Gamers on a strict budget can finally get a taste of the improvements found in Nvidia’s Turing GPU architecture. On Tuesday, Nvidia launched the GeForce GTX 1650, a $150 graphics card built to slip easily into pre-existing PCs and start playing 1080p games with minimal fuss. It joins the more powerful GTX 1660 and GTX 1660 Ti in the current GTX 16-series lineup.
Like those cards, the GeForce GTX 1650 will be available only in custom designs by Nvidia’s board partners—there won’t be a Founders Edition version. But unlike its cousins, the GeForce GTX 1650 will be available in some boards that consume a mere 75 watts of power, avoiding the need for a supplemental power connector from your power supply. That unique capability helped make its predecessors, the GTX 1050 and GTX 1050 Ti, such popular gaming options. Graphics cards that don’t need extra power connectors can turn even cheap prebuilt PCs into proper gaming rigs.
Nvidia’s radical new Turing architecture helps the GeForce GTX 1650 deliver plenty of game, though this particular GPU lacks the real-time ray tracing and machine learning capabilities of the far pricier RTX 20-series hardware. The GTX 1650 “is 2x faster than the GTX 950 and up to 70 percent faster than the GTX 1050 at 1080p resolution,” Nvidia says.
Working drivers weren’t provided to the press prior to the launch, so we weren’t able to verify the claim independently. We’d expect the best results to come in more modern titles that can leverage Turing’s newly separated integer and float point pipelines. Games that respond well to asynchronous compute capabilities, such as the ever-popular Rainbow Six Siege, play especially nice on Turing compared to Nvidia’s previous Pascal-based GPUs.
GeForce GTX 1650 specs and feature analysis
Here’s a high-level look at the GeForce GTX 1650’s tech specs:
CUDA cores: 896
Base clock: 1,485MHz
Boost clock: 1,665MHz
Memory: 4GB of GDDR5 memory with a 128-bit memory bus
Memory bandwidth: 128GB/sec.
The GeForce GTX 1650 comes with 256 more CUDA cores than the GTX 1050 and 128 more CUDA cores than the GTX 1050 Ti. That alone would give the newer card a performance bump. Factor in the more potent Turing architecture in the new GPU, along with significantly higher boost clock speeds, and the GTX 1650 should pummel its predecessors despite sticking to the same 75W TDP.
Nvidia’s latest card also ups the VRAM capacity from the GTX 1050’s scant 2GB. Instead it offers the 4GB that’s become the de facto minimum for decent 1080p gaming these days.
This should be a huge upgrade, in other words—though as with the other graphics cards in Nvidia’s current lineup, it comes at a price. While the GTX 1050 launched at $110, and the speedier GTX 1050 Ti cost $140 before 2018’s nasty cryptocurrency-fueled pricing crunch, the GeForce GTX 1650 starts at $150.
Note, however, that AMD’s graphics card requires an 8-pin power connector. If the GTX 1650 fails to match its rival—again, Nvidia didn’t provide reviewers with drivers to test these cards prior to release—than it’s clear that the company prioritized the plug-and-play nature of a 75W card over beating the Radeon RX 570’s performance. Stay tuned for reviews.
Speaking of design decisions, there’s a sizeable price gap between the $149 GeForce GTX 1650 and the $220 GTX 1660. There’s a big gap in CUDA core counts, too, with a mere 896 in this new GPU compared to a whopping 1,408 in the GTX 1660. Nvidia didn’t announce a GeForce GTX 1650 Ti today, but rumors of one have been swirling. We’d be shocked if one weren’t announced in the future to fill in those gaps.
Look for a wide variety of custom GeForce GTX 1650 graphics cards from the usual manufacturers, on store shelves today in the usual places. If past releases are any indication, some overclocked variants may require an extra 6-pin power connector. Be sure to check the fine print before you buy if you need a model that can be powered by your motherboard alone.
Note: When you purchase something after clicking links in our articles, we may earn a small commission. Read ouraffiliate link policyfor more details.