AMD's next generation RDNA3 graphics architecture generation could see a near-quadrupling in raw SIMD muscle over the current RDNA2, according to a spectacular rumor. Apparently, the company will deploy as many as 15,360 stream processors (quadruple that of a Radeon RX 6800), and spread across 60 WGPs (Workgroup Processors), and do away with the compute unit. This is possibly because the RDNA3 compute unit won't be as independent as the ones on the original RDNA or even RDNA2, which begins to see groups of two CUs share common resources.

Another set of rumors suggest that AMD won't play NVIDIA's game of designing GPUs with wide memory bus widths, and instead build on its Infinity Cache technology, by increasing the on-die cache size and bandwidth, while retaining "affordable" discrete memory bus widths, such as 256-bit. As for the chip itself, it's rumored that the top RDNA3 part, the so-called "Navi 31," could feature a multi-chip module design (at least two logic dies), each with 30 WGPs. Each of the two is expected to be built on a next-gen silicon fabrication node that's either TSMC N5 (5 nm), or a special 6 nm node TSMC is designing for AMD. Much like the next-generation "Lovelace" architecture by NVIDIA, AMD's RDNA3 could see the light of the day only in 2022.

On the road to its rumored August market availability, AMD could unveil its Radeon RX 6600 XT graphics card on July 30, at the ChinaJoy expo. The RX 6600 XT is a middle-of-the-market product that AMD likely expects to sell in heaps, in markets such as China. It competes with NVIDIA's GeForce RTX 3060 series. According to VideoCardz, the RX 6600 XT could see a July 30 unveiling, which usually comes with some teasers as to what buyers can expect from the card; followed by an August 6 unboxing embargo. The card launches on August 11. The RX 6600 (non-XT) is expected to launch later, in September. The RX 6600 XT is based on the 7 nm "Navi 23" silicon, and is rumored to feature 2,048 stream processors across 32 RDNA2 compute units. It also comes with 8 GB of GDDR6 memory across a 128-bit wide memory bus, and 32 MB of on-die Infinity Cache memory. AMD is likely targeting the 1080p AAA and 1440p e-sports crowds with this card.

AMD is allegedly preparing for an August 2021 debut of its Radeon RX 6600 XT graphics card in the retail segment, according to tech YouTuber Coreteks. Released exclusively as custom-design cards, through the company's AIB partners, the card will reportedly come with an MSRP price of USD $399, or roughly $100 less than that of the RX 6700 XT (which is being scalped for north of $800). The lack of a reference-design (MBA) model in the retail channel means that the card will not be sold through the AMD website.

The Radeon RX 6600 XT will reportedly be based on the 7 nm "Navi 23" silicon, although it remains to be seen if it maxes the silicon out. 8 GB will be the standard memory amount, across a 128-bit wide GDDR6 memory interface, cushioned by a 32 MB Infinity Cache. As for performance, Coreteks predicts that the RX 6600 XT will perform ±5% of the RX 5700 XT and the GTX 1080 Ti, which could make it a formidable card for AAA gaming at 1080p, or at 1440p with FSR.

AMD today announced the AMD Radeon PRO W6000 series workstation graphics, delivering exceptional performance, stability and reliability for professional users. The new graphics products were designed and optimized to power demanding architectural design workloads, ultra-high resolution media projects, complex design and engineering simulations, and advanced image and video editing applications.

Built on award-winning AMD RDNA2 architecture, the foundation of next generation, high-powered PCs, laptops and many of today's game consoles, the new product line features the AMD Radeon PRO W6800, the fastest AMD RDNA workstation graphics card ever. It also includes the AMD Radeon PRO W6600 graphics card, meticulously engineered for ultra-high performance workflows, and the AMD Radeon PRO W6600M GPU, designed to power professional mobile workstations.

Graphics cards have been developed over the years so that they feature multi-level cache hierarchies. These levels of cache have been engineered to fill in the gap between memory and compute, a growing problem that cripples the performance of GPUs in many applications. Different GPU vendors, like AMD and NVIDIA, have different sizes of register files, L1, and L2 caches, depending on the architecture. For example, the amount of L2 cache on NVIDIA's A100 GPU is 40 MB, which is seven times larger compared to the previous generation V100. That just shows how much new applications require bigger cache sizes, which is ever-increasing to satisfy the needs.

Today, we have an interesting report coming from Chips and Cheese. The website has decided to measure GPU memory latency of the latest generation of cards - AMD's RDNA 2 and NVIDIA's Ampere. By using simple pointer chasing tests in OpenCL, we get interesting results. RDNA 2 cache is fast and massive. Compared to Ampere, cache latency is much lower, while the VRAM latency is about the same. NVIDIA uses a two-level cache system consisting out of L1 and L2, which seems to be a rather slow solution. Data coming from Ampere's SM, which holds L1 cache, to the outside L2 is taking over 100 ns of latency.

AMD on April 1st published a new patent application that seems to show the way its chiplet GPU design is moving towards. Before you say it, it's a patent application; there's no possibility for an April Fool's joke on this sort of move. The new patent develops on AMD's previous one, which only featured a passive bridge connecting the different GPU chiplets and their processing resources. If you want to read a slightly deeper dive of sorts on what chiplets are and why they are important for the future of graphics (and computing in general), look to this article here on TPU.

The new design interprets the active bridge connecting the chiplets as a last-level cache - think of it as L3, a unifying highway of data that is readily exposed to all the chiplets (in this patent, a three-chiplet design). It's essentially AMD's RDNA 2 Infinity Cache, though it's not only used as a cache here (and for good effect, if the Infinity Cache design on RDNA 2 and its performance uplift is anything to go by); it also serves as an active interconnect between the GPU chiplets that allow for the exchange and synchronization of information, whenever and however required. This also allows for the registry and cache to be exposed as a unified block for developers, abstracting them from having to program towards a system with a tri-way cache design. There are also of course yield benefits to be taken here, as there are with AMD's Zen chiplet designs, and the ability to scale up performance without any monolithic designs that are heavy in power requirements. The integrated, active cache bridge would also certainly help in reducing latency and maintaining chiplet processing coherency.

AMD has revealed the Infinity Cache size for the upcoming Navi 23 GPU, as well as its absence in the next-generation Van Gogh APU, which features Zen 2 cores and an RDNA GPU. The reveal comes via a new patch done by AMD to the AMKFD, a Linux kernel HSA driver for AMD APUs. The patch file doesn't list Infinity Cache per se, but does clarify the last-level cache for AMD's GPUs - L3, which is essentially the same.

The patch reveals L3 size for Sienna Cichlid (Navi 21), Navy Flounder (Navi 22), and Dimgrey Cavefish (Navi 23). Navi 21 features 128*1024 (128 MB) of Infinity Cache, the just-released Navi 22 has 96 MB, as we know, and according to the file, Navi 23 is bound to feature 32 MB of it. Considering that Van Gogh lacks an infinity Cache, it would seem that it's making use of previous-gen Navi graphics, and won't leverage RDNA2, of which the Infinity Cache is a big part of. It remains to be seen if Van Gogh will materialize in an APU product lineup or if it's a specific part for a customer. It also remains to be seen which RX product will Navi 23 power - if an AMD RX 66000 series, or 6500 series.

TUL Corporation, a leading and innovative manufacturer of AMD graphics cards since 1997, today announced the new PowerColor AMD Radeon RX 6700 XT graphics card. Powered by the groundbreaking AMD RDNA 2 gaming architecture and featuring AMD Infinity Cache, AMD Smart Access Memory and other advanced technologies, the new graphics card delivers the ultimate in 1440p gaming performance.

The PowerColor AMD Radeon RX 6700 XT graphics card is built upon 7 nm process technology and AMD RDNA 2 gaming architecture, designed to deliver the optimal combination of performance and power efficiency. AMD RDNA 2 based graphics cards are optimized to deliver real-time lighting, shadow and reflection realism with DirectX Raytracing (DXR). When paired with AMD FidelityFX, developers can combine rasterized and ray-traced effects to provide the ideal balance of image quality and gaming performance.

AMD today announced the Radeon RX 6700 XT, its fourth RX 6000 series graphics card based on the RDNA2 graphics architecture. The card debuts the new 7 nm "Navi 22" silicon, which is physically smaller than the "Navi 21" powering the RX 6800/RX 6900 series. The RX 6700 XT maxes out "Navi 22," featuring 40 RDNA2 compute units, amounting to 2,560 stream processors. These are run at a maximum Game Clock frequency of 2424 MHz, a significant clock speed uplift over the previous-gen. The card comes with 12 GB of GDDR6 memory across a 192-bit wide memory interface. The card uses 16 Gbps GDDR6 memory chips, so the memory bandwidth works out to 384 GB/s. The chip packs 96 MB of Infinity Cache on-die memory, which works to accelerate the memory sub-system. AMD is targeting a typical board power metric of 230 W. The power input configuration for the reference-design RX 6700 XT board is 8-pin + 6-pin.

AMD is marketing the RX 6700 XT as a predominantly 1440p gaming card, positioned a notch below the RX 6800. The company makes some staggering performance claims. Compared to the previous-generation the RX 6700 XT is shown beating the GeForce RTX 2080 Super. NVIDIA marketed the current-gen RTX 3060 Ti as having the same performance outlook. Things get interesting, where AMD shows that in select games, the RX 6700 XT can even beat the RTX 3070, a card NVIDIA marketed as matching its previous-gen flagship, the RTX 2080 Ti. AMD is pricing the Radeon RX 6700 XT at USD $479 (MSRP), which is very likely to be bovine defecation, given the prevailing market situation. The company announced a simultaneous launch of its reference-design and AIB custom-design boards, starting March 18, 2021.AMD's performance claims follow.

AMD has filed a patent whereby they describe a MLA (Machine Learning Accelerator) chiplet design that can then be paired with a GPU unit (such as RDNA 3) and a cache unit (likely a GPU-excised version of AMD's Infinity Cache design debuted with RDNA 2) to create what AMD is calling an "APD" (Accelerated Processing Device). The design would thus enable AMD to create a chiplet-based machine learning accelerator whose sole function would be to accelerate machine learning - specifically, matrix multiplication. This would enable capabilities not unlike those available through NVIDIA's Tensor cores.

This could give AMD a modular way to add machine-learning capabilities to several of their designs through the inclusion of such a chiplet, and might be AMD's way of achieving hardware acceleration of a DLSS-like feature. This would avoid the shortcomings associated with implementing it in the GPU package itself - an increase in overall die area, with thus increased cost and reduced yields, while at the same time enabling AMD to deploy it in other products other than GPU packages. The patent describes the possibility of different manufacturing technologies being employed in the chiplet-based design - harkening back to the I/O modules in Ryzen CPUs, manufactured via a 12 nm process, and not the 7 nm one used for the core chiplets. The patent also describes acceleration of cache-requests from the GPU die to the cache chiplet, and on-the-fly usage of it as actual cache, or as directly-addressable memory.

AMD has launched its RDNA 2 based graphics cards, codenamed Navi 21. These GPUs are set to compete with NVIDIA's Ampere offerings, with the lineup covering the Radeon RX 6800, RX 6800 XT, and RX 6900 XT graphics cards. Until now, we have had reviews of the former two, but not the Radeon RX 6900 XT. That is because the card is coming at a later date, specifically on December 8th, in just a few days. As a reminder, the Radeon RX 6900 XT GPU is a Navi 21 XTX model with 80 Compute Units that give a total of 5120 Stream Processors. The graphics card uses a 256-bit bus that connects the GPU with 128 MB of its Infinity Cache to 16 GB of GDDR6 memory. When it comes to frequencies, it has a base clock of 1825 MHz, with a boost speed of 2250 MHz.

Today, in a GeekBench 5 submission, we get to see the first benchmarks of AMD's top-end Radeon RX 6900 XT graphics card. Running an OpenCL test suite, the card was paired with AMD's Ryzen 9 5950X 16C/32T CPU. The card managed to pass the OpenCL test benchmarks with a score of 169779 points. That makes the card 12% faster than RX 6800 XT GPU, but still slower than the competing NVIDIA GeForce RTX 3080 GPU, which scores 177724 points. However, we need to wait for a few more benchmarks to appear to jump to any conclusions, including the TechPowerUp review, which is expected to arrive once NDA lifts. Below, you can compare the score to other GPUs in the GeekBench 5 OpenCL database.

TUL Corporation, a leading and innovative manufacturer of AMD graphic cards since 1997, today is announcing its new PowerColor Fighter RX 6800 based on AMD's latest RDNA2 architecture and with 16 GB of GDDR6, offering solid 4K gaming performance for the most price conscious gamers. Powered by AMD RDNA 2 architecture, PowerColor Fighter RX 6800 graphics card features 60 Compute Units, 128 MB of all new AMD Infinity Cache and 16 GB of dedicated GDDR6 memory, is engineered to deliver ultra-high frame rates and serious 4K resolution gaming.

PowerColor Fighter RX 6800 graphics card is designed for gamers on a budget but with powerful gaming performance in mind. Fighter RX 6800 is paired with effective triple cooling fan design, Mute Fan Technology, BIOS switch button, and reinforced back plate, so the Fighter can fight with you to enjoy exceptional gaming experiences.

Not to be outdone in comparison to almost all other AAA game releases in recent times, Far Cry 6 has recently been delayed from its February 18th release to May 25th 2021. However, that extra time may come to serve the game nicely, in that it may allow for all the planned features to be integrated. As part of AMD's partner showcase, Ubisoft has revealed that Far Cry 6 will make extensive use of DirectX 12 Ultimate features, featuring raytracing, AMD's Contrast Adaptive Sharpening (CAS), Variable Rate Shading (VRS), as well as Hybrid SSR (Stochastic Screen Space Reflections).

Ubisoft's head of 3D programming Oleksandr Polishchuk had this to say: "We were very impressed with the latest AMD technologies and joined forces to bring FidelityFX CAS, DXR raytracing and Variable Rate Shading to Far Cry 6. We are working together to ensure a smooth 4K viewing experience. This requires a lot of bandwidth, memory, and a Radeon RX 6000 with Infinity Cache that can handle it easily while maintaining high FPS rates." Check out the AMD partner showcase video below.

AMD could follow up on its RX 6800 series and RX 6900 XT launches with the RX 6700 series, which logically succeeds the RX 5700 series, and competes with NVIDIA's RTX 3060/Ti. Patrick Schur on Twitter, who has a high hit-rate with specs of upcoming AMD products, put out possible specs of the RX 6700 series. Both are based on the new "Navi 22" silicon, with an interesting set of specifications.

Apparently 12 GB could be AMD's new memory amount for the mid-range. It's unknown whether the 12 GB is running over a 192-bit wide memory interface (6x 16 Gbit chips), or whether AMD is using mixed-density chips over a 256-bit wide memory bus (think 4x 16 Gbit and 4x 8 Gbit), because even the fastest JEDEC-standard GDDR6 chips, running at 16 Gbps, would only yield 384 GB/s memory bandwidth, which is less than the 448 GB/s the RX 5700 series enjoy. Perhaps an Infinity Cache is deployed to make up the difference?

Unlike NVIDIA, AMD still relies on its add-in board (AIB) partners to sell reference design (made by AMD) graphics cards, and Sapphire just announced its lineup. The company unveiled its reference-design Radeon RX 6800 XT and RX 6800 cards. The RX 6800 XT is characterized by its triple-slot cooling solution, while the RX 6800 makes do with a slimmer dual-slot one. Both cards are based on the 7 nm "Navi 21" silicon and feature 16 GB of 16 Gbps GDDR6 memory over a 256-bit wide memory interface, cushioned by 128 MB of on-die Infinity Cache.

The RX 6800 XT is configured with 72 out of 80 RDNA2 compute units on the "Navi 21" silicon, working out to 4,608 stream processors, 72 ray accelerators, 288 TMUs, and 128 ROPs. The engine clock of the RX 6800 XT boosts up to 2.25 GHz. The RX 6800, on the other hand, features 60 out of 80 RDNA2 compute units, which make up 3,840 stream processors, 60 ray accelerators, 240 TMUs, the same 128 ROPs, and the same memory subsystem as the RX 6800 XT. Given that these are reference cards, Sapphire could price them at AMD's baseline, with the RX 6800 XT going for $649, and the RX 6800 at $579.

AMD's RDNA2-based cards are just around the corner, with the company's full debut of the secrecy-shrouded cards being set for October 28th. Rumors of high clocks on AMD's new architecture - which were nothing more than unsubstantiated rumors up to now - have seemingly been confirmed, with Patrick Schur posting on Twitter some specifications for upcoming RNDA2-based Navi 21 XT. Navi 21 XT falls under the big Navi chip, but likely isn't the top performer from AMD - the company is allegedly working on a Navi 21 XTX solution, which ought to be exclusive to their reference designs, with higher clocks and possibly more CUs.

The specs outed by Patrick are promising, to say the least; that AMD's Big Navi can reach clocks in excess of 2.4 GHz with a 250 W+ TGP (quoted at around 255 W) is certainly good news. The 2.4 GHz (game clock) speeds are being associated with AIB cards; AMD's own reference designs should be running at a more conservative 2.3 GHz. A memory pool of 16 GB GDDR6 has also been confirmed. AMD's assault on the NVIDIA 30-series lineup should embody three models carved from the Navi 21 chip - the higher performance, AMD-exclusive XTX, XT, and the lower performance Navi 21 XL. All of these are expected to ship with the same 256 bit bus and 16 GB GDDR6 memory, whilst taking advantage of AMD's (rumored, for now) Infinity Cache to make up for the lower memory speeds and bus. Hold on to your hats; the hype train is going full speed ahead, luckily stopping in a smooth manner come October 28th.