Intel Foundry has been delivering some interesting research lately, and the newest installment is the world's thinnest gallium nitride (GaN) chiplet, with the base silicon measuring just 19 micrometers in thickness. The company manufactured a 300-millimeter GaN-on-silicon wafer, which is one of the first combinations of GaN with traditional silicon logic on a single chip. In traditional manufacturing setups, GaN is reserved for power electronics and is not mixed with silicon-based computational logic. However, Intel has managed to combine GaN semiconductors for power delivery with silicon-based compute logic, meaning that power chips can now perform basic computations and actions on their own, without needing a separate chip to dictate behavior.

Intel manufactured the GaN wafer with silicon combined on the 30 nm process, which reportedly demonstrated excellent properties. This includes stable current carrying, very low power loss, and the ability to block voltages up to 78 V without leakage. What is fascinating is that GaN is usually used for wide bandgap applications, with high radio frequency performance exceeding 300 GHz. GaN also provides much better power delivery due to its superior material properties and operates more stably at higher temperatures. This is especially significant for workloads like electric vehicles or data center point-of-load delivery, where space is limited, temperatures can reach about 150°C, and stability is crucial.

NVIDIA's upcoming N1 SoC, designed for gaming laptops, has officially appeared for the first time alongside its motherboard intended for a laptop. Featuring 128 GB of LPDDR5X memory running at 8,533 MT/s, the SoC is paired with SK hynix modules. NVIDIA's OEM partners for this launch, including Dell and Lenovo, have confirmed that the new NVIDIA laptop platform will be available this year, and we have seen early benchmark submissions of the N1 SoC. As NVIDIA prepares to enter the gaming laptop market, the company initially released its GB10 SoC for the DGX Spark mini workstation. This chip allowed NVIDIA to test its collaboration with MediaTek on developing new low-power SoCs that offer unusually high-performance graphics for AI applications, as well as very high RAM capacity.

In the latest images that appeared on Goofish, we see a laptop motherboard equipped with the NVIDIA N1 SoC and 128 GB of LPDDR5X memory closely packed in a specialized PCB area. This area might be designated for the cooler, which could potentially also cool the LPDDR5X memory modules. NVIDIA plans to introduce its N1/N1X SoCs with Arm-based CPU cores around Computex 2026. These will offer consumers 20 CPU cores, consisting of 10 Cortex-X925 and 10 Cortex-A725 CPU cores based on the Armv9.2 ISA, along with a "Blackwell" GPU optimized for low power settings with up to 6,144 CUDA cores in the higher-end form. You can also see the port selection on the side of the PCB, which seems to be USB-C or Thunderbolt 4/5, USB-A, and an HDMI connection. Since we don't know which laptop model this is, we have to wait a bit more before getting final details.

In the latest series of leaks about Intel's upcoming "Nova Lake" chips, we learn that the desktop "Nova Lake-S" designs will feature a new socket mounting and independent loading mechanism (ILM). According to exclusive information from VideoCardz, the NVL-S desktop designs on the LGA-1954 socket will include a new 2L-ILM socket mechanism to improve contact with CPU coolers. Given that Intel's upcoming NVL-S desktop processors will have up to 52 high-frequency, power-hungry cores, a better mounting mechanism is definitely needed. This 2L-ILM is a two-level independent loading mechanism, using one lever on each side of the CPU to secure it into the LGA-1954 socket. Applying pressure on both sides of the socket ensures improved flatness while the locking mechanism does its job.

Intel categorizes its desktop sockets in various forms. One of the most basic is the Default-ILM, along with the RL-ILM. As seen with "Arrow Lake," Intel's motherboards divide the socket locking mechanism into the ILM and RL-ILM designs, applied depending on the sector. For the lower-end sector, the Default-ILM socket is used, while the RL-ILM is applied to higher-end overclocking motherboards, providing better pressure and ensuring a flatter surface contact for a cooler. Noctua and Cooler Master already differentiate these two in their CPU coolers, but the mounting hardware is generally the same for both. The RL-ILM is simply a better design for overclocking due to the improved surface contact it provides.

Apple's MacBook Neo, MacBook Air, and the latest MacBook Neo series of laptops are now susceptible to a significant performance drop if they remain powered on for 49.7 days without a reboot. Specifically, after exactly 49 days, 17 hours, 2 minutes, and 47.296 seconds of continuous operation, the entire TCP/IP networking stack shuts down and stops establishing new connections. This unusual bug stems from a kernel-level issue in Apple's XNU kernel involving tcp_now, a 32-bit unsigned integer that tracks elapsed milliseconds since boot for the TCP stack. When it reaches its maximum value of 4,294,967,295, the counter stalls instead of resetting, disrupting the arithmetic macOS uses to expire and discard closed connections.

These inactive connection sessions accumulate quickly, increasing CPU usage as the system struggles to handle an ever-growing backlog. Once the available port pool, typically around 16,384 ports, is fully used, no new connections can be opened. Applications that rely on network access begin to fail, although already-established sessions continue to function, and the machine still responds to pings, albeit much more slowly. This initially delayed the diagnosis of the bug. A reboot resets the counter and restores normal operation, but the issue will recur on the same schedule after another 49.7 days of continuous use. There are potential solutions to fix the bug by using larger 64-bit integers, but this would require more extensive code changes, performance testing, and additional work compared to a simple fix.

Valve seems to be incorporating more AI integration into its Steam gaming platform to assist gamers with various aspects of their experience. In the latest development of this AI integration, Valve is reportedly developing the SteamGPT AI system to assist with customer support queries, while also integrating many additional features into the system. As you know, Valve's Steam platform receives thousands of support questions daily, ranging from refunds and platform issues to payment processing problems and many other inquiries. Valve's support staff is often overwhelmed, especially during major sales events. If Valve creates a customized AI system for chatting, support, and other infrastructure tasks, the company could alleviate a significant portion of the daily issues.

Additionally, recent source code leaks mention some connection to Valve's Trust systems, which enhance matchmaking quality by grouping players with similar levels of trustworthiness in games like Counter-Strike 2. This is an algorithmic process where an AI system could improve grading, as AI can naturally solve these tasks by grouping players more effectively than any custom algorithm. Furthermore, it could also detect cheating patterns performed by players and activate the anti-cheat system. However, while an AI system can assist with customer support queries, it may still make errors, necessitating human oversight to ensure the validity of support resolutions.

Intel has released its latest Arc GPU Graphics Drivers, version 101.8629 WHQL, introducing the gaming world to its newest Arc Pro B70 and Arc Pro B65 discrete GPUs. These new WHQL drivers primarily focus on providing gaming support for these professional-grade GPUs, which means the drivers include minimal game optimizations and retain some existing issues. Since no extensive optimization was necessary, the recently announced BMG-G31-based GPUs only required a simple driver enablement rather than the lengthy optimizations typically needed. The "Battlemage" architecture has been around for a while, so previous optimizations have already been implemented. The performance of the new GPUs is now achieved by linearly scaling the Xe2 cores within the GPU. When workstations take a break from their usual AI development or inference tasks on Arc Pro B70 GPUs, they can seamlessly transition to gaming. Interestingly, with gaming now an option, the Arc Pro B70 becomes Intel's largest discrete GPU for gaming tasks.DOWNLOAD: Intel Arc Graphics Driver 101.8629 WHQL

Yesterday, Qualcomm introduced its Snapdragon X2 Elite series of chips to consumers in partnership with major OEMs like ASUS and HP. On the same day, ASUS released additional laptop models based on Intel and AMD SoCs. However, just a day after some media outlets published their reviews, ASUS increased the prices of its products through its launch partner distributor, Best Buy, significantly affecting some reviewers' conclusions. Reportedly, ASUS raised prices by $100 to $350, depending on the laptop model and configuration, regardless of the platform. This is a substantial change; for example, a $350 increase on a $1,000 model like the Zenbook 14 represents a one-third increase in the device's advertised price. Reviewers, such as those at TechPowerUp, evaluate the model based on its value. If a model is aimed at budget users, the perspective of a budget user is applied, and if a model is aimed at workstation users, a small price increase may not be a dealbreaker. However, in the $1,000 range for a laptop, ASUS is now pushing these laptops into a completely different category.Below is the price change, broken down by model. On the left is the launch pricing listed by ASUS, and on the right is the new pricing from Best Buy as of today.

Late last week, we reported on a new series of rowhammer bit-flip attacks targeting GDDR6-based NVIDIA GPUs. Most of these attacks can be mitigated by enabling IOMMU through the BIOS, which restricts the memory regions the GPU can access on the host system, thereby closing the primary attack path. However, researchers from the University of Toronto have introduced "GPUBreach," which can bypass IOMMU and enable CPU-side privilege escalation, unlike the previous "GDDRHammer" and "GeForge" attacks. In most typical server, workstation, and even PC configurations, IOMMU restricts the GPU's access to the CPU's physical addresses, preventing direct memory access. These are the typical DMA-based attacks that the Input-Output Memory Management Unit protects users from. However, the new "GPUBreach" operates differently.

For example, "GPUBreach" exploits memory-safe bugs in the actual GPU driver and corrupts them. When IOMMU confines the GPU's direct memory access to driver-assigned buffers, the new exploit corrupts metadata within these permitted buffers. This causes the driver, which has kernel privileges enabled on the CPU host, to perform out-of-band writes to the buffer, effectively bypassing any protection IOMMU can offer. This logic is built into the kernel by default, as the GPU driver is one of the most trusted components of the operating system. Hence, IOMMU bypass is possible when the metadata is corrupted. Since "GPUBreach" grants an attacker full root privilege escalation, the attack differs significantly from previous rowhammer attacks.

Intel has officially announced its participation in Elon Musk's "Terafab" project, which aims to reimagine chip manufacturing. Specifically, Intel Foundry plans to join this ambitious initiative, leveraging its significant manufacturing capabilities as one of the strategically important companies in the U.S. However, the specifics of Intel's involvement remain unclear, as it is not yet known how Intel will officially contribute to the Terafab project. The company has stated that since Terafab aims to produce 1 terawatt per year of compute power for AI and robots serving xAI, SpaceX, and Tesla. Intel will assist in designing silicon, manufacturing it, and providing some of the world's most advanced packaging technologies, such as EMIB. It is likely that some of Intel's facilities, which are currently being expanded, will become part of the network needed for the Terafab project, while the Terafab facility itself conducts custom work guided by Intel.

The goal of Terafab is to consolidate the entire chip manufacturing process under one roof. The plant is expected to integrate several stages of semiconductor production at a single site, including logic fabrication, memory, packaging, testing, and mask production. This setup is unusual, as these steps are typically spread across multiple specialized facilities and companies. The original idea behind Terafab is that consolidating these processes could accelerate development by enabling engineers to design, test, and revise chips with fewer delays, essentially allowing for rapid prototyping. This contrasts with the traditional, lengthy process of manufacturing chips at one site, packaging them at another, and testing them in-house. Elon Musk visited Intel's CEO Lip-Bu Tan over the past weekend securing a deal.

Microsoft has been addressing the recent wave of "Microslop" criticism that has emerged online in response to the forced integration of AI into its products. Specifically, Microsoft has been promoting its Copilot applications, products, and even Copilot-branded hardware like Copilot+ AI PCs to consumers. However, this is just the scratching the surface, as the actual number of Copilot variants is much higher than what the average PC enthusiast might consider. If you've ever wondered how many Copilot applications exist, the official count stands at 80 Copilot applications, products, services, and hardware that the Redmond giant has developed. Across every Microsoft vertical, there is a Copilot icon in some form, even present on Copilot+ PCs with its own dedicated Copilot key. This represents the biggest branding overhaul in Microsoft's history, as the company traditionally distinguished products with unique features and names.

However, the popularity of its ecosystem is at an all-time low, particularly within the PC community, which interacts most with the Windows 11 operating system and the Microsoft 365 suite of applications, formerly known as the Office package, including Excel, Word, PowerPoint, and others. Regular consumers are largely unaware of the extent of the Copilot branding, as Microsoft has extended its AI narrative to consumer and business chatbots, developer tools, desktop applications, Copilot applications within other applications, enterprise platforms, hardware, and business software serving the enterprise sector. At some point, the community narrative suggests that the branding is being pushed a bit too aggressively, as Windows 11 users, who interact daily with the world's most widely distributed operating system, have openly discussed the drawbacks of the forced Copilot integration.

Intel's upcoming "Nova Lake" CPU generation, part of the Core Ultra 400 series, will be a major refresh of the company's P and E-core hybrid design. While many specifications have been largely leaked, the exact integrated GPU configuration remained a mystery until now. One of the most reliable Intel leakers, Jaykihn, has revealed that Intel plans to use the Xe3 generation of graphics, which is found in the current "Panther Lake" Arc B300 series of integrated GPUs. The display and media engine will come from the Xe3P "Crescent Island." Previously, we reported the source's claim that the "Nova Lake" display and media engine would incorporate some IP elements from the Xe4 "Druid" generation of graphics. However, the actual underlying hardware is not related to Xe4 and instead borrows IP from Xe3P.

Intel's plans for "Nova Lake" are focused on late 2026, with the entire lineup expected to roll out in early 2027. The platform will support DDR5 memory at 8,000 MT/s out of the box, without any overclocking. This indicates an improved integrated memory controller on the Nova Lake platform, which seems ready to handle those speeds even before XMP or factory-overclocked modules are considered. It also suggests that Intel is pushing memory support further than its current controller, which reaches DDR5-7200 on the current "Arrow Lake Refresh," alongside the new core IP and updated configuration.

Steam is reportedly in the process of adding a "Frame Estimator" tool that can estimate your PC's performance before you purchase a game. As you know, Valve's Steam platform is the largest gaming platform in the world, with access to millions of PCs. The Steam Client application offers an option to include your PC in Valve's telemetry system, which processes data such as your PC's specifications and game information, including your library. Using these data points, Steam will estimate how many frames per second your PC can generate in any game, depending on your configuration. For example, for a specific CPU, GPU, and available system memory, the Steam Client will indicate whether a game can reach 60 FPS at 1440p using high settings, or whatever your preference is. We can only speculate at this point about what the feature will look like, as Steam is still refining it before the public beta release.

Additionally, Valve has already started asking users for anonymous FPS data collection about a month ago whenever they run a game. With this data pool, likely involving millions of participants, Valve aims to build a system that estimates your FPS output based on your specific PC configuration, without needing to run a game first. Reportedly, this feature will appear in the Steam Client and show how much performance your PC can deliver before you even purchase a game. This is a classic recommendation system that will indicate what your configuration typically delivers at specific game settings and resolutions.

Intel has recently released a new video showcasing its latest Texture Set Neural Compression (TSNC) technology, which delivers textures up to 18 times smaller while maintaining visual quality with little to no noticeable difference compared to the industry-standard compression. Using AI-based neural networks, Intel's graphics team processes input data from industry-standard BCn textures. These textures are compressed through an AI model encoder, encoded in the latest space values, and then decoded by a network decoder to decompress the textures. The result is output data textures that are up to 18 times smaller, with some quality loss at maximum compression settings. As with any neural technology, TSNC is trained on millions of standardized textures to create an AI model that can replace traditionally compressed textures in the BCn format. This results in new, much smaller game textures that load faster, use less VRAM, and perform better thanks to modern GPU technology.

There are several ways to apply TSNC neural compression, depending on the desired outcome, whether it's saving game installation size, reducing VRAM usage, or improving performance. Variant A, as Intel calls it, can achieve up to 9 times compression of the standard texture set, with little to no difference in visual quality—almost an unnoticeable drop. However, when the goal is maximum efficiency and requires up to 18 times texture compression, Intel offers Variant B of the TSNC neural network. This variant provides a significant performance boost, with the trade-off being a modest visual change. Using NVIDIA's FLIP tool to measure quality drop in generated images, Intel notes that Variant A experiences a 5% visual quality drop, while Variant B sees up to a 7% quality drop, which is noticeably more.You can judge for yourself by viewing the comparison images below.

Intel has unexpectedly discontinued the official XeSS plugin for the Unity game engine, leaving the Unity ecosystem without XeSS frame generation, temporal super sampling, and antialiasing technology. This decision comes just a month after Intel released its official XeSS 3.0 software development kit for game studios, which includes features like multi-frame generation and the ability for XeSS 3.0 to use external memory heaps for GPU memory allocated by the game engine. This allows XeSS and the engine to operate on the same VRAM blocks instead of each reserving separate ones. However, it is unclear if XeSS 3.0 works with the latest Unity 6 engine, as official support has been withdrawn and the repository now serves as a public archive on GitHub. Similarly, AMD abandoned the Unity platform years ago, leaving only FSR 2.0 support since the last update. The focus now seems to be on other game engines like Unreal Engine 5 and its future versions, which are receiving all the latest advancements from both Intel and AMD.

Intel on GitHubIntel will not provide or guarantee development of or support for this project, including but not limited to, maintenance, bug fixes, new releases or updates. Patches to this project are no longer accepted by Intel. If you have an ongoing need to use this project, are interested in independently developing it, or would like to maintain patches for the community, please create your own fork of the project.

Microsoft is now force-installing the latest version of Windows 11, the 25H2, for users of the older 24H2 version of the operating system. Through what Microsoft calls a "machine learning-based intelligent rollout," the company is automatically updating 24H2 users to Windows 11 25H2 when their devices are ready. This forced update is part of a staged update plan for Windows 11, where Microsoft is gradually updating systems worldwide to phase out the 24H2 version, as support for this operating system will officially end on October 13, 2026. It makes sense to upgrade users to an operating system with a longer support period. However, Windows 11 users might be frustrated with the constant forced upgrades that Microsoft is pushing, which seem to offer little real-world benefit beyond regular security maintenance.

Regarding Windows versions, the situation has been a bit confusing recently. Microsoft has dedicated the 26H1 update exclusively for Windows-on-Arm devices, supporting new silicon, including the Snapdragon X2 Elite and X2 Plus platforms, as well as the potential launch of NVIDIA's N1/N1X laptop SoCs. The latest feature and security updates have been delivered through both the 24H2 and 25H2 versions, which have been the primary focus for x86-based PC systems. However, as support for 24H2 is coming to an end, the focus is gradually shifting to 25H2. There is still time before October of this year, and Microsoft is using AI for a staged rollout. Importantly, users cannot refuse this update. They can only schedule the restart for the update to be applied or postpone it slightly before it is finally installed.

NVIDIA has released more details about its Neural Texture Compression (NTC) technology, which significantly reduces GPU VRAM usage by up to seven times. In a technology demo presented during one of the GTC 2026 sessions, NVIDIA revealed that its Neural Texture Compression can reduce VRAM usage from 6.5 GB to just 970 MB in certain scenes. This was demonstrated in a video comparing a Tuscan Villa and its interior. With virtually no difference in texture appearance, Neural Texture Compression represents a major advancement in maintaining graphics fidelity while freeing up GPU memory for more game content. For instance, in both the exterior of the Tuscan Villa and the interior demo showcasing detailed tableware, NTC technology provides users with high-quality textures while maintaining excellent material quality.

NTC technology is an AI-driven texture output that allows games to feature high-quality complex materials without a performance penalty. Games can benefit from the substantial VRAM reduction that NTC offers while maintaining the same texture quality. Traditionally, games use block-compressed formats like BCn, such as BC5, BC6, or BC7, which are commonly applied in 4x4 pixel formats and rendered by the GPU. However, NVIDIA has trained small neural networks that can produce the desired pixel format and texture appearance at a fraction of the size of traditional texture compression formats. Instead of using gigabytes of VRAM for textures, NTC drastically reduces VRAM usage by emulating textures, allowing for either much lower VRAM consumption or significantly enhanced material appearance, depending on the game developer's goals. This enables games to incorporate much more complex scenery without any performance penalty, relying on NVIDIA's AI technology to handle the workload. Below is the demonstration of Tuscan Villa, which shows just how the scene looks.

Intel's long-awaited gaming CPU king, the Core 200 "Bartlett Lake-S," is an intriguing CPU generation that isn't actually meant for gaming. The top version, the Core 9 273PQE, is a 12-core, 24-thread processor launched exclusively for commercial and industrial PC OEM markets as an edge/embedded design, and it's not available for the DIY PC crowd. Intel didn't plan for these CPUs to run on commercial motherboard chipsets, and major motherboard manufacturers have confirmed this. However, a modder has achieved something remarkable by booting Intel's "Bartlett Lake" Core 9 273PQE processor into the Windows operating system on a standard Z790 motherboard. Previously, these chips could only reach the POST stage on consumer LGA 1700 boards, but they couldn't run a fully functioning operating system.

Built on the Intel 7 node and designed for Socket LGA1700, "Bartlett Lake" is a non-hybrid chip featuring up to 12 "Raptor Cove" P-cores without any E-core clusters. It was not intended to be compatible with consumer Intel Z790 chipset motherboards. However, main issue was never the socket itself. "Bartlett Lake" shares the same physical LGA 1700 layout as Intel's "Raptor Lake" chips, but that doesn't automatically make it compatible with consumer boards. Typically, the absence of official BIOS and microcode support prevents proper CPU initialization. Earlier attempts usually resulted in startup failures, blank screens, or error codes before progressing further. This recent success changes that narrative. By modifying the motherboard firmware to accept the processor during early boot, the modder overcame the hurdles that had previously caused failures. Once the system accepted the chip at initialization, Windows loaded successfully.

Intel is reportedly preparing yet another CPU price hike, adding to the previous increases implemented in February and March. According to the latest research and channel checks from Chinese market research firm Minutes Logic Society, Intel plans to add another price increase to its existing ones. In February, Intel implemented a first round of CPU price increases ranging from 10% to 15%, depending on the segment and SKU. Just a month later, the company introduced another increase in March, around 15%, with earlier reports suggesting a 10% hike for the consumer CPU sector, like the Core Ultra family of CPUs. This time, we are expecting another increase in May across the overall CPU portfolio, meaning that Intel will again raise prices by a few more percent, depending on the CPU sector, whether it is a Core Ultra CPU or a Xeon server processor.

The total cumulative goal for the price hike is about 30% higher compared to 2025 pricing. Interestingly, Intel is facing a significant problem with CPU supply that it can't address immediately. While a large portion of CPU production is internal, with Intel Foundry handling a bulk of orders, some CPUs require TSMC's silicon for Intel to ship these CPUs. Especially with multi-die packaging, where some parts are manufactured on Intel's node and others on TSMC's node, shipping is impossible until every part arrives and Intel assembles it with its advanced packaging.

Rowhammer attacks exploit vulnerabilities in DRAM hardware by causing targeted bit-flips, allowing attackers to bypass memory isolation and gain control over a device. These attacks were initially focused on CPUs and their associated DDR memory, such as DDR4. However, recent research indicates that NVIDIA GPUs are also vulnerable due to the fragile nature of the GDDR6 memory they use, which directly compromises the CPU host. Two independent research teams have discovered ways to exploit this decades-old memory vulnerability against modern graphics hardware, with trouble extending beyond the GPU itself. The "GDDRHammer" and "GeForge" groups have each developed functional exploits that use Rowhammer bit-flips in NVIDIA GPUs with GDDR6 memory to gain complete control over the host CPU's memory. This attack can perform bit-flips on some NVIDIA GPU models, ranging from the "Ampere" to "Ada Lovelace" families of cards.

An attacker who succeeds in this can read and write anything stored in the machine's main memory. Both teams have also introduced new Rowhammer techniques specifically designed for GPU architecture, achieving a significantly higher rate of bit-flips on GDDR6 memory than previous methods. The critical step in both exploit chains involves targeting the GPU's memory allocator, using controlled bit-flips to corrupt the GPU's page tables. Once these page tables are compromised, the attacker gains arbitrary read and write access to CPU memory, breaking down the security boundary between the graphics subsystem and the rest of the machine. The end result is a full system compromise, as the attacker can manipulate memory at will and gain root access, achieving total control without interacting with privileged software paths. The affected GPUs include the GeForce RTX 3060, which experienced 1,171 bit-flips, and the RTX 6000 "Ada" GPU, which saw 202 bit-flips from the attack.

Intel Corporation today announced the appointment of Aparna Bawa as EVP, chief legal & people officer. Bawa will report directly to CEO Lip-Bu Tan and will lead Intel's global legal, ethics, compliance, people, and culture organizations as the company accelerates its transformation and execution agenda.

"The role of legal and people leadership has never been more critical as Intel drives cultural transformation with discipline, speed, and integrity," said Intel CEO Lip-Bu Tan. "Aparna brings a rare combination of operational rigor, business judgment, and people-first leadership. Her experience helping scale global technology companies through periods of significant change will be invaluable as we build a stronger, more agile Intel."

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.

On your Windows PC, the Unified Extensible Firmware Interface (UEFI) uses Secure Boot certificates to ensure that only trusted software initiates the startup sequence. The certificates currently in use were originally issued in 2011 and are set to expire in late June 2026. To address this, Microsoft has been quietly rolling out updated certificates through Windows Update. Starting in April 2026, users can check their device's status via a new indicator in the Windows Security app. By navigating to Device security and then Secure Boot, a color-coded badge will show whether your device is fully updated, awaiting an update, or requires immediate attention.

The badge system is simple yet significant. A green checkmark indicates that the new certificates are installed and no further action is needed. A yellow caution badge, which will start appearing in May 2026, means the update is either pending or has been blocked by a hardware or firmware limitation. A red stop icon is the most serious state and could appear as early as June 2026, once older certificates start expiring. When this occurs, the device will no longer be able to receive critical boot-level security updates. The same status is reflected in the Windows Security system tray icon, so warnings are visible even when the app is closed.

As we enter a new month, Steam's Hardware and Software Survey data has been processed, providing us with a clearer view of the overall gaming market that uses Steam platform. Today, the most notable change in the Steam Survey is the increase in Linux gamers, who have moved from their historically low single-digit market share to mid-single digits. As of March, Linux-based operating systems were running Steam on 5.33% of all polled systems. This represents an impressive 3.10% increase over February's data, which showed a dip in Linux market share from January's 3.5%. Fortunately, the numbers have now been revised upwards, marking a significant improvement for the community that has been steadily implementing improvements and making Linux-based gaming more accessible to everyone.

What might not be surprising is that a large portion of those 5.33% Linux installations run on Valve's customized SteamOS operating system. With a 24.48% share, the use of SteamOS grew by 0.65% last month alone, while other Linux distributions also contributed significantly. Other Windows alternatives like macOS are gaining momentum as well, with Apple seeing a 1.19% month-over-month increase to 2.35%. Although Linux now holds more than twice the market share of macOS, its growth within the Steam install base is a significant change, nearly doubling in just a month. Perhaps these alternative operating systems are now attracting enough attention from big game studios to encourage them to release native ports instead of relying on translation tools like Wine/Proton.

AMD's RDNA 4-based Radeon RX 9070 and RX 9070 XT graphics cards have finally reached reasonable pricing, as German retailers report that these GPUs are now selling below MSRP. In Germany, the European MSRP for Radeon RX 9070 cards is €629, including 19% VAT. For its bigger sibling, the Radeon RX 9070 XT, the European MSRP is listed at €689, also including the sales tax. However, according to multiple listings from German online retailers, both cards are trading below their European MSRP pricing, marking the first occurrence since the memory shortage fiasco began, which took a toll on the gaming community. The cheapest Radeon RX 9070 non-XT model is listed at €539.00 in the form of the ASUS Prime Radeon RX 9070 OC SKU, while the cheapest Radeon RX 9070 XT model is listed at €640 for the ASRock Radeon RX 9070 XT Challenger GPU. Do note that this is not the first time that these GPUs are priced below MSRP in Germany, but just the first time since the memory shortages started.

Interestingly, this price drop in Germany is not consistent with the markets in the United States, where GPU pricing for the Radeon RX 9070 and RX 9070 XT still ranges around $810-$820 for the non-XT SKU and about $880-$890 for the Radeon RX 9070 XT model. This represents a large premium in the U.S. market, considering that the Radeon RX 9070 and Radeon RX 9070 XT graphics cards have MSRPs of $549 and $599, respectively. Perhaps a fresh supply of GPUs has hit the German market, causing supply to overwhelm demand and significantly pushing prices down. In the U.S., that is not the case, where prices remain high and on an upward trajectory, according to PCPartPicker. In contrast, the German market is experiencing some of the lowest pricing in recent months, finally giving gamers a break.

Late last week, Microsoft released its KB5079391 non-security feature update for Windows 11, which was officially pulled due to widespread installation errors. Today, the company is issuing the out-of-band KB5086672 update to address this problem, as Microsoft has identified the source of the issue and the update can now be safely applied. This latest out-of-band KB5086672 update includes the KB5079473 package released on March 10, KB5085516 released on March 21, and the previously pulled KB5079391 released on March 26. Microsoft has combined all of these into the new KB5086672 package, which addresses the issues that appeared and introduces a variety of new features. Finally, the old installation error message, "Some update files are missing or have problems. We'll try to download the update again later. Error code: (0x80073712)," has been resolved for good.

Microsoft notes that this out-of-band update is available through Windows Update for devices running Windows 11 that have already installed KB5079473 or a later update. It is also available for manual download from the Microsoft Update Catalog. Currently, there are no known issues with this update, and if any arise, Microsoft will highlight them on their support documents website. Interestingly, KB5086672 is one of the first steps by Microsoft toward resolving the issues users have experienced with Windows 11 updates, and hopefully just the beginning of the overhaul that Microsoft has promised. Future non-security feature updates could also focus on other quality-of-life improvements, and installation errors should become less common.

Intel's "Wildcat Lake" processors, part of the Core 300 series non-Ultra family, have been leaked by a reputable source Jaykihn0 on X, revealing the entire lineup across various configurations and SKUs. The lineup includes six SKUs across the Core 3, Core 5, and Core 7 tiers, all designed to operate within a 15 to 35 W TDP range. Each model features a hybrid core configuration, pairing two "Cougar Cove" P-cores with four low-power efficiency cores, completely omitting the traditional "Darkmont" E-cores. Boost clocks range from 4.3 GHz on the entry-level Core 3 304 up to 4.8 GHz on the Core 7 360. All six SKUs share 6 MB of L3 cache, a single NPU tile, and integrated Xe3 graphics. The leak suggests that Intel is bringing architecture closely related to the Core Ultra 300 "Panther Lake" mobile platform into the embedded and industrial space, or perhaps into low-cost laptop configurations that don't require the power of "Panther Lake," appealing to buyers seeking budget-friendly options.

The 2P+0E+4LPE core layout is a deliberate trade-off, prioritizing efficiency over raw multithreaded performance, which suits the thermal constraints common in edge and IoT deployments. NPU performance figures range between 15 and 17 TOPS across the lineup. While this won't power the largest LLMs, it may be more than sufficient for on-device inference in industrial or automation settings. The Core 3 304 deserves special mention: it reduces to a single P-core and one Xe graphics unit, creating a clear cost-optimized option at the bottom of the lineup. SIPP certification, important for buyers needing stable, long-lifecycle platform support, is available on the Core 7 360 and Core 5 330 but not consistently across the lineup. Notably, there is no vPro support on any SKU, clearly distinguishing "Wildcat Lake" from Intel's enterprise mobile portfolio.

The NVIDIA App update today introduced some interesting features, such as DLSS 4.5 dynamic multi-frame generation and a 6x mode. Additionally, the app now includes a new beta version of NVIDIA Auto Shader Compilation (ASC). This feature takes DirectX 12 shaders from games and quietly compiles them while the system is idle or not running any graphically intensive tasks. Typically, when you start a game, you have to wait for all assets to load and shaders to compile before you can begin playing. However, with ASC, NVIDIA aims to shorten this process by pre-compiling shaders to reduce loading times and, interestingly, decrease in-game stuttering, which can occur when shaders don't load properly. NVIDIA states that this feature is opt-in within the NVIDIA App and can be enabled by navigating to the Graphics Tab > Global Settings > Shader Cache. Once in the menu, users can access a range of settings, including the option to turn on Auto Shader Compilation.

Since ASC uses a separate folder, users will need to allocate sufficient disk space to store the shaders that ASC will access. In the NVIDIA App, gamers can choose the "Compile Now" option to pre-compile all game shaders immediately by clicking on three dots, or they can wait for the system to do it automatically when it becomes idle. As compiling shaders requires some computing power, there are settings to control system utilization, with the default set to medium. The NVIDIA App will also display the date of the last compilation. Interestingly, ASC will perform its functions once a game is downloaded and after a new driver update is installed for optimal performance. NVIDIA requires GeForce Game Ready Driver 595.97 WHQL or newer for ASC to work, and more optimizations are expected as the beta testing concludes in the coming weeks.

The Intel Binary Optimization Tool (BOT) has been launched alongside the "Arrow Lake Refresh" series of processors, which includes the Core Ultra 5 250K Plus and Core Ultra 7 270K Plus models. While the tool is beneficial for gamers looking to extract a few extra frames from their setups, it may be a nightmare for makers of benchmarking tools like Geekbench by Primate Labs. Recently, Primate Labs testing found that BOT changes the way .exe applications run and concluded that Geekbench runs will now flag these BOT-enhanced runs. However, in deeper testing, Primate Labs discovered that Intel's BOT may deliver significant boosts in some applications like Object Remover and HDR, increasing performance by up to 30%. This is thanks to the deep vectorization that the BOT performs behind the scenes to optimize performance.

For example, Primate Labs used Intel's own Software Development Emulator (SDE) to measure how many instructions were executed and which types of instructions the program executed. Without BOT, Geekbench 6 required a total of 1.26 trillion instructions to finish, while a BOT-enhanced run completed with 1.08 trillion instructions. This is an impressive 14% reduction. However, when examining the execution by type, we see that BOT makes heavy use of vector instructions like SSE2 and AVX2. The number of scalar instructions needed to execute a program fell from 220 billion to 84.6 billion, while the number of vector instructions increased from 1.25 billion to 18.3 billion, a 13.7x increase. This means that Intel BOT finds a way to turn inefficient scalar code into vectorized instructions that are processed much more efficiently inside Intel CPUs. These techniques indicate a very complex behind-the-scenes process than was originally believed. The Geekbench v6.7 update will include a flag for BOT, allowing future Geekbench results to be easily distinguished as BOT-enhanced or not.

AMD has quietly renamed its Anti-Lag 2 technology as part of the FSR package, now calling it "FSR Latency Reduction 2.0." This move aligns with AMD's recent trend of rebranding FSR-related technologies. The AMD Radeon marketing team has successfully unified FidelityFX Super Resolution under the FSR branding, although Anti-Lag 2 was previously an exception, bundled with other AMD technologies. The advanced graphics technology, once known as FidelityFX Super Resolution, is now simply called "FSR." This change is reflected on AMD's official product page, which notes that FSR stands for "formerly AMD FidelityFX Super Resolution." However, AMD has not formally announced this rebranding. These changes occurred before the official launch of the FSR "Redstone" product in late December last year. Now, every new announcement features the standard FSR language, suggesting that this renaming might be part of a broader update to Anti-Lag 2.

Since FSR is aimed at gamers, it is now included in the FSR package as "FSR Latency Reduction 2.0." With FSR Redstone, AMD has already grouped four technologies under the FSR "Redstone" name: FSR Upscaling, FSR Frame Generation, FSR Ray Regeneration, and FSR Radiance Caching. If the renaming becomes more than just a label update, FSR Latency Reduction 2.0 could become the fifth component of the FSR "Redstone" suite. Technologies like AMD Anti-Lag 2 are specifically designed to reduce latency by improving CPU and GPU coordination. Even without frame generation, it can lower latency in a game, but it may be especially useful when synthetic frames are involved, helping to keep latency at a level where any added delay is far less noticeable.

NVIDIA has finally launched its long-teased Dynamic Multi Frame Generation (MFG) and Multi Frame Generation 6x mode today through a new NVIDIA app beta update. This marks the full public release of NVIDIA's DLSS 4.5 technology suite, which enables the GPU to generate up to five additional frames following each traditionally rendered frame using generative AI. Using the new MFG 6x mode results in a 6x performance uplift, meaning a game that traditionally runs at 60 FPS can now reach 360 FPS. Users will need to enable "beta and experimental features" in the NVIDIA app's Settings menu, and the GeForce Game Ready Driver 595.79 WHQL or newer is required to access all features. This will give a limited set of games (for now) a massive performance uplift, which includes ARC Raiders Flashpoint, Marvel Rivals Season 7, 007 First Light, CONTROL Resonant, and Tides of Annihilation. More games will get the official support as NVIDIA is working with game studios.

However, for setups where a monitor is maxed out at 240 Hz or 144 Hz, as many gaming panels are, using 6x MFG would be overkill. This is where Dynamic MFG comes into play. The technology determines which MFG multiplier is needed based on the display's refresh rate capability and the input framerate from the upscaler. NVIDIA calls this the "automatic transmission" for MFG, drawing a parallel to modern vehicle automatic transmission systems that switch gears based on demand. In graphically intensive scenarios, the multiplier can scale up to 4x, 5x, or 6x, while lighter scenes like settings menus or static sequences may only require a 2x multiplier to hit the target frame rate.

NVIDIA is reportedly experiencing manufacturing issues with its next-generation "Rubin Ultra" GPU design, one of the company's most ambitious chip development projects, due to the limitations of modern packaging technology. The world's largest company is already shipping customer samples of the standard "Rubin" GPUs, with mass shipments set to begin this summer. However, the current roadmap for the upgraded "Rubin Ultra" design may be encountering technological limitations, as NVIDIA's design goals are too ambitious for TSMC's packaging capabilities. Reportedly, NVIDIA plans to double the regular "Rubin" two-die package with 8 HBM4 modules into a new "Rubin Ultra" package that will include four silicon dies and 16 HBM4E modules in a single package. This configuration is scheduled for 2027, but the sheer volume of silicon may be too much for TSMC's packaging, according to Global Semi Research.

In a typical CoWoS package, TSMC usually combines multiple smaller dies and multiple HBM memory modules into a unified package that supports the entire AI build-out. However, with the ambitious "Rubin Ultra" design, NVIDIA planned to use CoWoS-L, which was expected to handle the design and concept that "Rubin Ultra" was based on. It is rumored, however, that in a 2+2 die package—meaning four dies as in this architecture—TSMC is encountering warping issues. The package—which includes a substrate—is bending in multiple directions, causing the compute dies of "Rubin Ultra" to not make complete contact with the underlying substrate. This instability means that TSMC has to explore alternatives within its packaging portfolio. One of these alternatives is a panelized approach called CoPoS, which stands for Chip-on-Panel-on-Substrate.

NVIDIA's latest GeForce NOW update has introduced 90 FPS streaming to various VR headsets, and Apple users are in for a treat. For the Apple Vision Pro headset, GeForce NOW will deliver 90 FPS at 4K resolution, offering a noticeable improvement for anyone using NVIDIA's game streaming service with their Vision Pro headset. The Apple Vision Pro features two displays, one for each eye, capable of running at a resolution of 3,660 × 3,200 and up to 120 Hz. It's great news that NVIDIA has updated its GeForce NOW service to officially support at least 4K resolution, running at 90 FPS. While it's unclear how many gamers use the Apple Vision Pro as their gaming display, the addition of support by NVIDIA suggests a significant number. Available as part of the Ultimate package, members can stream at 90 FPS on other VR headsets as well, but at lower resolutions.

Additionally, everyone can stream at 1080p and 90 FPS, while 1440p is reserved for Pico and Meta Quest. Currently, only the Apple Vision Pro can handle 4K and 90 FPS output from GeForce NOW. Although not many games can run at 4K resolution and 90 FPS on their own, NVIDIA's DLSS technology can boost the frame rate and deliver impressive visuals, ensuring a smooth 4K mode at 90 FPS. Finally, NVIDIA has also scheduled the rollout of H.265 video decoding support for browsers, which will greatly enhance streaming efficiency and visual quality from NVIDIA's virtual gaming server.