On October 18, 2013, in Montreal, NVIDIA announced a fix for a problem PC players had lived with for as long as they'd had graphics cards. The company's press release promised that tearing, stutter and lag would "become artifacts of the past." The technology was called G-Sync. By NVIDIA's own later account, gaming monitors "evolved rapidly" after its launch, and the rival approach from AMD turned variable refresh into a fight between a proprietary product and an open standard.
The problem nobody could switch off
The trouble came from two clocks that didn't agree. As NVIDIA's announcement explained, a standard monitor refreshed at a fixed rate, typically 60Hz, while a graphics card produced frames at whatever pace the game allowed, faster in an empty corridor and slower in a firefight.
Players had two bad options. Leave the two out of step and the monitor would start drawing a new frame partway through the old one, slicing the image with a visible tear. Turn on V-Sync to make the card wait for the monitor, and, in NVIDIA's words, it "eliminates tearing but causes increased lag and stutter," because the card is held back to match a refresh rate it can't always hit.
PC Gamer summed up the situation in 2014: if the frame rate doesn't match the refresh rate, "all kinds of visual artifacts will follow."
NVIDIA's answer: make the monitor wait
G-Sync flipped the relationship. Instead of the graphics card waiting on the monitor, the monitor would wait on the graphics card, refreshing whenever a new frame was ready. NVIDIA said this synchronized the display to the GPU's render rate so each image appeared as soon as it was rendered.
The way NVIDIA did it mattered as much as the idea. G-Sync relied on an NVIDIA-designed module built into the monitor, working with supporting hardware in certain Kepler-based GeForce cards. ASUS, BenQ, Philips and ViewSonic were among the first monitor makers lined up, and the launch press release carried endorsements from Epic's Tim Sweeney, DICE's Johan Andersson and id Software co-founder John Carmack.
Tom's Hardware later explained what that module involved: NVIDIA had designed a custom display scaler built around an expensive component called an FPGA. That gave NVIDIA tight control over how G-Sync behaved, but Tom's Hardware reported in 2015 that it also left G-Sync displays costing $150 to $200 more than their AMD equivalents.
AMD's answer: make it a standard
AMD's response, FreeSync, took the opposite route. Rather than a proprietary module, it leaned on an industry standard. In May 2014, as PC Gamer reported at the time, the Video Electronics Standards Association added Adaptive-Sync to version 1.2a of the DisplayPort specification, letting a DisplayPort monitor match a graphics card's frame rate on the fly.
That was no coincidence. AMD told PC Gamer that "the DisplayPort Adaptive-Sync specification was ported from the Embedded DisplayPort specification through a proposal to the VESA group by AMD." Its pitch was blunt: "G-Sync uses expensive and proprietary hardware," while FreeSync would use the industry-standard specification "to promote wider adoption lower cost of ownership and a broad ecosystem of compatibility," as AMD's statement read. The catch, PC Gamer noted, was time. AMD expected compatible monitors to be six to twelve months away, while G-Sync screens were already available.
Tom's Hardware later spelled out the distinction that confuses buyers to this day. Adaptive-Sync is, strictly, a DisplayPort protocol addition; FreeSync is AMD's implementation across the whole chain of graphics card, connection and display. Instead of building its own scaler, AMD signed up scaler manufacturers to support it. That kept costs down, but Tom's Hardware reported it also gave AMD little control over how monitor makers implemented the feature, and early displays had quality problems such as flicker, which AMD answered with its own eligibility process for the FreeSync logo.
The turning point: NVIDIA opens the door
For GeForce owners, the dividing line moved at CES in January 2019. In a GeForce post, NVIDIA said GeForce players had been asking for a "trimmed down" variable refresh experience, and introduced "G-SYNC Compatible": NVIDIA would test VESA Adaptive-Sync monitors and switch on variable refresh automatically for those that passed on GeForce GTX 10-series, GTX 16-series and RTX 20-series cards. Its tests looked for blanking, pulsing, flickering and ghosting, and required a variable refresh range of at least 2.4:1, such as 60Hz to 144Hz.
The bar was high. NVIDIA said 12 monitors were validated at launch in January 2019, and by March only five more had been added, for a total of 17, after it had tested hundreds. Owners of everything else got a control panel option to try it anyway, with NVIDIA's own warning that "it may work, it may work partly, or it may not work at all." NVIDIA kept recommending its own G-Sync and G-Sync Ultimate monitors, with dedicated processors, for the best experience.
Where it landed
After 2019, a GeForce card could run variable refresh on certified Adaptive-Sync monitors as well as on NVIDIA's own G-Sync displays, and the standard itself kept developing. In 2022, VESA launched its own Adaptive-Sync Display certification, an open logo program with more than 50 automated tests, which Tom's Hardware suggested could eventually supplant the G-Sync and FreeSync labels on monitors.
The sequence, then: NVIDIA first, with a proprietary module and a price premium; AMD next, with a standard built on its own proposal and weaker control over how monitor makers used it; and NVIDIA, in 2019, validating Adaptive-Sync monitors for its own cards.
Mason's take: This reads to me like a familiar platform story. The company that gets there first builds a closed version and charges for it; the challenger wins on price by turning the idea into a standard; and once the standard is everywhere, the leader finds a way to live inside it. Whatever the motives, players got the best of it: with a variable refresh monitor and a supported card, the old trade between tearing and V-Sync stutter is one they no longer have to make.
Reporting notes