News Posts matching #GPU

Return to Keyword Browsing

AMD FSR Leak Accidentally Tips Potential Support for Older GPUs

AMD recently released its FidelityFX Super Resolution SDK 2.0 to accompany the shift to the machine learning-based FSR 4, but in a bizarre twist, the company also seemingly accidentally open-sourced the libraries to FSR 4 via its AMD GPUOpen project, giving tech-savvy browsers a sneak peek at the inner workings of FSR 4. The files have since been removed from AMD's GPUOpen library, but quick fingers managed to snag screenshots of the file directory, which reveals enough. One of the biggest takeaways was the inclusion of 8-bit integer (INT8) libraries, which use less resources but are also less precise than those used in RDNA 4, indicate that there is or may have been some intention to release FSR 4 for AMD's RX 7000 GPUs. FSR 4 was initially announced to be exclusive to AMD's Radeon RX 9000 series GPUs, but that may not be the case for long.

There have previously been hints that AMD FSR 4 could be coming to older RDNA hardware, and there have even been mod projects that have successfully forced it to work on older hardware, but this is a substantial official signal that AMD may be working on porting it to older hardware. There is also speculation that the INT8 libraries may be destined for the PlayStation 5 Pro, but the file type seems to indicate otherwise—PS5 uses .PSSL, while these shaders were packaged as .HLSL, suggesting they were for PC use.

UL Releases 3DMark Solar Bay Extreme, The Ultimate Ray Tracing Benchmark

UL Benchmarks has released Solar Bay Extreme, a major update to the cross-platform 3DMark Solar Bay benchmark that cranks up ray tracing demands for both desktop and mobile devices. The update is now available as a free upgrade for existing 3DMark owners and is designed to showcase the capabilities of current hardware in pushing modern lighting techniques. Everyone can now run the test to compare scores on 3DMark leaderboards and utilize the built-in stress mode to monitor for throttling and stability issues during extended runs. Solar Bay Extreme runs on native APIs for each platform, utilizing Vulkan on Android, Metal on Apple platforms, and DirectX 12 on Windows, ensuring results that accurately reflect how real games behave.

The new edition brings more detailed scenes and a heavier ray tracing workload, including ray-traced specular reflections on rough and glossy surfaces, glass reflections and soft ray-traced shadows for directional lights. UL also added a second assembly bay to the scene where a rival team builds a solar array, which adds geometric complexity and varied lighting that exposes differences between GPUs and drivers.

ASUS Next-gen Handhelds ROG Xbox Ally and Xbox Ally X to Debut at Gamescom 2025

At Gamescom 2025 in Cologne, visitors can test the ROG Xbox Ally and ROG Xbox Ally X for the first time. Republic of Gamers (ROG) will also present the full lineup of NVIDIA GeForce RTX 50 Series laptops for 2025, offering an exclusive preview of the next hardware generation. Ned Luke and Shawn Fonteno, stars of the latest ROG campaign film and known from Grand Theft Auto V, will also be on hand to interact with fans.

ROG and Xbox present: The new ROG Xbox Ally handhelds
ROG and Xbox have collaborated to develop the ROG Xbox Ally X and the ROG Xbox Ally - two powerful, portable gaming handhelds for maximum performance on the go. The ROG Xbox Ally is powered by a highly efficient AMD Ryzen Z2 A-series processor with four Zen 2 cores, eight threads, and eight RDNA 2 GPU cores. Complemented by 16 GB LPDDR5X-6400 RAM, a 512 GB M.2 SSD, and a 60 Wh battery, intense gaming sessions on the go are no problem. The powerful ROG Xbox Ally X features the AMD Ryzen AI Z2 Extreme, a new 8-core/16-thread Zen 5 APU with 16 RDNA 3.5 GPU cores and an integrated NPU. It also features 24 GB LPDDR5X-8000 RAM, a 1 TB M.2 SSD, and a larger 80 Wh battery for extended playtimes, even with demanding titles.

ASUS Announces New ROG XG Station 3 With Thunderbolt 5

The ROG XG Station 3 (Thunderbolt 5) external graphics dock is a plug-and-play solution that merges the power of an external GPU with the expandability of a Thunderbolt dock, instantly transforming any Thunderbolt 5-enabled laptop into a gaming or rendering powerhouse. The incredible 80 Gbps bandwidth of Thunderbolt 5 allows users to upgrade to external GPU computing power, including NVIDIA GeForce RTX 50 Series, enabling high-speed gaming, video editing, 3D rendering, or AI application development on their laptop.

Along with the Thunderbolt 5 port, the ROG XG Station 3 offers three USB Type-C ports with up to 10 Gbps data-transfer speeds. It also features ASUS Aura Sync lighting and the BTF hidden-connector design that provides a clean look while improving cable management.

ASUS Unveils Quad-Fan GeForce RTX 5090 ROG MATRIX GPU

Last week, we received a few teasers from ASUS about its upcoming 30th anniversary special edition GPU, which will be available in the form of the GeForce RTX 5090 ROG MATRIX SKU. Today, ASUS officially showcased what we thought was an air-cooled design. Confirming that the card is air-cooled, it features four massive fans, which closely resemble ASUS's RTX 5090 Astral GPU, with three fans in the front and one in the back, performing a push-pull air cooling design. The GPU shroud features a massive ROG red design with a distinct ring on one fan, carrying an acrylic plate with the ROG MATRIX logo. ASUS also offers an option for BTF-supplied power delivery, along with a 16-pin 12V-2x6 connector, allowing for a total power setting of up to 800 Watts, which significantly exceeds the regular 575 Watts of the Founders Edition GPU. ASUS claims this supplies a 10% performance boost on its own, and the boost clock frequency is confirmed at 2,730 MHz.

For maximum performance, ASUS utilizes 3 Oz copper PCB layers to enhance heat dissipation and minimize voltage drop across componentry, resulting in improved overall performance and potentially higher overclocking capabilities. For die-to-heatsink contact, liquid metal is used to maximize heat dissipation in this beast, which is necessary given its ability to consume 800 W of power. ASUS still hasn't announced pricing and final specifications, like memory clock, but we assume it will also be overclocked out of the box. For serious tuning, the card has yet to be put through professional overclocking labs. Given that only 1,000 units will be sold, it is a question whether anyone will attempt any extreme OC record on the card.

New Mac Mini Models on the Horizon With Apple M5 and M5 Pro SoCs

Apple appears to be planning to refresh its Mac Mini line-up, according to an exclusive report by AppleInsider. The new Mac Mini will reportedly feature Apple's M5 SoCs, which will supposedly bring improved GPU performance to compete with the likes of AMD's Strix Halo APUs, thanks to 3D stacking tech. Given that the Mac Mini line-up only just received a hardware overhaul, it seems unlikely that there will be any physical changes for the M5 versions.

So far, two versions of the Mac Mini refresh have been unofficially confirmed, the J873s with the M5 Pro SoC and the J873g, which looks like it will feature the base model M5 SoC. An actual launch date is unclear at this point, but it is expected that MacBook models featuring the M5 chips may not launch until as late as 2026. That's not necessarily to say that the Mac Mini won't get an update this year, but traditionally the M-series Apple Silicon has launched in the company's MacBook line-ups. Apple notably skipped the M3 SoCs when it came to the Mac Mini, but it seems as though competition in the mini PC space might be pushing the tech giant to release more frequent updates.

PALIT Unveils NVIDIA GeForce RTX 5060 Infinity 2 OC Graphics Card

As a leading innovator in graphics card technology, Palit Microsystems Ltd. is excited to unveil its latest GeForce RTX 5060 Infinity 2 OC graphics card. This new addition to the Palit lineup delivers streamlined aesthetics alongside robust thermal performance and enhanced capabilities for gamers and creators.

Affordable Power, Overclocked Performance
The GeForce RTX 5060 Infinity 2 OC comes factory-overclocked, offering higher performance straight out of the box—ideal for users who want an edge without manual tuning. The Infinity 2 OC embodies a philosophy of simplicity and power. Its design, highlighted by the Infinity symbol (∞), not only offers a visually clean aesthetic but also represents smooth airflow, allowing users to game and create without limits. Designed without any lighting effects, and built with a compact 2-slot form factor, it offers a focused aesthetic that fits seamlessly into a wide range of builds.

New Intel "Lunar Lake" Core Ultra 7 254V SKU Appears

Intel's unannounced Core Ultra 7 254V has begun appearing in benchmark databases, and the figures suggest this may be a deliberately trimmed member of the "Lunar Lake" family, aimed at more affordable laptops. PassMark lists a multicore score of 17,327 and a single-core score of 4,089, which places the 254V noticeably behind the 256V and 258V in multithreaded work, while maintaining single-thread performance in line with other Core Ultra 7 chips. The listing also shows 12 MB of L3 cache and confirms an eight-core layout, matching Lunar Lake's standard design. Separate Furmark traces point to integrated Arc 140V graphics running at approximately 2.0 GHz, a GPU clock similar to that of higher-tier 7-series SKUs, which suggests that Intel may be binning an existing 256V to create a lower-cost option rather than designing something entirely new.

What remains unclear is the memory configuration, as Lunar Lake utilizes LPDDR5X packaged inside the SoC, and suffixes have so far indicated RAM capacity, with 2x8V mapping to 32 GB and 2x6V mapping to 16 GB. A 2x4V suffix would imply an even smaller memory option, but no confirmed devices are shipping with the 254V yet, and we don't know if a version with less than 16 GB memory has ever been planned/shipped. If Intel does release it, the chip could widen Lunar Lake's reach by offering efficient battery life and acceptable day-to-day performance at a lower price, making it an attractive choice for potential budget-focused buyers.

Report: Cloud-hosted Desktops Now Cheaper Than Laptops for Businesses?

According to the technology analyst firm, Gartner, we learned this week that hosted desktops, known as Desktop-as-a-Service (DaaS), are now often less expensive to operate than traditional laptops and are expected to become cost-effective for most employees within two years. That could force many enterprises to invest more in DaaS rather than their local PC infrastructure. In its Magic Quadrant for DaaS, Gartner forecasts that by 2027, approximately 20% of workers will use a cloud-hosted machine as their primary workspace, up from around 10% in 2019. The firm also expects DaaS to be cost-effective for approximately 95% of use cases/roles by 2027, compared with roughly 40% in 2019. Gartner cites lower total cost of ownership as a key factor, especially when organizations pair DaaS with light, inexpensive thin-client endpoints. New deployments are increasingly cloud-first, and many firms are choosing hosted solutions over building or expanding on-premises virtual desktop infrastructure. Especially with AI buildout requiring expensive GPUs, it could be more viable for enterprises to use a hosted acceleration when needed, instead of buying expensive hardware for local processing.

The report shows several market leaders and practical questions for IT teams. Microsoft is named the standout vendor for its wide cloud reach, workplace tools, and AI features, though buyers are advised to consider the differences between Windows 365 and Azure Virtual Desktop. Citrix is praised for its strong management and protocol technology, but is warned for offering bundled packages and long subscription terms. Omnissa is recognized for scalable offerings, while some customers worry about global support. AWS earns credit for integrating management tools into Workspaces. Beyond vendor rankings, experts highlight a new use case that could transform desktop strategy. As AI tools and autonomous agents begin to handle routine computer tasks, they will likely need their own optimized virtual workspaces. That raises questions about how those spaces will be designed, secured, and licensed. However, the takeaway is clear: budget planning, procurement, and desktop strategy must adapt to recurring cloud costs, endpoint choices, and a future where both people and software agents coexist in the digital workplace.

Intel "Lunar Lake" Updated PL2 Setting Can Yield Up to 30% Higher Gaming Performance

We recently published our MSI Claw 8 AI+ A2VM review, which is the only handheld gaming device to utilize the Intel Core Ultra 7 258V "Lunar Lake" CPU, featuring four E-cores and four P-cores, paired with the Intel Arc 140V Xe2 GPU. During our testing, we manually set different TDP values via the Manual power profile slider to see what the results are in gaming and how it compares to other, mostly AMD-based, handhelds. However, Intel notified us about performance levels used in testing, and that the official specification for Lunar Lake states that PL2 should always be at least 1 W higher than PL1, or always set at 37 W (the maximum TDP value for Lunar Lake chips). This means that whenever a CPU can extract an additional boost, it will, yielding up to 30% more performance, as we have seen in our own testing.

Initially, we used the Manual power profile, which had allowed owners to match PL2 to PL1. However, Intel noted that when PL2 is set to at least a single W higher than PL1, there is an expected 10% performance increase across the suite of games, but depending on the scenario, it can lead to a 30% increase in some instances. An update has arrived for the MSI Center M app, which locks PL2 to always be at least 1 W higher than PL1. As a result, you can no longer match the two. It was only possible before the update. In our retest, Cyberpunk 2077 and Space Marine 2 achieved nearly a 30% performance uplift with this setting. The Lunar Lake-based MSI Claw AI+ A2VM handheld is the highest-performing device we have tested across the board, even surpassing AMD-based devices in our testing suite. In other games, baseline improvement is set at 10%, just as Intel had notified. Our testing confirmed that Lunar Lake is now the most powerful gaming CPU for handheld devices that we own and have tested.

15:57 UTC: We updated some of the claims to better reflect the PL2 setting and that we manually set TDP values, too.
Below you can find the updated performance graphs.

New AMD Radeon GPU Tool Allows Users to Force-Enable FSR 4 in More Games

AMD Radeon GPU owners can now attempt to enable FSR 4 in more games, thanks to a new "AMD GPU Profile Manager" community tool that overrides the driver whitelist. Published on GitHub by developer Mikhail and demonstrated by content creators including Ancient Gameplays, the utility exposes AMD Adrenalin profile databases. It allows users to add or modify entries that typically require official driver support. The manager targets Radeon RX 9000 series cards and works by allowing titles that already support FSR 3.1 to try FSR 4. Users can enable FSR 3 first, either in the game or before launch, and then watch Adrenalin for signs that FSR 4 injection is active. The tool also offers quick add buttons for recent test builds and a one-click option to import profiles from AMD tech preview drivers, which is particularly handy for those who have reverted to older drivers and lost newly added entries. It aims to restore functionality and give power users a way to experiment while AMD completes its official whitelisting process.

There are important cautions alongside the convenience. The utility does not guarantee success in every title, and forcing FSR 4 into games with anti-cheat protections or online multiplayer can cause crashes, false positives, or account penalties. Community-maintained lists, demonstrations, and project notes highlight which games are confirmed to work and which are incompatible. Users are advised to back up driver profiles before making changes and to avoid applying forced profiles in competitive environments until developers add official support. The developer says more profile editing features are planned, and the project could grow into a general profile manager similar to well-known tools in other GPU ecosystems.

ASUS Announces the NUC 15 Performance

ASUS today announced NUC 15 Performance, a cutting-edge barebone mini PC that features the latest Intel Core Ultra 9 / 7 processors (Series 2) and the latest NVIDIA GeForce RTX 5070 / 5060 Laptop GPUs. This compact powerhouse delivers exceptional performance in a sleek 3-liter design, and is ideal for high-vision computing in commercial and industrial applications.

Next-level performance
Powered by either an Intel Core Ultra 9 275HX or Ultra 7 255HX processor, NUC 15 Performance delivers exceptional AI performance and efficiency, offering up to an 18% performance boost over previous NUC models for smooth multitasking while running the most demanding apps. Designed for business and creative professionals, NUC 15 Performance ensures seamless, high-efficiency computing to enhance productivity.

Intel Releases Arc GPU Graphics 101.6987 WHQL Drivers

Intel has released the latest 101.6987 WHQL version of its Arc GPU Graphics Drivers, bringing several new improvements and bringing Game On support for Battlefield 6 Open Beta and Mafia: The Old Country, as well as adding a performance boost to Doom: The Dark Ages running on Vulkan API and Arc B-series GPUs. Intel also added support for Shared GPU Memory Override control in Intel Graphics Software for built-in Intel Arc GPUs in Core Ultra (series 1 and 2) CPUs running on Windows 10 and 11, as well as added power management optimization for Intel Core Ultra 200V series (Arrow Lake H) CPUs, which should improve performance and frame pacing in low-power modes.

The latest 101.6987 Arc GPU Graphics Drivers fix a single issue where the Naraka Bladepoint game could freeze with Graphics Enhancements set to XeSS on Arc B-series GPUs. There are several known issues, including ones with Fortnite, PEAK, Call of Duty: Black Ops 6, Call of Duty: Warzone 2.0, Mafia: The Old Country, Marvel's Spider-Man 2, and other games and apps, but hopefully, Intel should fix those with the next driver release.

DOWNLOAD: Intel Arc GPU Graphics Drivers 101.6987 WHQL

Huawei Open-Sources AI Software Stack, Aiming to Rival CUDA

Huawei used its annual developer summit in Shenzhen this week to announce that it will release the Compute Architecture for Neural Networks (CANN) software stack, which powers its Ascend AI processors, as open source before the end of the year. Rotating chairman Eric Xu Zhijun said the goal is to make it easier for developers to get started and build a lively community around Huawei's chips. In the weeks leading up to the public launch, the company has collaborated with Tsinghua University, Peking University, AI startups such as StepFun and Infinigence, and the newly formed Model-Chips Ecosystem Innovation Alliance to agree on contribution rules and establish transparent governance. By putting CANN on public code repositories under a permissive licence, Huawei hopes to capture the same kind of grassroots enthusiasm that drove Linux and RISC-V adoption, only this time for hardware designed in China.

This announcement comes at a sensitive moment on both technical and political fronts. China's regulator has initiated a security review of NVIDIA's H20 GPUs, citing concerns about remote-management features that compromise data sovereignty. At the same time, US authorities have approved limited exports of the same chip. Unlike NVIDIA's CUDA framework, which now blocks third-party translation layers on non-NVIDIA GPUs, CANN will allow forks and cross-platform ports from day one. It will take years for CANN to match CUDA's two decades' worth of optimized libraries and tooling. If Huawei can keep contributors engaged, an open-source CANN may shift the company's role from chip maker to the neutral steward of a unified AI software ecosystem. The first public snapshot is scheduled for the fourth quarter, alongside the release of the Ascend 910C processors and a joint reference server design with Tencent Cloud.

NVIDIA Reassures Customers: No Backdoors, No Kill Switches, No Spyware in its GPUs

NVIDIA is shipping millions of GPUs across data center, consumer, and professional workloads that power every use case today. Since the infrastructure powering vital services, such as healthcare and finance, requires NVIDIA GPUs to run, some policymakers have demanded backdoors, kill switches, and spyware to prevent malicious actors from performing any harmful activity on an NVIDIA GPU. However, NVIDIA issued a blog post reaffirming that its GPUs don't contain any backdoors, kill switches, or spyware, and that such features will never be present on NVIDIA hardware.

The company argues that safety should stem from features such as diagnostics, performance monitoring, bug reporting, and timely firmware patching, rather than relying on a kill switch to power off a GPU. And all of that is happening with a user's consent and knowledge of what services run on their hardware. "Hardwiring a kill switch into a chip is something entirely different: a permanent flaw beyond user control, and an open invitation for disaster. It's like buying a car where the dealership keeps a remote control for the parking brake—just in case they decide you shouldn't be driving. That's not sound policy. It's an overreaction that would irreparably harm America's economic and national security interests," says NVIDIA blog. We hope that the company remains on this track, as promoting good values is essential for development in the hardware industry.

AMD Reports Second Quarter 2025 Financial Results

AMD (NASDAQ:AMD) today announced financial results for the second quarter of 2025. Second quarter revenue was a record $7.7 billion, gross margin was 40%, operating loss was $134 million, net income was $872 million and diluted earnings per share was $0.54. On a non-GAAP(*) basis, gross margin was 43%, operating income was $897 million, net income was $781 million and diluted earnings per share was $0.48. As previously announced, our second quarter results were impacted by the U.S. Government's export control on our AMD Instinct MI308 data center GPU products. For the quarter, these restrictions led to approximately $800 million in inventory and related charges. Excluding these charges, non-GAAP gross margin would have been approximately 54%.

"We delivered strong revenue growth in the second quarter led by record server and PC processor sales," said Dr. Lisa Su, AMD Chair and CEO. "We are seeing robust demand across our computing and AI product portfolio and are well positioned to deliver significant growth in the second half of the year, driven by the ramp of our AMD Instinct MI350 series accelerators and ongoing EPYC and Ryzen processor share gains."

Oxmiq Labs: Re-Architecting the GPU Stack: From Atoms to Agents

Oxmiq Labs Inc., the all-new GPU software and IP startup founded by one of the world's top GPU architects and visionaries, Raja Koduri, emerges from stealth after two years of intensive IP development. Raja has assembled a world-class team of GPU and AI architects with over 500 years of combined experience, hundreds of patents, and a collective track record of generating more than $100B in revenue at prior companies.

The Opportunity:
Modern computing has fundamentally shifted toward multimodal experiences where text, audio, video, images, and 3D environments seamlessly interact, establishing GPU architecture as the cornerstone of this transformation. Unlike fixed-function AI accelerators that handle specific tasks, GPUs provide the general-purpose computational flexibility required for these diverse modalities while maintaining deep integration with mainstream operating systems through standardized APIs and unified memory models. This architectural advantage positions GPUs as the essential compute engine for both current applications and the emerging landscape of multimodal AI, where heterogeneous workloads must be processed in harmony.

Xbox "Magnus" APU to Power Multiple Systems, Not Only the Main Console

Microsoft's next-generation Xbox will be a new design approach with the introduction of the "Magnus" APU, a chip built to power an entire family of devices rather than a single flagship console. Leaks and analysis shared by Digital Foundry have revealed that Magnus splits compute and graphics into separate dies linked over a high-bandwidth 384-bit fabric. This Lego-style modular design enables Microsoft to swap out upgraded CPU or GPU tiles without redesigning the motherboard. By moving away from a few-year console generations toward more frequent "phase" updates, Microsoft aims to maintain competitive performance and introduce new systems, ranging from palm-sized streaming boxes to even full-tower gaming rigs, on an annual basis.

For AMD, which architects the Magnus APU, this represents an opportunity to tap into a more predictable and potentially lucrative channel beyond the traditional DIY PC market. By planning an entire vertical of systems powered by Magnus APU, Microsoft is aiming to compete more with integrated PC systems than with consoles alone. If the Magnus APU can deliver the gaming performance required for both handhelds and tower-style consoles, Microsoft could capture a significant market share. For handheld versions, the design could likely be a cut-down version to maintain a lower power draw and keep thermals in check. However, for the regular console, the Magnus APU will likely be a powerful combination of Zen 6 / Zen 6c, along with UDNA graphics. We are yet to see what Microsoft plans as we move closer to the next-gen Xbox launch.

AMD Officially Launches Radeon RX 9060 non-XT GPU

Yesterday, we covered a leak that seemingly revealed the final specifications of AMD's upcoming Radeon RX 9060 non-XT GPU SKU, indicating that it appears to be a unit very similar to its XT sibling. However, AMD has officially posted a product page with detailed specifications, providing us with something a bit different from what we initially expected. Featuring 28 RDNA 4 Compute Units, the card boasts 1,792 Streaming Processors and 64 ROPs, alongside 112 Texture Mapping Units. This is a significant downgrade from the Radeon RX 9060 XT SKU, which featured 32 CUs for 2,048 SPs. What we believed would be a 32 CU design initially turned out to be a cut-down variant built on the same Navi 44 XL die, most likely coming from a lower bin.

For memory, AMD utilizes 8 GB of GDDR6 memory operating at 18 Gbps on a 128-bit bus, resulting in 288 GB/s of bandwidth. All of that is packed onto a 132 W TBP board, with a minimum PSU requirement of 450 W. AMD advertises the card as the 1080p powerhouse, delivering high frame rates at ultra settings. On its website, AMD lists the card performing 108 FPS in Assassin's Creed Mirage, 98 FPS in Call of Duty: Black Ops 6, 153 FPS in DOOM Eternal (RT), 67 FPS in Dragon Age: The Veilguard, 188 FPS in F1 24, 106 FPS in Horizon Zero Dawn Remastered, 120 FPS in God of War: Ragnarok, and 100 FPS in Resident Evil 4 (RT). All of these were tested at native 1080p ultra settings, without the use of any frame generation technologies. Final clock speeds, pricing, and availability remain unknown.

MoreClockTool 2.0 Debuts with Redesigned UI and Advanced AMD Radeon RX GPU Controls

The developer behind the original AMD Radeon GPU utility tool, MoreClockTool, has officially launched version 2.0. The new MCT 2.0 is now a complete Windows app, featuring a light or dark theme to choose from, along with new features that allow you to customize your AMD Radeon RX GPU in every possible way. Everything, from clock rates and power limits to GFX voltage, memory timing levels, and tuning profiles, is available for customization. For users of RDNA 2-based Radeon RX 6000 series GPUs and newer, tuning and fan settings are fully supported, with every feature in MCT 2.0 working. For this functionality, a driver-only install is sufficient, and AMD Adrenalin Software is not required for the tool to work correctly, as the driver provides the necessary interface for MCT 2.0 to direct the GPU's internals.

The tool allows up to 32 custom tuning and fan profiles in total. MCT 2.0 now features smooth sliders for customizable fan/voltage/power curve, allowing you to shape it with just your mouse, no keyboard required. Users can grab a corner of the window and drag to resize the controls, which grow or shrink right along with it. Shrink it down, and the sliders quietly disappear, leaving nothing but a slim little panel. Stretch it wide, and let the tool take over the whole screen. Being Windows-native now, it's much more user-friendly. Given that the MCT 2.0 development required a significant amount of engineering hours, the developer published the tool in the Microsoft Store for a small fee of $2.89. Check out the screenshots of the tool below, and click the link here to download it.

Bolt Graphics' Zeus GPU: Dev Kits Arrive 2026 for Gaming, HPC, and CAD

When Bolt Graphics made its appearance early this year, it was quite a shock to see a new entrant in the already very competitive consumer/prosumer GPU market. However, Bolt promises that its Zeus GPU is an entirely different beast, focused on areas such as path tracing in games, CAD workloads, HPC simulations, and film/TV. Its Zeos GPU is a unique design too, with expandable memory. There is a 32/64/128 GB soldered memory capacity, with 2x or 4x DDR5 SO-DIMM slots, supporting up to 384 GB of memory. In addition to regular video outputs, the Zeus GPU features a 400 GbE QSFP-DD port for advanced networking on render/simulation farms, allowing thousands of interconnected GPUs that bypass a NIC. Bolt Graphics notes that the developer kits are on track to arrive in 2026, with mass production set for 2027.

When it comes to gaming performance, Bolt uses consumer-grade GPUs such as NVIDIA GeForce RTX 5090 for comparison. Until now, developers have often had to compromise on visual quality to meet their frame-rate targets when using real-time ray tracing and path tracing. With Zeus, they no longer have to choose between speed and fidelity, delivering full-quality 4K path-traced graphics at a smooth 120 FPS with at least 25 samples per pixel. On the compute side, Zeus excels in heavy scientific and engineering tasks, achieving up to 20 FP64 TeraFLOPS while consuming less than 400 W. It tackles electromagnetic-wave simulations a staggering 300 times faster than the B200, without accuracy downgrade. We are yet to see if the Zeus will live up to its hype, so first dev kits will uncover a lot about it next year.

MSI Claw 8: Intel Lunar Lake vs AMD Ryzen Z2 Extreme

MSI offers two versions of its Claw handheld gaming console: the Claw A8, based on the AMD Ryzen Z2 Extreme SoC, and the Claw 8 AI+, based on the Intel "Lunar Lake" CPU. In the latest gaming benchmarks, with both processors capped at the same power envelopes, the AMD-based Claw A8 delivers a slight win. In a detailed comparison, courtesy of Golden Pig Upgrade, we see the AMD-based Claw handheld competing against the Intel-based model across ten AAA titles in terms of average FPS and 1% lows at 1080p resolution, as well as at 17 W capped TDP and 30 W capped TDP for both processors.

MSI's Claw A8 handheld powered by the Ryzen Z2 Extreme pulled ahead of the Intel-based Claw 8 AI+ in many titles, with Monster Hunter Wilds seeing up to a 27% frame-rate boost, while average gains stayed under 5% in most games. At 1080p and 30 W, it ran about 5.7% faster, rising to 7.4% at 17 W, and its superior 1% lows frame rates delivered noticeably smoother gameplay under stress. Under the hood, the Z2 Extreme built on TSMC's 4 nm node packs eight cores and sixteen threads that boost up to 5.0 GHz, plus a Radeon 890M-equivalent iGPU with 16 RDNA 3.5 CUs, and offers a flexible power envelope from 17 W up to 35 W for either tuned performance or extended battery life. Intel's Lunar Lake is giving a good battle with only eight Xe2 cores, showing that the potential Lunar Lake successor can put up a strong fight with next-gen Xe3 cores.

NVIDIA DLSS 4 Comes To Clair Obscur: Expedition 33, Dead Take and Rune Factory: Guardians of Azuma

More than 800 games and applications feature RTX technologies, and each week new games integrating NVIDIA DLSS, NVIDIA Reflex, and advanced ray-traced effects are released or announced, delivering the definitive PC experience for GeForce RTX players. Last week, DLSS 4 with Multi Frame Generation was available at launch in Killing Floor 3 and WUCHANG: Fallen Feathers, and was added to The First Descendant Season 3 demo. This week, DLSS 4 with Multi Frame Generation comes to Clair Obscur: Expedition 33, Dead Take, and Rune Factory: Guardians of Azuma. And 7 Days to Die now features support for DLSS Super Resolution. Plus, you can check out our new EA SPORTS F1 25 video captured with immersive path-traced effects enabled and maxed out, and with performance accelerated by DLSS 4 with Multi Frame Generation. Read on for all the details.

Clair Obscur: Expedition 33 Adds DLSS 4 With Multi Frame Generation
Once a year, the Paintress wakes to paint a cursed number upon her monolith and everyone of that age instantly turns to smoke and fades away. Year by year, that number ticks down. Tomorrow, she'll wake and paint "33." Join Gustave, and his fellow Expeditioners as they embark on a desperate quest to break the Paintress' cycle of death once and for all in Sandfall Interactive and Kepler Interactive's Clair Obscur: Expedition 33. Retracing the steps of dozens of failed Expeditions before them, players will overcome powerful enemies and fight to save the future of their world. Set in a dreamlike world, the story unfolds as fantasy and art converge, backed by an enchanting soundtrack to create a breathtaking story unlike any other.

Arm Q1 FYE26 Revenue Exceeds $1 Billion for Second Consecutive Quarter

Today Arm (NASDAQ:ARM) published a letter to its shareholders containing the company's results for its first quarter of fiscal year 2026 ended June 30, 2025. The infographic below provides the key highlights:
Here are what the results mean and how they were achieved:
  • Record revenues for an Arm first quarter, crossing the $1 billion quarterly revenue milestone for the second consecutive quarter and delivering the second-best revenue quarter ever.
  • Royalty revenue rose 25% year-over-year to $585 million, driven by the growing adoption of the Armv9 architecture, increasing number of chips based on Arm Compute Subsystems (CSS), and use of Arm-based chips in datacenters.

Ayaneo Reveals Next 2: AMD Strix Halo Gaming Handheld Slated for 2025 Launch

Just a few days ago, we got our first taste of what's to come with AMD's Strix Halo Ryzen AI Max+ 395 APU coming to the GPD Win 5, but now Ayaneo has confirmed on YouTube that it will be joining the Strix Halo-powered handheld battle, with the Ayaneo Next 2—only the second confirmed gaming handheld to feature the powerful AMD APU. The new Windows gaming handheld will be the company's next true flagship system, and is technically the successor to the Ayaneo Next, powered by the AMD Ryzen 7 5820U—although Ayaneo does have more powerful handhelds that feature the AMD Ryzen 7 8840U. No official launch date has been announced, but Ayaneo claims that the Next 2 will launch sometime in 2025. Unlike the aforementioned GPD Win 5, the Ayaneo Next 2 will have a built-in battery, which might be slightly more limiting than GPD's backpack design, and it will also feature a dual-fan cooling system—both of which make the prototype assembly shown off in the stream look exceptionally chunky.

The specs for the Ayaneo Next 2 are few and far between, but what Ayaneo has shown off looks like at least an 8-inch display, a "high-capacity battery," and back-mounted paddles similar to those found on the Steam Deck and the Lenovo Legion Go. It's unclear just what the Ayaneo Next 2 will look like, since most of the images shared by the company are labelled as renders of an early design, but the likelihood is strong that it will need a sizeable battery pack for the 16-core APU and Radoen 8060S iGPU. There's also no word on whether there will be a version of the Next 2 with the Ryzen AI Max 385. Ayaneo also teased the potential for a mini PC based on the same motherboard as the upcoming Next 2, which would compete with the likes of Aoostar and the recently announced Mini-ITX motherboards that we covered.
Return to Keyword Browsing
Nov 22nd, 2026 17:44 CST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts