For gaming, the RTX 3050 is a fine gets-you-in-the-door GPU, providing good 720p and decent 1080p performance. Since much basic photo editing still isn’t very GPU-intensive, a fast, high-core-count CPU still gives you a lot more performance value for the money than a higher-powered graphics card. But the GPU does matter for the experience and smooth display rendering, but for smallish images and single-screen editing you shouldn’t have any issues. The GPU is being used increasingly to accelerate AI-based features, like automagic object selection, so it’s worth getting a more recent card — especially since many of the older GTX models have left the “older so it’s cheaper” phase and entered that “going out of stock so let’s jack up the prices because scarcity” phase.

For a little less, at $250 there’s also the better-performing Intel Arc A750. Despite Intel’s frequent updates to the drivers, though, I’m hesitant to make a blanket recommendation; Intel’s new to the discrete GPU market, and its cards seem to be experiencing first-generation growing pains. Also, you really want resizable BAR support on your system and it has higher power requirements than a typical entry-level card, which means it’s not necessarily suited to older or entry-level PC upgrades. But if you’re feeling adventurous, at least there aren’t a raft of third-party options to choose from — just ASRock and Intel’s own Limited Edition model.

The RTX 3050 card I tested was an EVGA model, and EVGA has since left the GPU biz.

This one’s a bit difficult to call at the moment, since the current generation of technologies from AMD and Nvidia for this class of card hasn’t been announced yet — though I think both are imminent. You can get last-gen cards that will perform perfectly well for these uses, and possibly older, but I hate making those recommendations without knowing what the prices will be on the newer cards, especially since the newer Nvidia Ada and AMD RDNA 3 architectures add quite a bit over their predecessors. 

Still the AMD RX 6700 XT is a good meat-and-potatoes card if you need to buy something now, pretty much outperforming the rest in its price class. The AMD cards tend to get complaints about audible whining, and I admit I hear it myself periodically, so if that’s a potential issue then the more expensive RTX 3060 Ti is probably a good bet as well.

A lot of old or low-end gaming PCs, as well as compact systems, can’t handle the size or power requirements of many of the newer GPUs. The RTX 3060 is the highest-end GPU that can fit usually into dual-slot systems which only have room for short (one- or two-fan) cards and with power supplies of 500 w or less. That doesn’t necessarily account for overclocking, so if you’re planning on pushing it to the limit, make sure the card you get — when overclocked — combined with your CPU’s power draw won’t exceed the cooling or power limits of your PC.

The RTX 4070 is one of the most flexible performers on the market, with gaming performance that spans 1080p through entry 4K — though for decent 4K performance you (and your games) need to take advantage of Nvidia’s DLSS 3 optimization technology. DLSS 3 performs much better than the previous generation, and support for it is one of the things that makes the RTX 40- series notably better than its predecessor. 

While it’s not cheap, the price is also not at the top of the range for its competitive class either.

The RX 7900 XT isn’t always faster than competitor RTX 4070 Ti, but it has more memory for the same money (20GB vs. 12GB), which gives it a leg up for games and applications that can take advantage of it. AMD’s RX 7000 series of GPUs also have DisplayPort 2.1 connections, while Nvidia’s are still using DP 1.4. While there aren’t a lot of DP 2.x monitors yet, the standard supports higher refresh rates at 4K than earlier specifications.

The RTX 4070 Ti isn’t excessively more expensive than the 4070 and it provides noticeably better performance. But it’s a lot bigger and draws a lot more power as well — unsurprising, since it began life as a stripped-down RTX 4080 — which makes it a little less flexible than the RTX 4070 as a card that I’d consider a general recommendation. But it’s also a lot smaller, cheaper and less power hungry than the RTX 4080 and 4090. That combination makes it my favorite overall, even though the 4070 is my top recommendation.

For just about everything, the RTX 4090 is the fastest consumer GPU on the market, with a big enough 24GB frame buffer to handle high-res games and video editing as well as streaming.

The RTX 4080 is a cheaper alternative in the $1,200-$1,600 range, though it’s roughly the second fastest currently on the market and it’s tempting to drop to the much less expensive 4070 Ti. And the Radeon RX 7900 XTX is a better choice than the 4080 for compute-intensive pro graphics and can outpace the RTX 4080 on some games, plus it supports higher-bandwidth DisplayPort 2.1 — though it still lags a lot of Nvidia cards for ray tracing. It can be found for as low as $1,000, though, which may be worth the tradeoff.

If you’re sensitive to screen artifacts caused by a disconnect between the rate at which your monitor updates and the frame rate at which you’re playing, or you’re interested in proprietary technologies like Nvidia’s Latency Analyzer to help improve your gameplay by reducing lag, then you should definitely at least look into what each of them offers. Otherwise, get the appropriate GPU for your needs and work with what you get.