Enlarge / AMD’s Radeon RX 7600.
Andrew Cunningham
Earlier this month, AMD briefed the press on its first mainstream RX 7000-series card, the RX 7600. A mostly incremental upgrade over the original RX 6600 but with many of the new features from the RX 7900 XTX and XT, it would come with a price cut, from the RX 6600’s $329 to $299. Nvidia then briefed the press on its new mainstream RTX 4060 series. The prices for the higher-end 8GB and 16GB RTX 4060 Ti were already set at $399 and $499. The price for the lower-end RTX 4060 was left undisclosed.
A few days later, presumably having caught wind of AMD’s pricing plan for the RX 7600, Nvidia announced the price for the RTX 4060: also a surprisingly low $299. (This entire time, review embargoes and briefings have been shifting by a few days here and there as the companies maneuver around each other.) Then, around 36 hours before this article was published, a new update came from AMD: The RX 7600 would now be launching for $269, $30 less than the RTX 4060 and $60 less than the old RX 6600.
This is what competition in the midrange GPU market looks like after a years-long cryptocurrency-and-scalper-fueled shortage and many more months of Nvidia and AMD focusing on their pricey flagships. These are new, modern cards with modern features available at a price that can at least be called “literally affordable” even if they aren’t quite “budget.”
But the most common critique of them—and it’s a reasonable one—is that they don’t do much to change what you get when you buy a $250-to-$400 graphics card. Whether you’re talking about the RX 7600, the RX 6600 series, the RTX 3060 series, or the RTX 4060 series, you’re pretty much always getting a decent 1080p card, with 1440p within range with the right settings. Whether you’re an obsessive who talks yourself into upgrading every year or a more patient gamer who waits two or three generations between updates, the RX 7600 (like yesterday’s RTX 4060 Ti) doesn’t reward you much for your patience.
The RX 7600: RDNA 3 gets smaller
| RX 7900 XTX | RX 7900 XT | RX 7600 | RX 6600 | RX 6600 XT | RX 6650 XT | RX 6750 XT | |
|---|---|---|---|---|---|---|---|
| Compute units (Stream processors) | 96 (6,144) | 84 (5,376) | 32 (2,048) | 28 (1,792) | 32 (2,048) | 32 (2,048) | 40 (2,560) |
| Boost Clock | 2,500 MHz | 2,400 MHz | 2,600 MHz | 2,490 MHz | 2,589 MHz | 2,635 MHz | 2,600 MHz |
| Memory Bus Width | 384-bit | 320-bit | 128-bit | 128-bit | 128-bit | 128-bit | 192-bit |
| Memory Clock | 2,500 MHz | 2,500 MHz | 2,250 MHz | 1,750 MHz | 2,000 MHz | 2,190 MHz | 2,250 MHz |
| Memory size | 24GB GDDR6 | 20GB GDDR6 | 8GB GDDR6 | 8GB GDDR6 | 8GB GDDR6 | 8GB GDDR6 | 12GB GDDR6 |
| Total board power (TBP) | 355 W | 315 W | 165 W | 132 W | 160 W | 180 W | 250 W |
This is only the third desktop GPU that AMD has released based on the RDNA 3 architecture (not counting laptop GPUs and integrated GPUs, of which there have been several), so it’s worth recapping what it brings to the table compared to older RDNA 2 cards.
The biggest behind-the-scenes shift for the higher-end RDNA 3 GPUs was a move to a chiplet-based design like the one AMD uses for Ryzen processors. The 7600, based on the “Navi 33” GPU die, is still a monolithic design, with the memory controllers and all the GPU hardware on the same die. Navi 33 does double-duty as a laptop GPU in chips like the Radeon RX 7600M, and for both laptops and midrange desktop GPUs it’s possible that the added manufacturing complexity of a chiplet-based design wasn’t worth the extra flexibility it affords. Navi 33 is also built on a 6nm process, rather than the 5nm process of the 7900 XTX and XT or the 7nm process of most RX 6000-series GPUs.
Unlike Nvidia’s cards, AMD’s reference cards stick with the 8-pin PCIe power connector.
Andrew Cunningham
In theory, the move to a newer manufacturing process should save power, and the 5nm process is part of what makes the RX 7900 XTX and XT a good bit more efficient than the last-generation 6900-series flagships. The RX 7600 doesn’t seem to benefit as much from 6nm, using more power than the vanilla RX 6600 and about the same amount as the RX 6600 XT (the 6600 XT and 7600 have a very similar hardware configuration, too).
The architecture’s big new feature of note is the addition of hardware-accelerated encoding for the AV1 codec, which is poised to become the new baseline for video streaming as adoption slowly improves. It confers most of the same benefits as h.265/HEVC video—higher resolutions or better quality at the same bitrate as h.264—but without the royalties of h.265.
AMD also says that RX 7600 cards can support DisplayPort 2.1 but that it’s an optional feature. AMD’s first-party cards and third-party cards that use AMD’s reference design will all include it, but some others will trundle on with DisplayPort 1.4a. When in doubt, check the spec sheet.
AMD’s RX 7600 reference card uses the same design language as the RX 7900 XTX.
Andrew Cunningham
Nvidia restricts the new DLSS Frame Generation feature to its 4000-series GPUs, but AMD doesn’t have an equivalent feature, at least not yet. Version 3 of the company’s Fidelity Super Resolution upscaling tech will include a similar sort of frame generation, but we don’t know when it will come out or what its hardware requirements will be (FSR 2 runs on most modern graphics hardware, including Nvidia’s).
As for the card itself, AMD’s RX 7600 reference card uses the same design language as the RX 7900 XTX and XT, with small red highlights on the heatsink and backplate. It’s a small, cute GPU with a single 8-pin power connector that should have no trouble sliding into most desktop PCs.
