News stories tagged with #NVIDIA
NVIDIA is reportedly planning the launch of the GeForce RTX 6090, 6080, and 6070 in 2027, based on the Rubin architecture with 3-nm manufacturing and GDDR7 memory. Meanwhile, Intel is unveiling a new Core Ultra 400 series based on Nova Lake, featuring up to 52 cores and bLLC cache for desktop CPUs. Both companies are investing in cutting-edge technologies to boost performance in gaming, AI, and everyday computing. These announcements signal a fierce competition between the two tech giants in the coming decade.
AMD's FSR 4.1 shows subtle but noticeable improvements in image quality over FSR 4.0, particularly in detail preservation and motion stability. The technology is limited to RDNA-4 graphics cards and Radeon RX-9000 series and was tested across seven games, including Death Stranding 2, which launched on PC with ray tracing and multiple upscaling solutions. While DLSS 4.5 delivers the best visual quality in Death Stranding 2, FSR 4.1 offers a viable alternative for AMD users.
DLSS 5: Controversy Over AI Graphics Technology in the Gaming Industry
NVIDIA's DLSS 5 has sparked intense debate within the gaming community and among developers. While users and experts criticize the use of AI-generated graphics, early demos in Starfield have received surprisingly positive feedback. Developers like Daniel Vávra defend the technology as a necessary evolution, while indie creators David Oshry and David Szymanski call for improved native GPU performance and more affordable hardware from NVIDIA.
NVIDIA DLSS 5: Revolution or AI Slop? New Technology Sparks Controversial Reactions
NVIDIA unveiled DLSS 5 at GTC 2026, a new technology described as 3D-guided neural rendering that enables fully AI-driven image generation. The system uses 2D color data and motion vectors to create photorealistic lighting and materials in real time, set to launch in autumn 2026 on the new GeForce RTX 5000 graphics cards. While praised for its technical innovation, DLSS 5 has sparked intense criticism for altering developers' artistic vision and creating a uniform, often unexpected visual style.
NVIDIA Unveils Vera-CPU Rack at GTC 2026: New Benchmark for CPU-Only AI Infrastructure
At GTC 2026, NVIDIA unveiled the Vera-CPU Rack, a system designed for CPU-only inference featuring 256 Vera CPUs based on a custom Arm v9.2-A core called Olympus. Each CPU includes 88 cores with SMT, FP8 support, and up to 1.5 TB of RAM. The rack provides 400 TB of LPDDR memory and a total memory bandwidth of 300 TB/s, with individual chips achieving 1.2 TB/s. Optimized for agentic workloads, reinforcement learning, and AI training, the solution is being co-developed with partners like HPE, with the Cray GXC240 supporting up to 640 Vera CPUs per rack.
Nvidia Integrates Groq 3 LPU into Vera-Rubin Platform: A New Era of Low-Latency AI Inference Begins
At GTC 2026, Nvidia announced the integration of Groq’s 3rd-generation Language Processing Unit (LPU) into its new Vera-Rubin-NVL72 platform to dramatically boost AI inference throughput with ultra-low latency. Designed specifically for inference workloads, the LPU leverages high SRAM and internal bandwidth for rapid token processing. This technology complements Nvidia’s existing GPU ecosystem and is deployed in new LPX racks. Partners such as HPE and Giga Computing showcased next-generation AI factories and high-performance computing infrastructure at the event, built around these advancements.
Nvidia has unveiled DLSS 5, a groundbreaking AI-driven technology that combines traditional rendering with neural rendering to deliver more realistic lighting, materials, and resolutions in real time. Available starting autumn 2026 for RTX 50-series GPUs, the technology aims to bridge the gap between real-time and cinematic graphics. Developers retain full control over visual effects while AI significantly enhances image quality and frame rates. The goal is to usher in a new era of photorealistic gaming visuals.
AMD has officially released FSR 4.1 with the Adrenalin Edition 26.3.1 driver for the Radeon RX 9000 series. The update delivers significant image quality improvements, especially at lower upscaling settings, and includes Day-0 support for Crimson Desert and Death Stranding 2. Early analyses reveal that FSR 4.1 offers sharper visuals and better detail fidelity compared to FSR 4, with nearly unchanged performance, and comes notably closer to NVIDIA’s DLSS 4.5 quality in Death Stranding 2.
Intel Xeon 6 Powers NVIDIA DGX Rubin NVL8: High Performance Without NVLink
The Intel Xeon 6 serves as the x86 host CPU in NVIDIA’s new DGX Rubin NVL8 system, despite lacking NVLink support. The processors leverage technologies such as Priority Core Turbo to efficiently manage data processing for the eight Rubin GPUs. This integration combines high-performance computing with advanced AI capabilities, targeting next-generation AI workloads and data center applications. Even without NVLink, the platform remains powerful and scalable.
MSI has released a new BIOS for the MAG X870 Tomahawk WiFi to fix a long-standing PCIe throttling issue on AM5 platforms, which has affected users for months. Meanwhile, the HammerHAI supercomputer at HLRS, powered by HPE and Nvidia, will deliver over 15 AI exaflops and become operational in 2026 as part of the EuroHPC JU initiative. In parallel, Bcachefs 1.37 has launched with Linux 7.0 support, stable erasure coding, and new commands, enhancing data redundancy and system management. These developments highlight key advancements in hardware, AI infrastructure, and storage technology.
TinyCorp Calls for 96-GB AMD GPU While Nvidia Unveils 9-GB RTX 5050
AI startup TinyCorp is calling on AMD to develop a 96-GB RDNA-5 GPU for around $2,500 to power a data center with 3,000 GPUs, while NVIDIA has unveiled the GeForce RTX 5050 with 9 GB of GDDR7 memory and a 130 W TDP, featuring a higher memory bandwidth of 336 GB/s. The card is positioned as the smallest variant of the RTX 50 series and is based on the larger GB206 GPU, with performance expected to match its predecessor despite reduced memory bandwidth. Meanwhile, the 2026 GPU benchmark ranking by PCGH compares 39 graphics cards from Nvidia, AMD, and Intel across four resolutions. TinyCorp's proposal faces skepticism due to the high cost of memory chips and current technological limitations.
NVIDIA Unveils Major Updates to GeForce NOW and Local AI Video Generation at GDC 2026
At GDC 2026, NVIDIA unveiled major updates to GeForce NOW and local AI video generation. The cloud gaming platform now offers enhanced streaming quality, expanded game library access, 90 FPS VR streaming, and CloudXR integration into visionOS for the Apple Vision Pro. Additionally, NVIDIA introduced powerful new features for locally developing AI-generated videos.
Nvidia has released a new beta driver for Linux, significantly boosting GeForce GPU performance through enhanced Vulkan extensions and optimizations. CEO Jensen Huang emphasized Nvidia’s pivotal role in shaping the modern gaming industry and hailed Openclaw as the most important software release of all time. Meanwhile, the Vulkan API introduced the VK_KHR_device_address_commands extension, enabling direct use of device addresses. Early benchmarks of the new 595 driver showed substantial improvements in OpenGL, Vulkan, and GPU compute tasks, particularly on GeForce RTX 50 series cards.
NVIDIA Releases GeForce 595.59 Drivers for Resident Evil Requiem – But with Fan Issues
NVIDIA released the GeForce 595.59 WHQL Game Ready drivers to support Resident Evil Requiem and Marathon, while fixing multiple game bugs and enhancing performance on GeForce RTX 50 series GPUs. The update also improved DLSS and Reflex support. However, the driver was soon removed from download due to a bug causing severe fan issues on RTX 3000 and RTX 5000 series cards, with reports of only one fan being recognized. AMD has also issued a dedicated driver for Resident Evil Requiem and Marathon.
Meta Strikes Multi-Billion Dollar Deal with Nvidia for AI Chips and Infrastructure
Meta has signed a multi-year, up to $50 billion deal with Nvidia to procure millions of Blackwell and Rubin GPUs, along with additional components such as CPUs, switches, and Vera processors for its AI data centers. The partnership also includes the development of complete rack systems and strengthens Meta's position in the AI race, displacing AMD as the primary supplier. This comprehensive agreement highlights Nvidia's strategic dominance in AI infrastructure.
NVIDIA Sells Remaining ARM Stakes: Strategic Move for AI Infrastructure and Agentic AI
NVIDIA has sold its remaining $140 million stake in Arm Holdings to free up capital for future investments in AI infrastructure. This follows the failed $40 billion acquisition in 2020, which was blocked by regulators over competition concerns. While ending its ownership, NVIDIA retains licensing rights to Arm’s IP and instruction set for its Grace and Vera CPUs. The move is less a retreat from Arm and more a strategic diversification to adapt to the evolving demands of the Agentic-AI era, where x86 platforms are increasingly important.