New technical details have appeared online about the upcoming PlayStation portable console, which Sony has not yet officially announced. According to a leak from insider KeplerL2, the device will be built on AMD gfx13 hybrid architecture (RDNA5/UDNA) and will be the first APU from AMD to support hardware AI upscaling.

The console will receive 16GB LPDDR5X RAM, since LPDDR6 is not yet ready for mass use. The memory bandwidth will be a third of the base PlayStation 5, but this is offset by improved data compression and the presence of second level cache (4 MB L2) и special MALL-cache of 16 MB. This cache is not available in current desktop PlayStation models.

Sony is already rumored to be integrating it into the PlayStation 5 and PS5 Pro firmware energy-saving mode, which will reduce performance - it may be related to a future portable device. While Sony is keeping quiet, Microsoft is already promoting the ROG Xbox Ally, hinting at Xbox's development in the portable PC segment.

Add a comment

MSI has released a new BIOS version for motherboards of the series X870 and B850, socket based AM5. Update on the base AGESA 1.2.0.3e brings several key improvements at once and hints at support for future processors AMD Ryzen, probably on architecture Zen 5.

The most important innovation is support 256GB DDR5 RAM, implemented through four 64 GB modules. Chips from are supported Samsung, Hynix and Micronand improved compatibility in 2DPC 1R configurations. This is an important step for workstations and enthusiasts where memory capacity is critical.

Additionally, MSI has optimized overclocking for Samsung 4Gx8 chips, which improves the stability of high-frequency assemblies. But the main thing is that the update description directly states: added support for future CPUs, which actually confirms the platform’s readiness for new generation Ryzen, possibly the Ryzen 9000 or Ryzen 9000G series.

The BIOS is now available for download and has a number 7E59v2A73. File size - 12,99 MB. Release date - 11 June 2025 year.

Key phrases:
BIOS AGESA 1.2.0.3e available for AM5, support 256GB DDR5, optimization for Samsung 4Gx8 chips, Improved 2DPC 1R compatibility, added support for future Ryzen CPUs

Add a comment

Two days before the official announcement at ISC 2025 in Hamburg, the characteristics of the accelerator series became known AMD Instinct MI350, built on architecture cDNA 4. At the exhibition stand were seen as MI350X, as well as a more powerful model MI355X, both of which are available in configurations with 8 GPU per node.

The series received support new AI formats FP8, FP6 and FP4, which allows for double the performance compared to FP8. According to the data from the showcase, FP4/FP6 reaches 18-20 PFLOPS, FP8 - up to 10,1 PFLOPS. It twice as high early benchmarks and is in line with the figures previously attributed to FP4/6.

The memory capacity is 288 GB HBM3E, and the throughput reaches 8 TB/s per OAM module... Model MI355X focused on liquid cooling, with a thermal package of up to 1400 WWhereas MI350X uses traditional solutions with power consumption of up to 1000 W.

Despite the impressive performance, the series does not offer scalability beyond 8 GPU, as in solutions NVIDIA NVL36/NVL72. This feature will only appear in a future series. MI400, whose release is expected in 2026. Confirmation of all characteristics is expected on June 13 at 18:30 at the official AMD presentation.

Add a comment

Modern business demands not only high performance from technology, but also reliability, portability and innovative features. ASUS offers a wide range of ExpertBook notebooks that meet all these requirements. From lightweight and powerful devices for executives to reliable and functional solutions for small and medium businesses - in the ASUS ExpertBook line you will find the optimal notebook for any needs. 

ExpertBook B9 – the embodiment of perfection  

ExpertBook B9  – the lightest (weighs only 990 grams) 14-inch business notebook with an incredible OLED display with WQXGA+ resolution (2880 x 1800), ultra-thin bezels and a 16:10 aspect ratio. In its maximum configuration, the model  (B9403CVAR)  Equipped with an Intel Core 7 processor and 64 GB of RAM, the sleek body of the device is made of magnesium-lithium alloy, which ensures lightness, portability and durability. With an eco-friendly design, a bright display, SI tools for improving the quality of video conferencing, the latest processor and numerous security features, this laptop is an ideal choice for business executives and those professional users who often travel on business. 

ExpertBook B1 – modern, reliable, easy-to-maintain laptops 

Inexpensive and versatile laptops  ExpertBook B1  are oriented to the needs of businesses of any scale. configurations with both Intel and AMD processors, with screen diagonals of 14 and 15,6 inches. For example, the new model  (B1503CVA)  can be equipped with processors up to Intel Core 7  150U  и  up to 64 GB of RAM. ExpertMeet, Copilot in Windows and intelligent noise reduction.

ExpertBook B3 – powerful, intelligent and secure devices

Laptops  ExpertBook B3  – is a powerful solution with support for modern artificial intelligence technologies. Models with 14 and 16-inch diagonals with an aspect ratio of 16:10, thin frames and an anti-glare screen coating are available. Equipped with Intel Core processors Ultra 7 with a neural processing unit (NPU) and a modern graphics core, these laptops provide high performance and comfortable work with applications using AI technologies. ExpertBook B3 is equipped with full security features, including hardware. This includes a Full HD IR camera, fingerprint and smart card scanners, a TPM module, a shutter for blocking the webcam and a slot for a Kensington Nano lock.

ExpertBook P5 – style and performance for small and medium businesses 

The ASUS business lineup includes the B-series for corporate users  P-series , aimed at the needs of small and medium-sized businesses, entrepreneurs and home office workers. Laptops in this series, including the flagship model  ExpertBook P5  based on Intel Core Ultra (Series 2) sets new standards for AI-driven productivity with the integration of ExpertMeet AI for more efficient online meetings (with translation and summarization) and ExpertPanel for automatic system optimization. The P-Series design combines lightness, style (particularly the full metal chassis of the P5 and P3) and ergonomics, including larger keys and an improved ExpertCool cooling system. In addition, these laptops feature robust multi-layered security down to the BIOS level (NIST SP800-155 compliant), offer comprehensive ASUS support, and meet MIL-STD-810H military-grade durability standards. With the use of recycled aluminum alloy, steel, and even recycled magnets, the ExpertBook P5 demonstrates a significant improvement in environmental friendliness and is a step toward a circular economy. 

The ExpertBook P5 laptops boast a 2,5K display with a 144Hz refresh rate. With two SSD slots, up to 32GB of RAM, and a high-capacity battery that supports up to 28 hours of battery life, these laptops give users maximum productivity in any situation. At the same time, the 14-inch ExpertBook P5 (P5405) weighs only 1,27 kg. 

ASUS ExpertBook laptops feature a 180-degree flip-up display, with some models (Flip series) offering a 360-degree flip-up display. They feature a mechanical webcam shutter and come with productivity-enhancing apps like ExpertMeet AI for video conferencing, including automatic translation, person identification, and meeting. 

ASUS Chromebook for Business and Education  

ASUS also offers Chromebooks for business and education. Lightweight Chrome OS devices like the 30g Chromebook CM610, the powerful Chromebook Plus CX14, and the affordable 12-inch Chromebook CR12 are great options for all-day computing. 

All ASUS business notebooks are tested according to the US military standard MIL-STD 810H and rigorous internal ASUS tests. For corporate customers, customization of configurations and appearance is possible, in particular, the application of the company logo on the cover of the notebook. ASUS ExpertBook notebooks are reliable and productive solutions for business, offering a wide range of functions for modern professionals. 

Add a comment

Nintendo has officially confirmed that console sales Switch 2 Surpasses 3,5 Million Units all in its first four days of release, making it the fastest-selling platform in Nintendo history, surpassing even the success of the original Switch.

According to analysts at DFC Intelligence, Nintendo is capable of selling up to 15 million Switch 2s by the end of 2025, despite possible trade restrictions from the US. Without the impact of tariffs, the estimate reaches 17 million units per year. Experts believe that the all console has a chance to break the world record for the speed of distribution in the first year of sales.

Early reviews are also positive. According to technical reviews, the Switch 2 offers significant improvements over the original model: a more robust case, output up to 4K in docking mode, a larger screen with high definition and improved quality of life. However, the main drawback is considered modest starting catalog of games and the need to purchase additional MicroSD Express memory card, since the underlying storage is limited.

Despite this, the console has already received the title Nintendo's best UK launch ever, although it still couldn't beat the PS5 or Xbox Series X/S by the same criterion. Nintendo is betting on a gradual expansion of the library and support for major releases in 2025, which should further accelerate the pace of sales.

Add a comment

Intel continues to develop its gaming performance monitoring tool on Windows — PresentMon (GameTechDev/PresentMon repository). The latest updates have added new FRAME_TYPE and APPLICATION_FPS metrics, improved command line processing and improved ETW event analysis mechanism for more accurate tracking of delays and frames.

The key feature was the clarification on hardware accelerated planning GPU (Hardware-Accelerated GPU Scheduling): When HWS is enabled, some key metrics such as msGPUActive, msUntilRenderStart, msUntilRenderComplete и msGPUVideoActive, can be overestimated by ~0,5–1 ms. This is important to consider when optimizing performance. In addition, metrics for OpenGL and Vulkan demonstrate reduced accuracy, as they have less ETW telemetry available.

Intel emphasizes that PresentMon requires the following to work correctly: administrator rights and membership in the “Performance Log Users” group — otherwise you will get a trace startup error. Without these conditions, the utility will not be able to record data even with a simple CSV log.

PresentMon includes three key components:

• PresentMon Console — console application for collecting CSV data,

• IntelPresentMon Service — service with API and telemetry GPU/CPU,

• IntelPresentMon GUI - convenient graphical interface with overlay and real-time metrics.

The platform is already used by Intel and is also actively integrated into third-party solutions, such as OCAT, CapFrameX, Microsoft PIX and NVIDIA FrameView, making it one of the multifunctional tools for analyzing graphics performance.

Add a comment

Micron has officially launched Deliveries of new generation HBM4 memory key customers developing AI platforms. The new 36GB stacks in a 12-layer layout demonstrate significant gains in throughput and energy efficiency, which is especially relevant for data centers and large-scale AI systems.

Developed on the basis of the 1β (1-beta) DRAM process technology, HBM4 delivers over 2,0 TB/s of throughput per stack thanks to the 2048-bit interface. This is on more than 60% higher, than the previous generation HBM3E. This leap allows for efficient processing of tasks related to training and inference of large language models, as well as supporting the operation of AI accelerators in critically loaded sectors.

Micron also announced improvements energy efficiency by more than 20%, which allows reducing overall energy consumption in data centers while maintaining high computing density. The company emphasizes that the release of HBM4 is strictly synchronized with the readiness schedules of customers, including developers of advanced AI platforms. Among them, the following have already been named NVIDIA Vera Rubin and AMD Instinct MI400, where the new memory will be used.

Micron reports that The products are already being delivered to customers and are being actively tested., and the first full-scale deliveries will begin as production scales up. This is another step towards strengthening the company's position in the AI ​​memory market. HBM4 to be a key component of future AI accelerators, especially in the medical, financial and transport segments, where both speed and energy efficiency are critical.

Add a comment

TrendForce analysts published a rating Top 10 Contract Chip Manufacturers by Q1 2025 Revenue, which demonstrates the continued dominance of TSMC and a noticeable decline in Samsung Foundry's positions. The Taiwanese company once again showed the best result - $ 25,5 billion in revenue, occupying 67,6% of the global contract manufacturing marketDespite a 5% decline in quarterly revenue, TSMC's position remains strong thanks to contracts with NVIDIA, Apple and AMD, especially in the AI ​​and server space.

Against this background Samsung Foundry Loses Share and Demand, despite its former status as TSMC's main competitor. At the end of the quarter, its share fell to 7,7% with revenue falling by 11,3% ($2,89 billion). The reasons cited are insufficient flexibility when switching to new technological processes and contract fulfillment difficulties. Meanwhile, China's SMIC increased quarterly revenue by 1,8%, reaching $2,25 billion and 6% market share, closing the gap with Samsung.

It is also worth noting the stability of UMC (1,76 billion, -5,8%) and GlobalFoundries' sharp decline (1,58 billion, -13,9%), which continue to lose competitiveness. Among the newcomers gaining momentum are HuaHong Group and Nexchip, driving growth in China's semiconductor sector.

The total revenue of all participants in the rating amounted to $36,4 billion against $38,5 billion in Q4 2024 — the decline in the first quarter was 1%, despite the overall growth of the AI ​​sector. Against this backdrop, TSMC continues to expand production in the US and is preparing to launch new N2 and A16 process technologies, while Samsung may yield second place to Chinese competitors as early as next quarter.

Add a comment

NVIDIA CEO Jensen Huang meets with UK Prime Minister Kiir Starmer as part of the opening of London Tech Week. The event discussed the future of artificial intelligence and its impact on the economy. Huang said that over the past decade AI systems' performance has increased a million-fold, emphasizing the pace of technological progress. Although the head of NVIDIA did not specify whether he meant hardware or software advances, everything points to an increase in power GPU and data centers.

An example of such a leap is the new generation of chips. Blackwell B200, which delivers up to 20 teraflops of inference—while GPU The 100 P2016 delivered just 19 teraflops of FP16. Blackwell, meanwhile, 42 times more energy efficient when processing each token. While this data does not provide an exact 1:1 ratio, it does support the idea that the overall performance of modern AI clusters is indeed many times greater than the performance of ten years ago.

At the same time, the UK is still lagging far behind the US in terms of infrastructure. The country is building a system Isambard-AI with 5500 Grace Hopper accelerators, while the American xAI already uses 200 Hoppers and plans to create a cluster on million GPU Blackwell. However, NVIDIA will invest in local development: the company will open in Britain AI Research Center, will support projects in robotics, environmental modeling and 6G, and will launch educational program on AI for developers. The focus will be on upgrading skills, collaborating with government agencies, and creating a secure environment for testing AI in the financial sector.

Add a comment

Financial performance AMD in Q2025 XNUMX will come under serious pressure due to US restrictions on the export of MI308, a specialized GPU for the Chinese market. As stated in the Q1 report, the company expects write-off of inventory and contractual obligations totaling approximately $800 million, which will directly affect its margins and profit for the current quarter.

According to AMD's own forecasts, Gross profit in the second quarter will fall to 43%, whereas without taking these one-off losses into account it could have been around 54%. Despite the fact that such expenses are usually treated as one-off and ignored when assessing operational efficiency, Most Wall Street analysts included these costs in their forecasts., which significantly adjusted earnings per share (EPS) expectations.

KeyBanc analyst John Vinh said his team was forced to AMD's Q2 earnings forecast was cut to $0.47 per share from the previous estimate of $0.90, to stay within the consensus. By comparison, the current average Wall Street forecast is $0.57. This change in perception suggests a possible doubts about the "one-time" nature of losses — Investors fear that the restrictions on MI308 will remain in place for a long time.

Meanwhile, U.S. officials have hinted at a possible easing of export restrictions in talks with China. However, National Economic Council Chairman Kevin Hassett said expansion of supplies GPU to China is unlikelyand The limitations on MI308, like on NVIDIA H20, will remain in place for the foreseeable future.

Add a comment