person puzzler Portal, which will receive ray tracing and DLSS in a new edition free to everyone who already owns it, was "reimagined" by Nvidia and Valve as a demonstration of some of the techniques (which is just about everyone by now). Looks really good:

Additionally, it has tensor cores designed specifically for AI-related tasks, which should enable the new GPUs to handle pure neural network tasks (like Stable Diffusion) and AI-powered graphics operations both more quickly and effectively.

The GPU has evolved from a device for maintaining high frame rates in computer games to a processing powerhouse for well-known, computation-intensive applications that consume CPUs and older GPUs.

For a variety of reasons, operating a contemporary text-to-image generator or a local instance of something like GPT-3 is appealing to many, and many of them will fork up a grand for the opportunity. Beginning in November, the cards will be accessible.