Apparently, AMD's next-generation gaming graphics card is closer to launch than anyone in the media expected, with mass-production of the so-called Radeon RX 8800 XT poised to begin later this month, if sources on ChipHell are to be believed. The RX 8800 XT will be the fastest product from AMD's next-generation, and will be part of the performance segment, succeeding the current RX 7800 XT. There will not be an enthusiast-segment product in this generation, as AMD looks to consolidate in key market segments with the most sales. The RX 8800 XT will be powered by AMD's next-generation RDNA 4 graphics architecture.

There are some spicy claims related to the RX 8800 XT being made. Apparently, the card will rival the current GeForce RTX 4080 or RTX 4080 SUPER in ray tracing performance, which would mean a massive 45% increase in RT performance over even the current flagship RX 7900 XTX. Meanwhile, the power and thermal footprint of the GPU is expected to reduce with the switch to a newer foundry process, with the RX 8800 XT expected to have 25% lower board power than the RX 7900 XTX. Unlike the "Navi 31" and "Navi 32" powering the RX 7900 series and RX 7800 XT, respectively, the "Navi 48" driving the RX 8800 XT is expected to be a monolithic chip built entirely on a new process node. If we were to guess, this could very well be TSMC N4P, a node AMD is using for everything from its "Zen 5" chiplets to its "Strix Point" mobile processors.

We've known since May that AMD is giving its next generation RDNA 4 graphics architecture a significant upgrade with ray tracing performance, and had some clue since then, that the company is working on putting more of the ray tracing workflow through dedicated, fixed function hardware, unburdening the shader engine further. Kepler_L2, a reliable source with GPU leaks sheds light on some of the many new hardware features AMD is introducing with RDNA 4 to better accelerate ray tracing, which should give its GPUs a reduced performance cost of having ray tracing enabled. Kepler_L2 believes that these hardware features should also make it to the GPU of the upcoming Sony PlayStation 5 Pro.

To begin with, the RDNA 4 ray accelerator introduces the new Double Ray Tracing Intersect Engine, which should at least mean a 100% ray intersection performance increase over RDNA 3, which in turn offered a 50% increase over that of RDNA 2. The new RT instance node transform instruction should improve the way the ray accelerators handle geometry. Some of the other features we have trouble describing include a 64-byte RT node, ray tracing tri-pair optimization, Change flags encoded in barycentrics to simplify detection of procedural nodes; improved BVH footprint (possibly memory footprint): and RT support for oriented bounding box and instance node intersection. AMD is expected to debut Radeon RX series gaming GPUs based on RDNA 4 in early 2025.

Phoronix head honcho, Michael Larabel, has noticed another set of interesting updates for AMD Graphics on Linux—albeit in preparation for next generation solutions: "engineers on Monday (February 5) posted a few new patch series for enabling some updated IP (intellectual property) blocks within their open-source AMDGPU Linux kernel graphics driver. This new IP is presumably part of the ongoing hardware enablement work for their next-gen RDNA 4 graphics." Team Red GitHub patches for "GFX12" targets appeared online last November, again highlighted by Larabel's investigative work—AMD engineers appear to be quite determined with their open-source software endeavors, as seen in LLVM-Project notes regarding GFX1200's enablement.

The new "IP block" updates included patches for the enabling ATHUB 4.1, LSDMA 7.0, IH 7.0, and HDP 7.0—presumably for next generation Radeon graphics solutions. Larabel provided a quick breakdown of these components: "ATHUB 4.1 is needed for clock-gating / power management features, LSDMA 7.0 is the latest IP for Light SDMA for general purpose System DMA (SDMA) on the GPU, IH 7.0 for the Interrupt Handler on the GPU, and HDP 7.0 for the Host Data Path support for CPU accessing the GPU device memory via the PCI BAR. As far as code changes, the big chunks of the work are from the auto-generated header files." He believes that AMD's engineers have largely moved on from current generation tasks: "The big version bumps for these IP blocks all the more are likely indicative of these bits being for next-gen RDNA 4 as opposed to further iterating on RDNA3 or similar." The patches could be merged into the upcoming Linux 6.9 release, possibly coinciding with a Radeon RX 8000 series rollout.

We've known since way back in August 2023, that AMD is rumored to be retreating from the enthusiast graphics segment with its next-generation RDNA 4 graphics architecture, which means that we likely won't see successors to the RX 7900 series squaring off against the upper end of NVIDIA's fastest GeForce RTX "Blackwell" series. What we'll get instead is a product stack closely resembling that of the RX 5000 series RDNA, with its top part providing a highly competitive price-performance mix around the $400-mark. A more recent report by Moore's Law is Dead sheds more light on this part.

Apparently, the top Radeon RX SKU based on the next-gen RDNA4 graphics architecture will offer performance comparable to that of the current RX 7900 XTX, but at less than half its price (around the $400 mark). It is also expected to achieve this performance target using a smaller, simpler silicon, with significantly lower board cost, leading up to its price. What's more, there could be energy efficiency gains made from the switch to a newer 4 nm-class foundry node and the RDNA4 architecture itself; which could achieve its performance target using fewer numbers of compute units than the RX 7900 XTX with its 96.

With NVIDIA's Ampere and Hopper GPUs enjoying a domination in the AI acceleration industry, compute companies are turning to AMD's Instinct CDNA series accelerators to look for alternatives. It seems like they've found one. This has financial market analysts excited, causing the AMD company stock to hit near record highs. AMD recently launched the Instinct MI300X and MI300A processors based on the CDNA 3 architecture, which the company claims beat NVIDIA's H100 "Hopper" processors at competitive prices, which has encouraged analysts from major financial institutions, including Barclays, KeyBanc Capital, and Susquehanna Financial Group, to increase their price targets for the AMD stock. As of market closure at Jan 17, 7:59:56 PM UTC, the AMD stock stood at $160.17, near its November 2021 record high of $164.46.

AMD's data center business looks to ramp up Instinct CDNA accelerators through 2024. These large chiplet-based GPUs are based on the same 5 nm TSMC foundry nodes to NVIDIA's H100 "Hopper," and to maximize the use of its foundry allocation, it's been reported that AMD might even forego large gaming GPUs based on its Radeon RX RDNA4 architecture, to maximize its allocation for high-margin CDNA3 chips. The Instinct MI300X features a colossal 304 compute units worth 19,456 stream processors capable of AI-relevant math formats, and 192 GB of 8192-bit HBM3 memory, with 5.2 TB/s of memory bandwidth on tap.

NVIDIA's next-generation GeForce RTX 50-series "Blackwell" gaming GPUs are on course to debut toward the end of 2024, with a Moore's Law is Dead report pinning the launch to Q4-2024. This is an easy to predict timeline, as every GeForce RTX generation tends to have 2 years of market presence, with the RTX 40-series "Ada" having debuted in Q4-2022 (October 2022), and the RTX 30-series "Ampere" in late-Q3 2020 (September 2020).

NVIDIA's roadmap for 2024 sees a Q1 debut of the RTX 40-series SUPER, with three high-end SKUs refreshing the upper half of the RTX 40-series. The MLID report goes on to speculate that the generational performance uplift of "Blackwell" over "Ada" will be smaller still, than that of "Ada" over "Ampere." With AI HPC GPUs outselling gaming GPUs by 5:1 in terms of revenues, and AMD rumored to be retreating from the enthusiast segment for its next-gen RDNA4, we get to see why this is the case.

AMD is working hard on delivering next-generation products, and today, its Linux team has submitted a few interesting patches that made a subtle appearance through recent GitHub patches for GFX12 targets, as reported by Phoronix. These patches have introduced two new discrete GPUs into the LLVM compiler for Linux, fueling speculation that these will be the first iterations of the RDNA4 graphics architecture, potentially being a part of the Radeon RX 8000 series of desktop graphics cards. The naming scheme for these new targets, GFX1200 and GFX1201, suggests a continuation of AMD's logical progression through graphics architectures, considering the company's history of associating RDNA1 with GFX10 and following suit with subsequent generations, like RDNA2 was GFX10.2 and RDNA3 was GFX11.

The development of these new GPUs is still in the early stages, indicated by the lack of detailed information about the upcoming graphics ISA or its features within the patches. Currently, the new GFX12 targets are set to be treated akin to GFX11 as the patch notes that "For now they behave identically to GFX11," implying that AMD is keeping the specifics under wraps until closer to release. The patch that defines target names and ELF numbers for new GFX12 targets GFX1200 and GFX1201 is needed in order to enable timely support for AMD ROCm compute stack, the AMDVLK Vulkan driver, and the RadeonSI Gallium3D driver.

AMD in its post Q3-2023 financial results call stated that it expects the Instinct MI300X accelerator to be the fastest product in AMD history to rake in $1 billion in sales. This would be the time it took for a product in its lifecycle to register $1 billion in sales. With the MI300 series, the company hopes to finally break into the AI-driven HPC accelerator market that's dominated by NVIDIA, and at scale. This growth is attributable to two distinct factors. The first of which is that NVIDIA is supply bottlenecked, and customers and looking for alternatives, and finally found a suitable one with the MI300 series; and the second is that with the MI300 series, AMD has finally ironed out the software ecosystem backing the hardware that looks incredible on paper.

It's also worth noting here, that AMD is rumored to be sacrificing its market presence in the enthusiast-class gaming GPU segment with its next-generation, with the goal of maximizing its foundry allocation for HPC accelerators such as the MI300X. HPC accelerators are a significantly higher margin class of products than gaming GPUs such as the Radeon RX 7900 XTX. The RX 7900 XTX and its refresh under the RX 7950 series, are not expected to have a successor in the RDNA4 generation. "We now expect datacenter GPU revenue to be approximately $400 million in the fourth quarter and exceed $2 billion in 2024 as revenue ramps throughout the year," said Dr. Lisa Su, CEO AMD, at the company's earnings call with analysts and investors. "This growth would make MI300 the fastest product to ramp to $1 billion in sales in AMD history."

AMD is rumored to be withdrawing from the enthusiast graphics segment with its next RDNA4 graphics architecture. This means there won't be a successor to its "Navi 31" silicon that competes at the high-end with NVIDIA; but rather one that competes in the performance segment and below. It's possible AMD isn't able to justify the cost of developing high-end GPUs to push enough volumes over the product lifecycle. The company's "Navi 21" GPU benefited from the crypto-currency mining swell, but just like with NVIDIA, the company isn't able to push enough GPUs at the high-end.

With RDNA4, the company will focus on specific segments of the market that sell the most, which would be the x700-series and below. This generation will be essentially similar to the RX 5000 series powered by RDNA1, which did enough to stir things up in NVIDIA's lineup, and trigger the introduction of the RTX 20 SUPER series. The next generation could see RDNA4 square off against NVIDIA's next-generation, and hopefully, Intel's Arc "Battlemage" family.

AMD's next-generation RDNA4 graphics architecture will retain a design-focus on gaming performance, without being drawn into an AI feature-set competition with rival NVIDIA. David Wang, SVP Radeon Technologies Group; and Rick Bergman, EVP of Computing and Graphics Business at AMD; gave an interview to Japanese tech publication 4Gamers, in which they dropped the first hints on the direction which the company's next-generation graphics architecture will take.

While acknowledging NVIDIA's movement in the GPU-accelerated AI space, AMD said that it didn't believe that image processing and performance-upscaling is the best use of the AI-compute resources of the GPU, and that the client segment still hasn't found extensive use of GPU-accelerated AI (or for that matter, even CPU-based AI acceleration). AMD's own image processing tech, FSR, doesn't leverage AI acceleration. Wang said that with the company introducing AI acceleration hardware with its RDNA3 architecture, he hopes that AI is leveraged in improving gameplay—such as procedural world generation, NPCs, bot AI, etc; to add the next level of complexity; rather than spending the hardware resources on image-processing.

AMD in its 2022 Financial Analyst Day presentation claimed that it will repeat the over-50% generational performance/Watt uplift feat with the upcoming RDNA3 graphics architecture. This would be a repeat of the unexpected return to the high-end and enthusiast market-segments of AMD Radeon, thanks to the 50% performance/Watt uplift of the RDNA2 graphics architecture over RDNA. The company also broadly detailed the various new specifications of RDNA3 that make this possible.

To begin with, RDNA3 debuts on the TSMC N5 (5 nm) silicon fabrication node, and will debut a chiplet-based approach that's somewhat analogous to what AMD did with its 2nd Gen EPYC "Rome" and 3rd Gen Ryzen "Matisse" processors. Chiplets packed with the GPU's main number-crunching and 3D rendering machinery will make up chiplets, while the I/O components, such as memory controllers, display controllers, media engines, etc., will sit on a separate die. Scaling up the logic dies will result in a higher segment ASIC.