Radeon 7900 XTX and XT Review: Faster, hotter, and cheaper than the RTX 4080

The three-fan cooler of the Radeon RX 7900 XTX.
Enlarge / The three-fan cooler of the Radeon RX 7900 XTX.

Andrew Cunningham

Nvidia’s RTX 4080 and 4090 GPUs are amazing performers. They’re also surprisingly expensive, starting at $1,200 and $1,500 path look for cards from partners like MSI, Gigabyte, and Asus. The 4080 is almost twice the price of the RTX 3080’s original $699 MSRP.

These price hikes are being driven in part by pandemic-era concerns like supply chain snarls and inflation, and in part by a cryptocurrency-fueled boom (which is thankfully over) that has encouraged a network of scalpers to snag every single high-end GPU out there they could find. A lack of competition and the increasing cost and complexity of building gigantic, monolithic chips using state-of-the-art manufacturing processes also played a role. Today AMD is trying to solve the last two problems with the launch of its Radeon RX 7900 series GPUs.

At $899 and $999, the RX 7900 XT and RX 7900 XTX are still objectively expensive – but as they don’t represent a further escalation beyond the starting price of the RX 6900 XT, both cards are considered bargains in today’s GPU market. If you’re looking for cards that can consistently handle 4K gaming at 60fps and beyond, these GPUs do it for less than Nvidia’s latest, and they’re good enough and fast enough that they’ll hopefully start to match Nvidia’s prices something to lower , too.

But Nvidia still retains some key advantages that complicate a simple David-and-Goliath narrative. These GPUs don’t quite feel like a Ryzen moment for AMD’s graphics division — a tipping point where a rowdy AMD manages to severely chip the market share of an established, complacent competitor. But if you can actually find them at their starting prices, they’re the first sign we’ve had in a while that some relief is coming for high-end but budget-conscious PC gamers.

Bigger sat nav

AMD's chiplet design can be seen in this shot - a large center chip with most of the compute resources and six smaller chips containing the cache and memory controllers.
Enlarge / AMD’s chiplet design can be seen in this shot – a large center chip with most of the compute resources and six smaller chips containing the cache and memory controllers.

AMD

The RX 7000 series is the third iteration of the RDNA GPU architecture, occasionally referred to as “Navi” after the codenames of the GPU chips themselves. RDNA 3 adds nothing that feels as significant as ray tracing support from RDNA 2, but AMD added a lot of extra hardware and made major changes under the hood.

The most significant is a new chiplet-based approach, similar to the concept AMD uses for its Ryzen CPUs. Instead of building the entire GPU chip in one manufacturing process – which increases the size of the chip and thus the chance that part or all of it could be defective – AMD builds the main Navi 3 GPU chip on a 5nm -TSMC fabrication process and a range of smaller memory controller dies (MCDs) in a 6nm process. These chips are all connected via a high-speed interconnect, which AMD says can transfer data at speeds of up to 5.3 terabytes per second.

The Main Graphics Compute Die (GCD) contains most of the hardware you think of when you think of a GPU – compute units, shaders, ray tracing hardware, the media encoding and decoding block, and the display output. Both the 7900 XTX and XT use the same Navi 31 GCD, but the XTX runs at higher clock speeds and has more CUs and stream processors enabled. The XTX has 96 CUs and 6,144 stream processors, while the XT has 84 CUs and 5,376 stream processors. Both cards represent a leap up from the Navi 21 die used in the RX 6900 series, which was maxed out at 80 CUs and 5,120 stream processors (and that’s before you factor in other performance-enhancing improvements).

RDNA 3 brings a little more of everything to AMD's top-tier GPUs, from CU counts to memory bandwidth.
Enlarge / RDNA 3 brings a little more of everything to AMD’s top-tier GPUs, from CU counts to memory bandwidth.

AMD

The MCDs all contain a single 64-bit memory controller and 16MB of AMD Infinity Cache, demonstrating the benefits of a chiplet-based approach. The 7900 XTX has a 384-bit memory bus and 96MB Infinity Cache, while the 7900 XT has a 320-bit bus and 80MB cache; All AMD has to do is remove an MCD. The exact same MCDs can be reused with all the different RDNA 3 GCDs that AMD is releasing, from low-end products with a single MCD to mid-range GPUs that use between two and four. Defects in MCD chips do not require the larger, more complex GCDs to be discarded or thrown into a bin, and vice versa.

When it comes to the brand new features for RDNA 3, there are three things worth noting. First, the GPUs contain new AI accelerators that could be useful for both the plethora of AI-assisted content creation that has emerged over the last year and AI-assisted upscaling (if AMD decides to include them in a future release FSR upscaling algorithm to be implemented by it; both DLSS and XeSS use AI for upscaling, but FSR 2.0 does not).

Second, the video encoding and decoding block supports hardware-accelerated encoding for the AV1 video codec, just like Intel’s RTX 4000 series and Arc GPUs. This should be useful for content creators as well as streamers who want to stream higher resolution videos or videos with the same resolution while using less bandwidth.

And third, the “Radiance Display Engine” adds DisplayPort 2.1 support to the GPUs. Monitors that take full advantage of the additional bandwidth provided by DisplayPort 2.1 don’t really exist as of this writing, but when they do, RDNA 3 GPUs can drive 4K displays up to 480Hz and 8K displays up to 165Hz .

Leave a Reply

Your email address will not be published. Required fields are marked *