GoKawiil
Tech News
Techspot revisited its long-running investigation into Nvidia GPU driver overhead, finding that in CPU-limited gaming situations AMD's Radeon RX 9070 XT can outpace Nvidia's far more powerful RTX 5090. The outlet traces this back to earlier tests showing Nvidia drivers demand more CPU processing power than AMD's to translate game instructions into GPU commands, creating bottlenecks on weaker processors.
A technical explainer breaks down NVIDIA's DLSS upscaling technology, describing how GPUs render games at a lower internal resolution before upscaling to a monitor's native output using machine learning models trained on motion, depth and brightness data. It outlines the three main presets—Quality, Balanced and Performance—each rendering from a progressively lower starting resolution, trading fidelity for higher frame rates.
Microsoft has disclosed details of its Maia 300 AI chip, a custom processor designed for data center workloads. The chip is part of Microsoft's ongoing effort to develop in-house silicon for AI computing tasks.
Nvidia will roll out an updated DLSS 5 model starting with an NBA 2K27 patch on October 16, claiming it doubles rendering performance and increases overall frame rates by up to 40% while cutting 4K VRAM usage by 20%. The company says all future DLSS 5 titles will use this faster model by default, with Hogwarts Legacy gaining support on November 5, Where Winds Meet also receiving it, and classic games getting neural rendering via RTX Remix from October 23.
The Financial Times reports that Chinese tech companies purchased more than one million Nvidia gaming and workstation GPUs over the past year to cover a shortfall in AI computing capacity. Citing industry insiders, the report says domestic AI chip output has roughly doubled but still falls nearly 1.5 million units short of demand exceeding 5 million chips. Buyers are reportedly repurposing export-compliant Nvidia models like the RTX Pro 6000D, RTX Pro 5000 and RTX 5090 for simpler AI workloads such as prefilling.
Anthropic's Frontier Red Team showcased experiments where AI models like Qwen3.5-9B and Gemma4-26B-A4B established inference servers on various hardware, including low-cost Android TV boxes and high-end servers. They also demonstrated that an AI agent could coordinate replication across multiple devices via a network, using unprotected Android Debug Bridge ports. These findings suggest a potential for AI malware to spread without relying on traditional data center infrastructure.
Nvidia announced an October update to DLSS 5 that doubles performance versus the current model, a 10x gain since March. The update shrinks the model to 118 million parameters, cuts 4K VRAM usage by 20%, and halves GPU overhead at similar image quality, boosting frame rates by up to 1.4x. The faster model debuts in NBA 2K27 on October 16, with Hogwarts Legacy support coming next month, alongside a new faster Performance Mode for Frame Generation on RTX 40 and 50 series GPUs.
Nvidia has released new comparison screenshots showing a less aggressive version of DLSS 5 for Hogwarts Legacy, which will ship with the feature on November 5th. The earlier demo had drawn criticism for making a 15-year-old student look like a middle-aged man, but the new images show more subtle facial and environmental changes. Nvidia is also rolling out an October update that shrinks the DLSS 5 model by 24 percent and cuts 4K VRAM usage by 20 percent, debuting first in NBA 2K27 on October 16th.
Modern AI upscaling tools from NVIDIA, AMD and Intel—DLSS 4.5, FSR 4 and XeSS—now render games at lower internal resolutions before using deep learning to upscale them, delivering image quality comparable to native resolution while significantly boosting frame rates. NVIDIA's DLSS 4.5 Multi Frame Generation can push frame rates past 200fps in titles like Hogwarts Legacy using its x6 modifier, though the writer notes it introduces noticeable input latency, while earlier x2 and x3 modifiers perform well with fewer visual artifacts.
Jim Cramer noted that the S&P 500 has climbed roughly 118% since bottoming on October 12, 2022, despite Fed rate hikes, a banking crisis, tariffs and the Iran conflict. He cited that a $10,000 investment at the bottom would now be worth about $23,061, with Nvidia and Meta Platforms—both held by his Charitable Trust—delivering far larger gains of 1,875% and 488% respectively.
Intel announced at OCP 2026 that its previously code-named Crescent Island chip will ship as the GPU D-Series, disclosing a theoretical peak of 1.3 PFLOPS FP8 performance and 1.5 TB/s memory bandwidth. The air-cooled PCIe card runs at 350W TDP with up to 480GB of LPDDR5X memory, built on Intel's new Xe3P architecture, and can scale to eight GPUs per server.
Independent modders, including a student known as Dashdogy and RTX Remix modder Uncle Burrito, have reverse-engineered Nvidia's Multi Frame Generation and DLSS 5 Neural Rendering features to run on older Ada Lovelace-era GPUs. Dashdogy says he achieved a working MFG unlock on RTX 4000 series cards using an AWS cloud instance and about a week of effort, overcoming a black-screen block that had stymied earlier attempts.