AMD finally revealed its much-awaited RDNA 3 GPU architecture with the Radeon RX 7900 XTX and RX 7900 XT. The company announced both the price and release date for the cards at their event in Las Vegas, NV.
A few months back, rival GPU maker Nvidia shocked the gaming world with the stratospheric pricing of its flagship GeForce RTX 4090, with CEO Jensen Huang stating that high GPU prices are here to stay. AMD it seems, is not following suit. The company revealed that the RX 7900 XT would have a suggested retail price of $899 (Php 52,000) while its flagship RX 7900 XTX would have an SRP of $999 (Php 59,000).
Both will undercut Nvidia’s $1,599 (Php 94,000) GeForce RTX 4090 by a significant amount, when they launch on December 13, 2022.
Despite the lower price, AMD’s new cards still come with some impressive specs. The RX 7900 XTX promises to deliver 61 teraflops (TFLOPS) of compute performance. This is nearly thrice the performance of AMD’s previous-generation flagship, the RX 6950 XT.
To achieve this performance, the RX 7900 XTX employs 96 RDNA 3 Compute Units (CUs), operates at 2.3 Gigahertz (GHz) and packs 24 gigabytes (GB) of GDDR6 video RAM. The result is a card that AMD promises to be 67 percent faster than the RX 6950 XT it replaces while maintaining the same power draw of 335 watts.
The RX 7900 XT, on the other hand, comes with 84 RDNA 3 CUs, operates at 2.0 GHz and has 20GB of GDDR6 memory, resulting in a card that runs at a more modest 300W.
Both cards will use dual 8-pin power connectors instead of the problematic 12VHPWR connector used by Nvidia’s 4000 series. AMD is also not precluding its add-in-board partners from adding a third 8-pin connector for overclocked cards.
To further whet gamer’s appetites, AMD released some details of the RX 7900 XTX’s performance in several games. (Read: GUIDE: Should you buy an AMD graphics card?)
AMD achieves this performance with the new cards through the use of a chiplet-based design—a first for GPUs. Using its experience with its Ryzen CPU line, RDNA 3 uses two new core components: the Graphics Compute Die (GCD) and Memory Cache Die (MCD).
GCN represents the actual rendering cores used for floating point and integer operations. Built on TSMC’s 5 nanometer (nm) process node, the GCD includes both the hardware for AI and ray tracing as well as media and display engines. Both the RX 7900 XTX and RX 7900 XT come with one GCD each.
For video producers and content creators, the GCD also includes support for both AV1 encode and decode, AVC and HEVC. At the same time, it also features better AI acceleration, with a 2.7x improvement over the previous-generation cards. Meanwhile, ray tracing performance is up by 50 percent per CU compared to RDNA 2.
The GCN in these cards will be surrounded by six MCDs. Built on TSMC’s 6nm process node, each MCD contains a 64-bit memory controller and AMD’s second-generation Infinity Cache. All six will be active on the RX 7900 XTX, however, only five will be functional on the RX 7900 XT with its lower price.
Beyond the increased performance, both the RX 7900 XTX and RX 7900 XT will support DisplayPort 2.1 via AMD’s new Radiance Engine. The feature—which Nvidia did not include in the RTX 4090—will allow for frame rates up to 480Hz at 4K and 165Hz at 8K.
As for what this all means in comparison to Nvidia’s flagship cards, AMD did not talk much about them during its stream. However, in a post-stream Q&A, AMD hinted that it wasn’t necessarily looking to compete with the RTX 4090.
“We’re competing in $1,000 or below,” stated AMD’s Radeon GM Scott Herkelman. He added that he believed in the “performance and what we’re offering in terms of price.”
Whether or not AMD has made the right decision to slightly compromise on performance to keep their cards below the $1000 mark will pay off remains to be seen. That said, with the fall in the price of previous-gen GPUs due to the crypto crash, the RX 7900 XTX and RX 7900 XT GPUs’ lower price may just help swing a few purchases their way.
If you like reading our content, why not show your appreciation by treating us to a cup of coffee? (or two, if you’re feeling generous)