Skip to main content

This trick improves the power efficiency of Nvidia’s RTX 4090

Nvidia’s latest flagship, the RTX 4090, has high power requirements, which is not surprising given its 450-watt total board power (TBP). However, there are ways to make the GPU much more power efficient.

To put that claim to the test, an owner of an RTX 4090 decided to try out power limiting and undervolting. The results were better than expected.

Nvidia's RTX 4090 undervolting chart.
QuasarZone

Before the RTX 4090 launched in October, it was the subject of a lot of speculation when it came to its power efficiency. Some sources warned about obscenely high TBP figures, in the range of 600W to 900W. Fortunately, that wasn’t the case, but the GPU is still quite power-hungry at 450W.

Recommended Videos

Even at 450W, the card requires a strong power supply and it eats up a whole lot of power. Still, it’s not as bad as expected — the RTX 4090 hardly ever utilizes all of that power. It’s currently too good for most games, even demanding AAA titles, so that helps.

Get your weekly teardown of the tech behind PC gaming
Check your inbox!

However, it’s no secret that undervolting a GPU can limit power usage while maintaining similar performance. QuasarZone tried out two ways of improving the RTX 4090’s power efficiency — adjusting the power limit and undervolting. They then put the card to the test in five games, all played at max settings and at 4K resolution. The list of titles includes Cyberpunk 2077, Marvel’s Spider-Man Remastered, Forza Horizon 5, Lost Ark, and PlayerUnknown’s Battlegrounds.

RTX 4090 performance at different power levels.
QuasarZone

By default, the card ran at 347W and a clock speed of 2,726MHz. Limiting the power to 60% lowered power consumption down to 268W while maintaining a clock speed of 2,443MHz. A 70% power limit seems like a sweet spot, with 305W and 2,602MHz, which is a negligible drop compared to running at 100%.

Undervolting has proved to be better than tweaking the power limit, but it’s also trickier to do. At 323 watts, QuasarZone managed to bring up the clock speed to 2,850MHz. On the lower end of the scale, they achieved 232W combined with 2,460MHz.

Regardless of the method, RTX 4090’s power efficiency was improved without hurting the card’s performance. If you own one of Nvidia’s best GPUs, this could certainly be something worth trying, especially if you worry about being hit by the melting connector fiasco.

Monica J. White
Monica is a computing writer at Digital Trends, focusing on PC hardware. Since joining the team in 2021, Monica has written…
Nvidia may have found a new way to bypass GPU export restrictions
The RTX 4090 graphics card sitting on a table with a dark green background.

Chinese gamers are expected to have limited access to Nvidia's best graphics card due to strict export restrictions. However, Nvidia may have found a way around it without cutting down its flagship GPU. A new leak suggests that the RTX 5090D will have the same hardware specifications as the worldwide version, and the solution lies in firmware adjustments.

As a quick refresher, the "D" in RTX 5090D stands for "Dragon." It marks GPUs made by Nvidia to bypass the export restrictions imposed by the U.S. on China, which limit the sale of high-performance graphics processors. We first saw it appear in the RTX 4090D as a remedy for the fact that the base RTX 4090 exceeds the performance thresholds set by these regulations.

Read more
Nvidia CEO in 1997: ‘We need to kill Intel’
NVIDIA CEO Jensen Huang at GTC

Those headline above includes strong words from the maker of the best graphics cards you can buy, and they have extra significance considering where Nvidia sits today in relation to Intel. But in 1997, things were a bit different. The quote comes from the upcoming book The Nvidia Way, written by columnist Tae Kim, and was shared as part of an excerpt ahead of the book's release next month.

The words from Nvidia CEO Jensen Huang came as part of an all-hands meeting at the company in 1997 following the launch of the RIVA 128. This was prior to the release of the GeForce 256, when Nvidia finally coined the term "GPU," and it was a precarious time for the new company. Shortly following the release of the RIVA 128, Intel launched its own i740, which came with an 8MB frame buffer. The RIVA 128 came with only a 4MB frame buffer.

Read more
Nvidia’s next-gen GPU plans could be good news for Intel and AMD
Two RTX 4070 Ti Super graphics cards sitting next to each other.

According to a new leak from Benchlife, Nvidia may launch the vast majority of the RTX 50-series in the first quarter of 2025 -- but one GPU is notably missing from the early lineup. That could be very good news for AMD and Intel. While Nvidia will rule the high-end market, the other two brands may get to swoop in with some of the best graphics cards for gamers on a budget and get some breathing room before Nvidia strikes back.

Benchlife reveals that we'll see many of the RTX 50-series staples arrive in the first quarter of the year. The flagship RTX 5090 and the RTX 5080 arriving in January feel like a sure thing at this point, but many leakers also suggest that we'll see other GPUs make their debut during CES 2025.

Read more