The next generation of AMD graphics cards is coming soon, and it’s rumored to finally include a true competitor to Nvidia’s high-end graphics cards. We already know AMD is supplying the graphics power behind the upcoming Xbox Series X and Playstation 5, but that same architecture will also be used in a high-end GPU that can compete with Nvidia’s recently launched RTX 3080 and 3090.
The architecture behind these new GPUs is known as RDNA 2, which will be used both in these new PC graphics, as well as in the new consoles. AMD has a proven track record of delivering the most bang for the buck when it comes to performance per watt, as evidenced by the success of the Radeon RX 5700 XT.
With what’s been referred to as “Big Navi,” AMD will reportedly be focusing on delivering both better power efficiency and performance gains, as well as adding support for ray tracing to its GPU lineup for the first time. Here’s everything we know so far.
Release date and price
AMD has started a countdown clock, and we can expect the company’s next-generation graphics cards to be unveiled on October 28, with the new Zen 3-based processors getting an October 8 announcement.
— Radeon RX (@Radeon) September 9, 2020
During AMD’s Financial Analyst Day earlier this year, CEO Lisa Su promised that its next-generation graphics architecture will be ready by the end of 2020. The timeline meshes well with Microsoft’s and Sony’s plans for their respective consoles, as both hardware makers have promised that the Xbox Series X and PlayStation 5 will be available for the 2020 holiday season.
This could offer the company a significant boost in mind share as it tries to narrow the gap with rival Nvidia. If AMD is able to deliver on consoles, Big Navi on that platform could serve as a halo product that will in turn generate future sales for PCs.
Though AMD has been aggressive on price in the past, PC gamers looking for a steep discount on Big Navi may be in for a disappointment. In press presentations throughout this year, Su had stated that as AMD realizes its investments in graphics deliver significant value to gamers, the company has slowly been increasing the prices of its GPUs in recent years in an attempt to generate better margins. We expect this will continue with Big Navi, especially since the demand for gaming and entertainment continues to climb amid new work, study, and play from home environments created by the global health pandemic.
For reference, pricing for Nvidia’s current GPU flagship, the Radeon VII, is very close to rival Nvidia’s pricing. Benchmarks showed that Radeon VII performed very similar to the $699 GeForce RTX 2080, and AMD’s pricing of $679 revealed that gamers didn’t get much of a discount by choosing Team Red. With Nvidia pricing the RTX 3080 at $699, we expect the Radeon RX 6000 price on PC to be just below that.
Nvidia’s aggressive price on the RTX 3080 and the card’s seemingly strong performance will likely put pressure on AMD to either deliver matching performance – if it wants to position Big Navi as a competitor to its rival’s Ampere architecture – or compete on price.
The RX 6000 brand confirmed
The “6000” branding was spotted in Fortnite as an Easter egg. Gamer @MissGinaDarling posted on Twitter that she was transported into a secret Radeon room inside the game with a special code of “6000,” which insiders speculated could be used for the Radeon RX 6000’s final branding. Following that buzz, Frank Azor, AMD chief architect of gaming solutions, confirmed the Easter egg and hinted more clues are coming.
— Gina (@MissGinaDarling) September 4, 2020
AMD’s Radeon RX 6000 family lineup is expected to include multiple cards, spanning entry-level, midrange, and premium GPUs. Like the 5000 series, AMD could offer an RX 6300 as a successor to the RX 5300 at the entry-level spot and position its RX 6500 and RX 6300 as midrange offerings to take on Nvidia’s GeForce RTX 3070. At the premium end, AMD could replace its RX 5700 with a new RX 6700. And given that AMD had offered a premium Radeon RX 5700 XT and $499 Radeon XT 50th Anniversary Edition, the company could offer high-end cards with a new RX 6900 model that will be designed to compete against Nvidia’s GeForce RTX 3080 and RTX 3090.
AMD also offers discrete mobile graphics for laptops, with models like the RX 5300M, 5500M, 5600M, and 5700M. These chips also come in Pro configurations for mobile workstations targeted at enterprise use. The RX 6000 series could similarly arrive in similar configurations for use in laptops and mobile workstations to take on Nvidia RTX Studio laptops.
It’s unclear what AMD’s strategy will be for the Mac, given Apple recently announced that it will begin transitioning from Intel’s processors to its own Apple Silicon for the Mac. The RX 5000 series is also available on the iMac as the Radeon Pro 5300, 5500XT, 5700, and 5700XT. When Apple’s transition is fully realized, it will use its own integrated GPU solution on its custom ARM-based processors.
RX 6000 coming to APUs
In addition to premium GPUs and consoles, Big Navi will also be headed into AMD’s APUs as an integrated GPU solution, according to AMD Chief Financial Officer David Kumar, It’s a move that’s similar to what rival Intel is doing with its Intel Xe graphics architecture.
The company’s Ryzen 4000 APUs with integrated Radeon graphics offer enough performance to play many 1080p titles using the game’s lowest graphics settings. This would allow basic desktops — and mobile laptops — to play basic games without requiring a costly or more bulky setup with discrete GPUs. When AMD compared its Ryzen 7 4700G desktop APU to Intel’s Core i7-9700, it showed that GPU performance was up to 274% higher than its competition.
The RDNA 2-based APUs will likely provide AMD with additional boost, but Intel’s new integrated Xe architecture on its 11th-Gen processors is also coming with a significant boost in performance. Headed to Tiger Lake-based laptops this fall, Intel’s integrated Xe graphics promises 2.7 times faster content creation, more than 20% better performance with Office tasks, and 2x faster faming performance over the prior Gen 11 integrated GPU. There are 96 execution cores for a total of 768 cores with 3.8MB of L3 cache on Intel’s Xe platform.
Intel’s 11th Gen Tiger Lake will likely compete against AMD’s Renoir and Nvidia’s discrete and lightweight GeForce MX in gaming, and Intel’s own numbers show that Xe is outperforming the competition, including the Ryzen 7 4800U. With Intel turning up the heat on AMD on integrated GPU performance, it will be interesting to see how a future Ryzen APU with integrated RDNA 2 graphics will compete against the new Xe architecture.
A recent tweet from @coreteks suggests that following Nvidia’s announcement of the GeForce RTX 3000 series, AMD could position the Radeon RX 6000 with 16GB of RAM – nearly 50% more than what’s available on the RTX 3080 flagship – at a price between $499 and $599, with $549 being the likely cost of ownership for this card.
Given that the new GDDR6X — or G6X — memory standard was developed by Micron in partnership with Nvidia, it will be an exclusive to the RTX 3080 and 3090. This means that AMD will have to rely on the G6 non-X GDDR6 memory format for its upcoming RDNA 2 card. Though you’re not getting as wide a bandwidth here, the new RDNA 2 cards could potentially come with even more video RAM — up to 16GB — which could offset some of the speed gains Nvidia is boasting with its 10GB of RAM on the RTX 3080. For reference, the GDDR6X memory can perform up to 760GB per second, while the 8GB GDDR6 memory on the Radeon RX 5700 XT 50th Anniversary Edition supports a maximum memory bandwidth of 448GB per second.
Previously, it was rumored that AMD’s new cards could come in HBM — or high bandwidth memory — versions, but more recent leaks reveal that those plans have been scrapped. HBM’s high cost would have likely contributed to a higher retail cost, which is something that AMD may want to avoid as it competes against Nvidia on both price and performance. It’s still unclear what AMD’s plans are for HBM, but the feature’s inclusion could make a high-end RX 6000 card competitive against Nvidia’s ultra-premium $1,499 RTX 3090 for enthusiast gamers and creative professionals.
In stark contrast to the plethora of leaks that emerged ahead of rival Nvidia’s GeForce RTX 3000 series announcements, design information for AMD’s upcoming Radeon RX 6000 cards hasnot surfaced ahead of an official unveiling. While Nvidia chose a rather controversial design language for its RTX 3000 cards in an effort to better control heat from its powerful card, it’s unclear what AMD’s strategy will be.
The company had already confirmed that it will be moving away from the blower-style heat sink this year in favor of a more attractive design. With more premium aesthetics, Team Red gamers can expect to shell out more for RGB backlighting kits to allow Big Navi to shine as a centerpiece on gaming rigs with tempered glass see-through panels. For reference, the metal-clad industrial aesthetic of the RX 5000 series was a bit dull, and the single-fan design made the card appear anemic.
With a more powerful offering this year, AMD could potentially add more fans to the card and give it a more aggressive design. It’s unclear, however, if the card will also grow in size. For comparison, the RTX 3090, which Nvidia CEO Jensen Huang dubbed as the BFGPU, comes in a triple-slot design. In general, the new RTX series debuted with a new 12-pin connector that’s designed to be more compact and will support the new power requirements of the cards.
Big Navi will use AMD’s next-generation GPU architectural core, which is a refinement of the RDNA architecture that debuted on the Radeon RX 5000 series graphics cards. Prior to the launch of Radeon RX 5000 series, AMD’s older Vega graphics relied on the GCN platform. According to AMD executives, Big Navi’s RDNA 2 architecture will provide a 50% uplift in performance per watt compared to RDNA, a similar improvement to what we saw with the transition from GCN to RDNA.
This means that RDNA 2 is expected to use up to 33% less power while maintaining performance levels similar to RDNA, or it will deliver 50% better performance with the same power consumption.
In 2020, AMD will likely only launch its RDNA 2 PC GPU in a reference design that will have a new cooler design, offering better and faster performance, according to TweakTown. Unfortunately, it looks like there won’t be enough time for third-party cards, and those designs could be pushed back to next year.
AMD’s new GPU will likely be manufactured using a more advanced 7nm+ node. Whereas Nvidia had turned to Samsung’s 8nm node for its RTX 3000 series, AMD’s new RX 6000 series will likely be made by partner TSMC. AMD had previously stated that the new RDNA 2 will utilize a redesigned microarchitecture that will feature an improved performance per clock.
AMD is also expected to implement Microsoft’s new DirectStorage feature, which allows the GPUs and SSDs to communicate directly with each other in an effort to eliminate load times for games, on its new Radeon RX 6000 cards. Rival Nvidia showed off the feature by calling it RTX I/O on its new RTX 3000 series.
The card will also bring ray tracing to AMD’s lineup for the first time, alongside support for variable rate shading, or VRS, that’s part of Microsoft’s DirectX 12 API on Windows. AMD will likely use its own optimizations and software to bring ray tracing to games. Even though the company has not revealed any details about ray tracing implementation, the company’s early discussions of Big Navi suggest that it will use the same Bounding Volume Hierarchy, or BVH, approach as its rival.
Early rumors, however, suggest that ray tracing won’t be available on all members of the Big Navi or Navi 2x family, which could indicate that the coveted feature will only be headed to AMD’s flagship cards. This could mean that ray tracing may require more complex hardware that may not be found on lowe- end cards, and enabling the feature across the board could lead to compromised performance, a situation that Nvidia found itself in with entry-level RTX 2000 series cards.
With RDNA 2, AMD has a shot at taking on Nvidia in the premium segment. Though AMD has an excellent track record in the midrange market, the RDNA 2 architecture will allow AMD to compete at all price points. While AMD’s current lineup is strong for gamers in the 1080p and 2K 1440p segments, Big Navi will bring “uncompromised 4K gaming,” according to a slide the company showed to investors.
“Nice launch from @Nvidia yesterday on their new graphics card,” Azor tweeted. “They are going to pair well with our latest @AMDRyzen CPUs. I can’t wait to show you all the great products our @Radeon team has been working on! What an awesome year to be a gamer!!!”
Early rumors suggest that Big Navi could deliver up to a 30% boost in performance compared to Nvidia’s Turing flagship, the GeForce RTX 2080 Ti, which is a $1,200 graphics card. Given the strong performance of the RTX 2080 Ti on games in 4K, a 30% uplift in performance compared to this card should give Big Navi plenty of headroom to handle ray tracing at higher resolutions. The launch of Big Navi and Nvidia’s RTX 3080 flagship should create an even bigger demand for 4K gaming, and we’ll likely see more 4K gaming-ready monitors as a result.
Given the 5,120 stream processors, which is assuming the same ratio thatAMD used with GCN, and a potential clock speed of 1,700 MHz, performance of Navi 2x should be 17.5 TFLOPs. This is twice the number of stream processors that’s available on the ultra-premium RX 5700XT 50th Anniversary Edition.
The performance could be even higher if clock speeds were higher or if AMD was able to squeeze more stream processors onto the architecture. This compares favorably to the 12 TFLOPs that Microsoft promised for its upcoming console. The PlayStation 5 is said to feature 11 TFLOPs of compute power and ship with 52 CUs clocked at 1,743 MHz.
Azor’s comment could foreshadow the performance that AMD has in store for Radeon RX 6000, which according to Igor’s Lab and reported by Hardware Times, could match Nvidia’s flagship RTX 3080 in performance.
The publication noted that at 300 watts, Big Navi matches the 320-watt RTX 3080 in performance. At 300 watts, RDNA 2 would match the power draw of the Radeon VII.
When used at 275 watts, Big Navi’s performance is between that of the RTX 3070 and RTX 3080. Unfortunately, though, it looks like AMD will not have a product to compete against Nvidia’s ultra-premium RTX 3090, which delivers 8K gaming at 60 FPS.
Big Navi is expected to feature a total of 5,120 cores, according to Wccftech. Twitter user @_rogame confirmed the core count in his own leak, adding that we should see up to four shader engines, up to 2 shader arrays per engine, up to five WGPs or 10 compute units per shader array, and up to four RBs per shader engine. In total, the tweet confirmed that we should see 80 compute units on a card labeled NAVI21, or double what’s available on the RX 5700 XT 50th Anniversary Edition.