Tag: Nvidia GPU

  • Adobe Premiere Pro 25.5 Adds 90+ New GPU-Accelerated Effects

    Adobe Premiere Pro 25.5 Adds 90+ New GPU-Accelerated Effects

    Key Takeaways

    1. Premiere Pro 25.5 will launch on September 12, 2025, with over 90 new GPU-accelerated effects and transitions.
    2. The software is popular among students and independent filmmakers due to its integration with other Adobe programs like After Effects and Photoshop.
    3. Key features include optimized timeline responsiveness, visible audio waveforms, multi-track clip fade adjustments, and a “Surprise Me” button for random effects.
    4. Enhanced integration with After Effects allows for easier creation of effects without constant switching between programs.
    5. A powerful Nvidia GPU is recommended for running Adobe Premiere Pro effectively.

    Adobe has revealed the launch of Premiere Pro 25.5, which features over 90 new and refreshed effects tailored for video editing. The software will be available starting September 12, 2025, coinciding with the company’s showcases at the 2025 IBC expo.

    Popular Editing Software

    Premiere Pro has established itself as a favored, subscription-based video editing tool for both Apple Macs and Windows PCs. Many students and independent filmmakers rely on its seamless integration with other widely-used Adobe programs like After Effects and Photoshop. Nevertheless, it has not managed to outshine Avid Media Composer, which remains the go-to software for Hollywood blockbusters grossing more than $500 million.

    Exciting New Features

    The standout feature of this update is the introduction of over 90 GPU-accelerated, real-time effects and transitions. To enhance the efficiency of adding effects, the timeline has also been optimized for better responsiveness through GPU acceleration. Adobe has added visible audio waveforms while dragging clips and multi-track clip fade adjustments to boost productivity. A novel “Surprise Me” button will randomly apply an effect to videos, providing a spark for creativity.

    Enhanced Integration

    Additionally, more capabilities from Adobe After Effects have been embedded directly into Adobe Premiere Pro. This allows editors to create simpler effects, like animating text and logos, without the need for constant round-tripping between the two programs.

    Source:
    Link

     

    Tags:
  • GALAX GeForce RTX 5090D HOF OC LAB Listed on eBay for $5,500

    GALAX GeForce RTX 5090D HOF OC LAB Listed on eBay for $5,500

    Key Takeaways

    1. Exclusive Availability: The GeForce RTX 5090D HOF OC LAB edition is exclusive to China and designed to meet U.S. export regulations, making it a rare find outside the region.

    2. High Price Tag: Prices for the RTX 5090D on eBay have reached around $5,500, significantly higher than the regular RTX 5090’s price, which is typically up to $3,000.

    3. Impact of Supply Shortage: The discontinuation of the RTX 5090D’s production by NVIDIA due to stricter export rules has led to a scarcity, driving up prices and creating a collectible status.

    4. Overclocking Capabilities: The HOF OC LAB edition supports extreme overclocking with a power draw potential of up to 2,000 watts, achieving impressive benchmark scores and GPU frequencies.

    5. AI Performance Limitations: The RTX 5090D has reduced AI workload capabilities compared to the standard RTX 5090, adhering to export guidelines, which diminishes its performance in AI tasks.

    GALAX’s GeForce RTX 5090D HOF OC LAB edition has been spotted on eBay with prices hitting around $5,500, making it one of the priciest GB202-based NVIDIA GPUs ever sold.

    Exclusive Graphics Card

    This 5090D model is exclusive to China, designed to fit U.S. export rules. The HOF edition, which is a high-end version, is priced much higher than the main RTX 5090. The regular versions of the RTX 5090 usually cost up to $3,000 at many US stores, even though Nvidia’s suggested retail price for the Founder’s Edition is $2000.

    Surprising Price Point

    The high cost of this GPU is quite shocking, especially since the RTX 5090D is merely a slightly downgraded version of the RTX 5090 in terms of AI workload capability. This could be linked to the limited supply in China, which elevates the GPU to a sort of ‘collectible’ status.

    Market Changes

    The shortage and rising prices for the RTX 5090D began after NVIDIA stopped producing the card in Q2 2025, due to stricter U.S. export rules. They decided to cease all exports of the RTX 5090D to China. Meanwhile, already placed orders were “temporarily canceled,” while Chinese industry insiders claimed that the RTX 5090D series was “actually done and the graphics cards wouldn’t be available in the second quarter.”

    Power and Performance

    The halt on exports led to scarcity, particularly for the limited GALAX HOF OC LAB edition, which has a dual 16-pin power connector setup that lets it draw a theoretical 1200 watts. Special BIOS updates allow users to increase this power limit for extreme overclocking, pushing the power draw to as high as 2,000 watts.

    The HOF OC LAB edition has set numerous records, with overclocking fans reaching GPU frequencies of 3,650 MHz using liquid nitrogen. In benchmarks like 3D Mark Port Royal, it scored a record 47,469 points, while it achieved 38,327 points in Uniengine Superposition and a record time of 39.434 seconds in GPUPI v3.3.

    High-End Features

    The premium price of the HOF OC LAB is backed by a 38-phase VRM setup, with 28 phases dedicated to the GPU itself. The GPU is constructed on a 12-layer PCB and shares the same memory specs as the standard RTX 5090, featuring 32 GB of GDDR7 memory running at 28 Gbps.

    In terms of AI performance, the RTX 5090D variant falls short compared to the base RTX 5090, with a reduction from 3,352 to 2,375 AI TOPs, a decrease of 29.15%. This limitation was put in place to adhere to U.S. export guidelines.

    Conclusion

    The current price of the RTX 5090D GALAX’s HOF OC LAB edition highlights not just how expensive high-end overclocking cards can become, but also illustrates how global political factors can impact gamers, enthusiasts, and those who don’t use AI performance.

    Source:
    Link

    Tags:
  • Samsung HBM4 Memory Begins Trial Production, Mass Production by 2025

    Samsung HBM4 Memory Begins Trial Production, Mass Production by 2025

    Samsung Electronics has completed the design for its upcoming High Bandwidth Memory (HBM4) logic chips and has begun trial production using a 4nm process. This is an important move that paves the way for mass production in the latter part of 2025, which is roughly six months earlier than the initial timeline set for 2026. This acceleration might help Samsung secure significant orders for future GPU platforms from Nvidia.

    Production Details

    Reports suggest that Samsung’s storage unit has finalized the logic chip designs, while their foundry unit is actively engaged in trial runs at the 4nm level. The HBM4 production takes place in a specialized facility known as D1c, which utilizes a cutting-edge 10nm DRAM process. Notably, Samsung has bypassed the traditional D1b development phase to expedite the process.

    Performance Enhancements

    Technical insights revealed at ISSCC 2024 indicate that HBM4 will provide a substantial increase in performance, achieving data transfer speeds of up to 2 TB/s—around 66 percent faster than the previous HBM3E. Additionally, it will feature a 2048-bit interface operating at 6.4 GT/s and a capacity increase up to 48 GB, which is 33 percent larger than its predecessor.

    Competitive Landscape

    On the other hand, SK Hynix, Samsung’s main competitor in the HBM market, is also reportedly accelerating its own HBM4 development to meet the same 2025 deadline. Analysts from Hanwha Investment & Securities predict that SK Hynix will maintain its market share lead and may be the first to deliver HBM4 samples to its clients.

    Beyond Nvidia, Samsung is tailoring HBM4 solutions for clients like Microsoft and Meta. This memory technology is crucial for Nvidia’s forthcoming “Rubin” GPU platform, expected in 2026, which will incorporate eight HBM4 chips and depend on TSMC’s 3nm process.

    Market Trends

    Samsung’s existing range of HBM products is experiencing significant growth. Sales in the third quarter of 2024 increased by over 70 percent compared to the previous month. HBM3E products, including both 8-layer and 12-layer versions, have entered mass manufacturing and are projected to comprise about half of Samsung’s total HBM sales by the end of 2024.

    Source: Link

    Image 1
    Tags: ,