AMD has begun integrating its next-generation graphics IP into the Linux kernel, releasing a series of eight patches that add GFX13 support to the open-source AMDGPU driver. These patches, submitted to the mailing list on November 9 by AMDGPU maintainer Alex Deucher, contain over 12,600 lines of code, excluding the typically large header files. GFX13 is anticipated to be the graphics core for RDNA 5, although AMD has not officially confirmed this connection. However, since GFX12 has been linked to RDNA 4, it is assumed that GFX13 is the correct codename for RDNA 5.

The patches introduce version 13 of three IP blocks: the "graphics" GFX block itself, the MicroEngine Scheduler (MES), and the IMU. On the compute side, the AMDKFD driver gains initial GC 13 support, a CWSR trap handler for saving and restoring compute wave state, and a temporary change that disables queue reset in the topology. The final patch enables the MGCG, CGCG, and CGLS clock gating features, which reduce power consumption when parts of the GPU are idle. Interestingly, the series mentions two specific targets, GFX 13.0.1 and GFX 13.0.4, indicating that AMD is preparing at least two graphics IP configurations. These version numbers do not correspond to any specific GPU models, so it is unclear which products they relate to. RDNA 5 is expected to be a significant advancement over RDNA 4, necessitating numerous patches to add support.

AMDLSS5, a free mod that runs NVIDIA's DLSS 5 Neural Rendering on AMD's Radeon RX 9070 and RX 9070 XT, has reached version 2.0 with an engine its developer, Wehaveall, built for the project. Earlier builds were based on Daniel Blanco's DLSS-NR-on-AMD, the tool that first got DLSS 5 working on Radeon cards back in September, but version 2.0 no longer needs files from any other mod. In a 60-second Resident Evil Requiem run at 4K with FSR on an RX 9070 XT, the new build averaged 101.2 FPS with 78.5 FPS 1% lows, up from 91.3 and 66.0 FPS on version 1.6.1. That is roughly 11% and 19% better, respectively.

The headline 64% figure comes from a comparison with unnamed "other mods," which managed 61.6 FPS in the same scene. These are the developer's own numbers from one game on one GPU, and there's no mod-off baseline showing what Neural Rendering costs. Game support now includes PRAGMATA, Nioh 3 and GTA 5 Enhanced, the latter only in offline Story Mode with BattlEye disabled. RX 9060 series cards should work but are untested, while RX 7000 support is an alpha that hasn't been tried on real hardware yet.

Leaker Kepler_L2 says AMD will bring RDNA 5+ to its Zen 7 "Grimlock" APUs in 2029. He posted on the NeoGAF forum, replying to a user who guessed RDNA 5 would trickle down to regular mobile chips in 2028 or later. His answer was "RDNA5+ in 2029 with Grimlock." He gave no details on how RDNA 5+ differs from RDNA 5, and AMD hasn't confirmed the name or timing. Grimlock is the codename for the Zen 7 Ryzen generation. This update follows Kepler_L2's January claim that AMD is splitting its APU roadmap in two. Products that don't need strong integrated graphics, like office laptops or laptops with a high-end discrete GPU, stay on RDNA 3.5 until 2029, while "premium" parts get RDNA 5. Medusa Halo is expected to be one of those.

The picture is messy, though. We previously reported that Medusa Point should get RDNA 4m, a tweaked RDNA 3.5 with FSR 4 support, and AMD's HIPRT 3.2 release notes list Medusa support for gfx1170, gfx1171 and gfx115E. The gfx115x family is RDNA 3.5, so it looks like at least one Medusa chip could stick with it. If this ends up being true, anyone waiting on a mainstream Ryzen laptop or handheld with RDNA 5 graphics may be waiting a long time. That's a long run for one iGPU architecture, but of course, this is all leak territory until AMD says something concrete.

It has long been rumored that Sony is working on a PlayStation 6 handheld with a powerful AMD APU codenamed Canis, which has been alleged to feature a combination of Zen 6 CPU cores—potentially with all-new Zen 6 LP cores—and RDNA 5 tech for the iGPU. Now, according to Richard Leadbetter on episode 287 of the Digital Foundry Direct Patreon-exclusive podcast, Sony may be the only gaming handheld to feature this RDNA 5 iGPU. However, it may not be an exclusivity agreement in the traditional sense of the word, since it appears to be as a result of AMD's requirements to build custom silicon for hardware OEMs.

Rumors, albeit disputed rumors, have mentioned that part of the reason Xbox cancelled its own in-house handheld console is because AMD wanted a minimum commitment of 10 million unit sales before it would engineer a custom APU for a console. This may also be why Valve's Steam Deck 2 will seemingly use an off-the-shelf component instead of another custom silicon design. This leaves questions about what CPU and GPU combo will be in the upcoming Steam Deck 2 and Xbox's previously leaked handheld, which now seems like it will be another third-party device, like the ROG Xbox Ally series.

AMD's next-generation gaming graphics cards featuring the RDNA 5 GPU architecture are unlikely to debut before the second half of 2027, according to well-known hardware leaker Kepler_L2. The leaker suggests that AMD is unlikely to release any RDNA 5 desktop graphics cards before Computex 2027, indicating that these GPUs will probably arrive in the latter half of that year. AMD is expected to lead with the silicon codenamed "AT2," which is part of a family of discrete RDNA 5 GPUs. This will be followed by "AT0," "AT3," and "AT4," which are expected to launch a few quarters later in 2028. Since "AT2" likely refers to a mid-range design, AMD might begin its rollout with a mid-range GPU first, followed by flagship and lower-end models in 2028.

Previously, we learned that AMD's RDNA 5 GPUs will incorporate GDDR7 memory with unknown capacities, as GDDR7 is now listed among the supported VRAM types in the AMDGPU driver. Currently, AMD's graphics cards use GDDR6 for the consumer RDNA 4-based Radeon RX 9000 series, while Instinct MI data center and workstation GPUs use HBM3E and HBM4 memory. Since AMD is developing data center GPUs with HBM and consumer cards are the only models using GDDR, the GDDR7 placeholder suggests future support for RDNA 5.

AnandTech forum user SixPie has spotted a previously unknown AMD codename, Gainsborough, in the Adrenalin 26.9.2 driver, and it may be the chip for Valve's Steam Deck 2. Import and export records show test equipment carrying the name was shipped to Asia several times in late July, and that it uses an FF6 BGA package with 1,179 contacts. AMD's partners who know about the Medusa APUs reportedly have no information on it, which would make sense for a custom part.

As for the Steam Deck 2 link, the name is the main clue. Valve's Steam Deck chips have used Final Fantasy VII codenames such as Aerith and Sephiroth before, and Gainsborough is Aerith's surname, as leaker Kepler_L2 pointed out. Separately, Moore's Law Is Dead claims it's a semi-custom APU built on TSMC N3P with Zen 6 and RDNA 5, possibly arriving in late 2027 or 2028, but none of that is confirmed. Moreover, there are reasons to be cautious. Leakers have reportedly tied the same FF6 package to AMD's upcoming Ryzen Z3 handheld chips, so there's conflicting information about who the customer really is.

AMD has submitted patches to its open-source graphics card drivers for Linux, adding support for GDDR7 memory, as anticipation builds for the company's next-generation RDNA 5 GPUs. In the AMDGPU open-source driver, GDDR7 is now listed among the supported VRAM types. Currently, AMD's graphics cards use GDDR6 for the consumer RDNA 4-based Radeon RX 9000 series, while Instinct MI data center and workstation GPUs use HBM3E and HBM4 memory. Since AMD is developing data center GPUs with HBM and consumer cards are the only models using GDDR, the GDDR7 placeholder indicates future support for RDNA 5.

Switching to GDDR7 will allow AMD to significantly boost the memory speeds of its graphics cards and potentially increase capacity, all while improving energy efficiency. The latest GDDR6 memory modules top out at 20 Gb/s, while GDDR7 can achieve 36 Gb/s today and even 48 Gb/s in future versions, according to JEDEC specifications. This could enable AMD's gaming and workstation GPUs with RDNA 5 technology to achieve significantly higher memory bandwidth, enhancing performance in memory-bound workloads.

According to hardware leaker Kepler_L2, known for accurate hardware predictions, the next generation of gaming GPUs might be skipped next year, with a full-stack refresh expected in 2028. Reportedly, NVIDIA's GeForce RTX 60-Series "Rubin" and AMD's RDNA 5-based Radeon RX 10000 series will only be available in 2028, with some select RDNA 5 GPUs from AMD being made available throughout 2027. The specific model is yet to be determined, but it reportedly carries the codename "AT2." Perhaps AMD will roll out this GPU as either a new flagship design and launch the remainder of the stack later on, or simply as a mid-range or low-end SKU to kickstart the RDNA 5 launch. Thus, we still need to wait and gather more information before making conclusions, so treat this only as a rumor for now.

For example, the NVIDIA GeForce RTX 50-Series SUPER refresh had been planned before, but the recent DRAM scarcity has turned the launch into speculation, with some believing it is entirely canceled, while others believe it will be a stopgap solution before RTX 60 "Rubin" arrives. For AMD, we hope to see RDNA 5 finally compete in the flagship SKU sector again and see more powerful designs from the start.

AMD's next-gen RDNA 5 architecture might get its own neural rendering feature, something in the same ballpark as NVIDIA's DLSS 5. That's according to Kepler_L2, a well-known AMD leaker, though it's worth being clear that this is a single-source claim and should be treated as a rumor for now. Posting on the Anandtech forums, Kepler_L2 said AMD will have its own version of "Neural Lighting" comparable to DLSS 5 with its next-gen hardware. Asked whether this would be limited to RDNA 5, the leaker said it's probably going to be the case, since this kind of workload is apparently pretty heavy. That's about as far as the concrete part of the leak goes.

Kepler_L2 also shared thoughts on how AMD's approach should differ from NVIDIA's. He thinks the model should have direct access to raw game data like geometry and depth buffers instead of having to guess at scene details, plus splitting lighting and shading into separate models instead of one pipeline. It's worth stressing that this part is the leaker's own wishlist, and not a description of what AMD is actually building. AMD hasn't confirmed RDNA 5 by name yet, and launch estimates currently range from mid-2027 to early 2028 depending on who you ask. Therefore, even if this pans out, we're probably looking at something that's early in the development pipeline.

According to a well-known leaker, Kepler_L2, AMD is reportedly targeting clock speeds ranging from 3.1 GHz to 3.5 GHz for RDNA 5. The leaker claims to have gained access to AMD's internal materials. This suggests that AMD's new silicon will not only feature significant microarchitectural changes but also offer a raw frequency boost of approximately 4-14% over the factory-rated Radeon RX 9070 XT GPU. The current out-of-the-box boost frequency for this top RDNA 4 GPU SKU is 2.97 GHz, indicating that AMD is planning a substantial increase in clock speeds.

However, it is unclear whether these clock speeds are intended for integrated graphics, dedicated graphics, desktop, or mobile platforms, leaving some questions unanswered. Additionally, RDNA 5 does not appear to be launching soon, with a supposed release date likely in mid-to-late 2027. As of now, we are less than a year away from the launch, it seems. As the launch date approaches, more details will emerge, allowing us to form a clearer picture of what to expect.

Today marks the day we see AMD's next-generation RDNA 5 GPU IP inside working silicon, but not in the form you might expect. Thanks to SemiAnalysis's die annotation of the Exynos 2700 SoC from Samsung, we see AMD's new GPU IP shine in a mobile form with the Xclipse 970 GPU. As earlier leaks reported, AMD and Samsung co-developed the Xclipse 970 GPU using AMD RDNA technology and adapted it for mobile. However, the GPU IP has now been upgraded to the latest RDNA 5, making this SoC technically the first RDNA 5-equipped chip that will be shipping. Built on Samsung Foundry's SF2 node, Samsung has designed this SoC with a unique 10-CPU core configuration and a strong GPU to compete with Qualcomm Snapdragon Elite SoCs.

On the CPU front, Samsung uses Arm's off-the-shelf CPU core IP, which includes C2-Ultra and C2-Pro. This setup features a single C2-Ultra big core running at 4.24 GHz, one C2-Ultra running at 3.36 GHz, four C2-Pro cores running at 3.74 GHz, and four C2-Pro cores running at 2.88 GHz, making a total of 10 CPU cores. On the GPU side, the Xclipse 970 packs eight GPU WGPs, each containing two Compute Units. This results in 16 CUs on the GPU front if everything remains the same with RDNA 5 and no changes to the WGP/CU ratio have been made by AMD since the last-generation RDNA 4.

AMD has released the latest set of patches for its RDNA 5 GPU IP, initiating the enablement of the Display Core Next (DCN) 6 display engine. Although the launch of AMD's next-generation RDNA 5 graphics cards is still a few months away, Linux support is already being developed to ensure every part of the stack is operational before the official release. According to the latest patch noted by Phoronix, the DCN6 display engine is associated with the AMD GFX13, which corresponds to the RDNA 5 GPU family. AMD's DCN display engine handles a variety of functions, including display output, image scaling, monitor ratio output, pixel manipulation, and much more. This engine is a highly complex component crucial for graphics output in AMD RDNA graphics cards.

For this DCN6 version, AMD has introduced new scaling integrity and costing surfaces, a UPSP pre-scaler to sample 4:2:0 and 4:2:2 content before scaling, and new pixel formats within the display logic. For the final GPU, this means that modern high-refresh and high-resolution gaming will be possible across a range of aspect ratios and configurations, with content being delivered smoothly. However, the final implementation will ultimately depend on what AMD includes in its actual GPU silicon.

Recent rumors revealed that Microsoft may be ditching APU customization for Xbox Project Helix, its upcoming PC-console hybrid that is expected to launch around the $1,000 mark, opting instead for an off-the-shelf AMD APU that will seemingly be available to other manufacturers as well. Now, per a new rumor by ubiquitous leaker Moore's Law Is Dead on the Broken Silicon podcast, it seems as though that high price may come with a significant performance boost.

Moore's Law Is DeadWhat's even more of a cause for optimism, though, is how much better PS6 and Helix will be. Helix is basically a high-end PC, even using the same silicon as AMD's 70-class or maybe 80-class RDNA 5 GPUs. It isn't an ultra-level of PC hardware, but still far more powerful than consoles usually have. It will have the biggest APU in console history... Yes it will be expensive, but guys, if it's even $1,200, it's like a $2-3,000 gaming PC. That's disruptive.

We've known for a while now that the upcoming Xbox Project Helix game console will be a hybrid game console, able to play both PC and Xbox games, but recently news broke that the Xbox Helix would be using an off-the-shelf AMD APU, followed by a leak from KeplerL2 on the NeoGAF forums that it would be a third-party console. The latter rumor claimed that the Xbox Helix would not be available directly to consumers from Microsoft itself, instead adopting something of a Steam Machine model—the original Steam Machine, not the 2026 reboot—and making the hardware design available to hardware partners similarly to how AMD and NVIDIA sell their GPUs to board partners who package them in their own way.

Xbox Vice President, Jason Ronald, chimed in with a response on X, saying simply that "Project Helix will be available as a 1st party Xbox console," and nothing more. While this somewhat debunks the rumor started by KeplerL2, it also leaves the door open to the possibility of third-party Helix consoles. This idea somewhat holds water, given the context of the previous rumor about the off-the-shelf APU. If Microsoft did not license the Xbox Helix design, and brands could simply buy the APU directly from AMD to make their own hardware platforms, similarly to what's been happening with other Radeon 780M-powered APUs that have been popping up in Windows gaming handheld devices since the Ryzen Z1 Extreme handhelds launched.

Valve's next-generation Steam Deck 2 handheld console is reportedly planned for release in 2028, with significant manufacturing changes expected for this sequel to the highly successful handheld gaming device. According to a well-known industry leaker, KeplerL2, posting in the NeoGAF community, Valve is targeting a 2028 refresh for the second-generation Steam Deck. However, the ongoing supply chain shortages of DRAM and NAND Flash could cause disruptions to these plans, potentially leading to delays. Interestingly, this period is when the shortages are expected to start easing, so the Steam Deck 2 could still be released on time, depending on Valve's sourcing capabilities.

One of the most significant procurement shifts for the Steam Deck 2 is Valve's choice of the computing base that will power the handheld. Instead of using a semi-custom AMD APU, Valve is expected to use an off-the-shelf AMD APU that won't require any custom tuning from AMD to meet Valve's needs. This is welcome news, as the latest Steam Machine showed that Valve's reliance on a semi-custom APU solution made the hardware "obsolete" quickly while the rest of the industry advanced. With any semi-custom solution, stockpiling silicon and waiting for DRAM/NAND modules to arrive puts pressure on Valve to ship a product that is significantly underpowered or too expensive. However, with an off-the-shelf solution, Valve could use the best available option at the time of shipping and optimize SteamOS around it.

AMD's RDNA 4m graphics might be the company's most mysterious GPU IP, as they are reportedly rebranding some of their RDNA 3.5 IP to fit INT8 data types and support FSR 4 technology. We previously reported that AMD designated the GFX1170 target for RDNA 4m. However, in the latest merge request for the LLVM compiler, AMD added two new software IDs: GFX1171 and GFX1172 GPUs. These targets are not true RDNA 4 GPUs, which belong to a GFX12 branch, but rather extensions of RDNA 3. What was thought to be RDNA 3.5 has now evolved into RDNA 4m, which will power AMD's Ryzen 500 "Medusa Point" series of APUs. With RDNA 3.5 / RDNA 4m expected to be used by AMD until 2029, it makes sense for AMD to adapt RDNA 3.5 into RDNA 4m with support for FSR 4 upscaling technology.

In contrast, "Medusa Halo" will utilize AMD's next-generation RDNA 5 / UDNA GPU microarchitecture. "Medusa Point" will introduce a desktop-exclusive RDNA 4 with the new RDNA 4m variant. Although we initially lack comparisons between the two, some instruction set extensions, such as WMMA and SWMMAC instructions, indicate support in the new "GFX1170" GPU, which should be associated with the GFX11 generation, also known as RDNA 3. Currently, this is believed to be an upgraded RDNA 3 with many RDNA 4 modules, enabling FSR 4 support even on the less powerful "Medusa Point" APU.

A fresh leak from Geekbench has surfaced listing an unannounced AMD processor with the OPN (Ordering Part Number) code 100-000001713-31. As for the platform, that is listed as Plum-MDS1, a somewhat direct hint to Medusa Point, AMD's next-gen mobile APU family based on Zen 6. The chip is a 10-core, 20-thread with a 2.4 GHz base clock, though actual test results show it running closer to 1.3-2 GHz, unsurprising for an engineering sample this early in development. Each core gets 1 MB of L2 cache, and the L3 comes in at 32 MB, up from 24 MB on Strix Point and 16 MB on Hawk Point. This is 50% more cache than the current 10-core parts like the Ryzen AI 9 365. The test system had 32 GB of memory installed, type unspecified. Benchmark numbers aren't worth reading into at this stage, as the chip spent most of the test hovering around 1.39 GHz and barely peeked above 2 GHz, well below what a final retail part on a 3 nm process would run at. Too early to draw any conclusions from the scores.

Shipping manifests from Planet3DNow linked to the "Medusa" codename suggest a 4C+4D layout, standard and density-optimized cores, though that doesn't exactly explain the 10-core count seen in Geekbench. One theory is that two additional low-power cores sit in the IO die, bringing the total to ten. The part is expected to be a 28 W TDP mobile chip for the FP10 socket. Medusa Point is shaping up to combine Zen 6 CPU cores with a mix of RDNA 5 and RDNA 3.5 graphics, plus an updated NPU. A launch around CES 2027 fits the AMD usual cadence, meaning there is likely a long wait ahead; however, this Geekbench entry confirms that testing is already happening, whether at AMD itself or at one of its early hardware partners.

AMD is refining its RDNA 5 architecture, which is likely in its final stages of design. Thanks to a submission to the LLVM compiler, we are learning that AMD's upcoming RDNA 5/UDNA architecture features architectural changes that will enhance compute utilization, resulting in significantly higher game shader performance. The codename for RDNA 5 is GFX1310, and it now implements a full Dual-Issue VALU pipeline for Wave32. This allows vector operations (VOPs) to be issued simultaneously to the GPU's X and Y arithmetic logic unit (ALU) lanes. The new design expands the range of fused multiply-add (FMA) and other VOP instructions eligible for dual-issue and relaxes some register constraints, enabling compilers and shader code to perform more intensive floating-point work in Wave32 mode. This means the FP32 compute utilization of RDNA 5 can be much higher than before, greatly benefiting applications that are FP32-heavy, which means gamers are about to see a significant performance boost.

AMD originally implemented Dual Issue VALU in the RDNA 3 architecture with the Radeon RX 7000 series graphics cards. However, the pipeline was not fully functional, as the dual-issue implementation supported only a limited subset of VOP instructions, excluded several important FMA variants, required Wave32 mode and strict register-bank separation, and was often bypassed by compilers and not fully exposed to drivers. As a result, many shaders could not exploit X/Y pairing, and measured FP32 throughput often fell well below the hardware's theoretical peak. This was detrimental to performance and reduced compute capability, especially in applications that rely heavily on FP32 compute. This is particularly true for modern games that process vertex and pixel shaders primarily using FP32 compute. With the new RDNA 5 design, gaming performance should align more closely with theoretical compute performance.

AMD is reportedly developing a next-generation FSR update, codenamed "FSR Diamond," for the upcoming Xbox project "Helix" gaming console. With the Project Helix console expected to launch in 2027, the details of "FSR Diamond" remain a mystery. It's unclear what AMD aims to achieve with this AI-powered video generation technology, but it likely builds on previous advancements like Radiance Caching and Ray Generation, adding a new dimension to the company's graphics capabilities. We might see an AMD equivalent of the multi-frame generation technology found in NVIDIA's GeForce RTX 50-series GPUs and Intel's XeSS 3.0. Since Project Helix is anticipated to feature RDNA 5 / UDNA graphics IP, this feature could be exclusive to that generation, as AMD tends to tie its latest technologies to the current RDNA IP.

AMD already differentiates its latest FSR "Redstone" suite of technologies, with features like Ray Regeneration and Radiance Caching exclusive to RDNA 4 hardware in the Radeon RX 9000 series of GPUs. Other basic technologies, such as upscaling and frame generation, are supported on older RDNA 3/2/1 generations but use an FSR 3.1 fallback, with no FSR 4 support currently available. However, since INT8-based FSR 4 exists, it may only be a matter of time before the company extends this capability to older GPUs, though the expected performance might not be optimal. For multi-frame generation and potentially dynamic multi-frame generation, "FSR Diamond" would need specialized hardware. Even NVIDIA, with its MFG 6x mode and Dynamic MFG, keeps those features exclusive to the GeForce RTX 50-Series "Blackwell," which uses hardware flip-metering available only on the newest GPU generation. Similarly, RDNA 5 / UDNA could incorporate these hardware components as well.

Given the current state of the gaming industry, rumors have started to emerge about the next-gen gaming console launches, with previous rumors claiming that the PlayStation 6 would launch alongside a new standalone PlayStation Portable gaming handheld, and that the PS6 would have less graphical processing power than the upcoming Xbox console. In a new post on YouTube, ubiquitous leaker and industry insider, Moore's Law is Dead, claims that the PS6 Orion—the living room console—will feature 2.5-3× the rasterization performance of the PS5 and 6-12× the ray tracing performance of the PS5, or 3-6× faster ray tracing than the PS5 Pro and roughly twice the raster performance of the PS5 Pro. This is thanks to a significantly faster RDNA 5 GPU with 52-54 CUs running at around 2-3.6 GHz and delivering around 34-40 TFLOPS of theoretical performance.

The leaker also shot down rumors that recently surfaced claiming that the PlayStation 6 may be delayed to as late as 2029 due to the ongoing memory crisis, stating that the nature of memory and APU fabrication contracts make it unlikely that the PS6 will be delayed that long. Instead, we may see a brief period of increased pricing or scarcity at launch. Apparently, according to his sources, Sony has a contract with TSMC to start mass-producing the PS6 as early as Q2 2027. MLID also mentions the AMD "Canis" APU slated to arrive in the PS6 handheld console, claiming that it will feature four Zen 6c cores, two Zen 6 LP cores (for running the operating system), and 16 RDNA 5 CUs running at 1.6-2 GHz with a 15 W total board power, all being fed by LPDDR5X memory over a 192-bit bus. All of this will allegedly be targeting 1080p gameplay with a significantly higher power limit when docked. He also speculates that the handheld will feature "vastly better" ray tracing than the PS5.

AMD's next-generation Ryzen 500 series of APUs, which include the "Medusa Point" and "Medusa Halo" SoCs, are slowly taking shape, and we are learning more about the integrated graphics of "Medusa Point." According to the latest compiler patches, "Medusa Point" will feature a variant of RDNA 4 called RDNA 4m, likely a mobile version designed for laptops and mini PCs. In contrast, the "Medusa Halo" will utilize AMD's next-generation RDNA 5 / UDNA GPU microarchitecture. "Medusa Point" will introduce a desktop-exclusive RDNA 4 with the new RDNA 4m variant. Although we initially lack comparisons between the two, some instruction set extensions, such as WMMA and SWMMAC instructions, indicate support in the new "GFX1170" GPU, which should be associated with the GFX11 generation, also known as RDNA 3. Currently, this is believed to be an upgraded RDNA 3 with many RDNA 4 modules, enabling FSR 4 support even on the less powerful "Medusa Point" APU.

When it comes to the overall system configuration, AMD targets "Zen 6" CPU cores for both "Medusa" variants. Both "Point" and "Halo" versions will likely differentiate with the same factors we have in the current generation APUs, which is a mix of AMD's "Zen 6" and "Zen 6c" CPU cores for the "Medusa Point," while the more powerful "Medusa Halo" will feature a uniform "Zen 6" core cluster for maximum CPU performance on that side as well. Recently, we learned that the platform will adopt LPDDR6 memory, which will increase the memory bandwidth by 50% over the standard LPDDR5X memory we see implemented today.

It's been widely reported that the next-gen Sony PlayStation 6 is in the works, with AMD once again in charge of the APU that will power gaming experiences, even if rumors suggest that the next-gen console launch may be pushed back as far as 2029. It's also common knowledge by now that RDNA 5 will power the PS6, but a new rumor that originated from prolific leaker, KeplerL2 on the NeoGAF forum, suggest that the architecture in the PS6 will not be the full-fledged RDNA 5.

The leaker does not expand on this further, other than to remind that the PS5 doesn't use the full RDNA 2 architecture either, but consoles have always used custom APUs, so the idea that the PS6 won't use the full RDNA 5 feature stack isn't unreasonable. Some have postulated that one feature that wouldn't make the cut is the NPU, since that may not get used in a modern console. It's also possible that certain productivity features or video encoders, although even a lot of the AI improvements in RDNA 4 have been marketed as gaming upgrades.

The next major refresh of AMD's Ryzen AI MAX APUs is still far away, but now we are putting together the pieces of the "Medusa Halo" APU puzzle. According to a famous leaker, @Olrak29_ on X, AMD's next-generation "Medusa Halo" APU will be complemented by LPDDR6 memory. This is one of the first LPDDR6 memory SoCs we are learning about, making it unique. Based on previous rumors, the silicon could have a 384-bit bus powering LPDDR6 memory, which would translate into massive bandwidth powering the SoC's new CPU and GPU configuration. This includes up to 24 "Zen 6" CPU cores and 48 RDNA 5/UDNA compute units for the GPU configuration. Paired with the added bandwidth from LPDDR6 memory—which these APUs greatly benefit from—"Medusa Halo" will be one of the best-performing SoCs when it launches.

Interestingly, memory manufacturers like Samsung and Innosilicon are already supplying LPDDR6 modules to customers for validation. Innosilicon's LPDDR6 modules boast an impressive speed of 14.4 Gbps, significantly faster than Samsung's initial modules, which achieve 10.7 Gbps. Innosilicon's modules offer a 1.5x increase in IO speed capability compared to the 9.6 Gbps of LPDDR5X previously available, along with improved efficiency. The latest LPDDR6 also increases the number of bits per byte of IO from 8 to 12. This results in LPDDR6's bandwidth at a single-channel 24-bit I/O speed being double that of LPDDR5X at a 16-bit single-channel. The company is reportedly collaborating with TSMC and Samsung to ensure sufficient production capacity for LPDDR6 IP, while Samsung relies on its own fabs for manufacturing memory.

Microsoft's next-generation Xbox console is reportedly taking an unconventional route by running on a customized version Windows 11 OS instead of the specialized console OS that typically powers these devices. According to Windows Central, the system will function essentially as a gaming PC that boots into an Xbox interface by default. This UI is likely similar to what the current Xbox Full Screen Experience looks and feels like, and will likely provide the same performance boost. We have already seen that Xbox FSE mode brings about a 9.3% reduction in RAM usage and about an 8.6% higher FPS due to the smaller system overhead. Users could exit that interface to access the full Windows 11 operating system, meaning the hardware would support Steam, EPIC, and other competing game stores, as well as standard PC applications alongside Xbox games.

This is Microsoft's first radical departure from the walled-garden approach that has defined console gaming for decades. What it could translate to is the first hybrid system that serves multiple purposes, from traditional gaming to running productivity suites of Microsoft 365 apps like Word, Excel, and others, all from the same system. Teams from the Windows and Xbox divisions are reportedly collaborating closely to adapt the operating system for living room use. Microsoft is also working with hardware partners like ASUS to create multiple devices at different price points rather than releasing a single standard console. Plans for a first-party handheld device are still under consideration, though the traditional console appears to be the main focus.

AMD posted its record fourth quarter revenue of $10.3 billion in 2025, and during the earnings call, the company issued some guidance on the upcoming product portfolio. During the call, AMD confirmed that Valve's Steam Machine is on track and shipping early this year, while its custom SoC division that designs processors for PlayStation and Xbox consoles will deliver an RDNA 5-based SoC for the next-generation Xbox console. While the Steam Machine specifications are confirmed, Xbox "Magnus" SoC is still largely a collection of rumored specifications. The "Magnus" SoC is rumored to feature the largest APU ever designed for a consumer console, with a 408 mm² chiplet design. Of this, 144 mm² is dedicated to the SoC built on TSMC's N3P node, while the GPU occupies 264 mm². The AMD chip is expected to include up to 11 CPU cores—three Zen 6 and eight Zen 6c—alongside a substantial GPU setup with 68 RDNA 5 compute units, four shader engines, and at least 24 MB of L2 cache. Memory might expand to 48 GB of GDDR7 on a 192-bit bus. A dedicated NPU is rumored to offer significant on-device AI performance, with reports suggesting up to 110 TOPS.

Dr. Lisa Su, AMDFor 2026, we expect semi-custom SoC annual revenue to decline by a significant double-digit percentage as we enter the seventh year of what has been a very strong console cycle. From a product standpoint, Valve is on track to begin shipping its AMD-powered Steam Machine early this year, and development of Microsoft's next-gen Xbox featuring an AMD semi-custom SoC is progressing well to support a launch in 2027.