49.jili

49.jili

‘Several Hundred’ North Korean Casualties in Russia’s Kursk – U.S. Official

File: A stack of various daily newspapers. Photo: Sebastian Kahnert/dpa-Zentralbild/ZB (Photo by Sebastian Kahnert / dpa-Zentralbild / dpa Picture-Alliance via AFP) File: A stack of various daily newspapers. Photo: Sebastian Kahnert/dpa-Zentralbild/ZB (Photo by Sebastian Kahnert / dpa-Zentralbild / dpa Picture-Alliance via AFP) JOHANNESBURG - Once upon a time, people learnt about and understood their world through a newspaper. People would sit at the table or the couch reading the printed words. The first few pages generally contained the most important news of the day, followed by opinion pieces, then something lighter like lifestyle or travel and finally, sport. Before the sport pages came the classified section. It was akin to what we now experience on Facebook’s marketplace. These days, newspapers are disappearing. It is a trend that started many years ago and by the end of this year, a few more titles will disappear, either forever or just in print form. This week marked the end of City Press in the printed form. It will continue online. Mondli Makhanya, the editor, said the plan was for the title to go big on stories that were important. He accepts that there is strong competition online but believes maintaining good journalistic values will give the title the competitive edge. These values include accuracy, trust, and bold reporting. Another title to ditch its print edition and embrace the digital world was the Daily Sun – a newspaper that focused on the ordinary person. Amos Mananyetso, the editor, said: ‘Upon establishment, the philosophy of the Daily Sun was clear. It was a newspaper for the man in the blue overall, a factory worker, a taxi driver, a domestic worker, or a retail worker. These were commuters who bought their papers at taxi ranks, train stations, or bus stops’. ‘But suddenly, on 26 March 2020, that commuter market was not there. The country was on lockdown and suddenly the Daily Sun did not have a paper to sell to someone,’ he said. Most newspapers were not able to win back readers after COVID-19. The world had changed. Money was tight. And there were other platforms to get information. Mananyetso said there was a need to shift gears – moving from a traditional print media publication and into the online space. He said there were also advantages as journalists could use audio and video to tell their stories. ‘It is not the end of the word. It is a changing word. We are just using a different medium to tell the same stories of our community and the suffering of our people’. During the course of 2024, the Competition Commission conducted the Media and Digital Platforms Market Inquiry to investigate the plight of smaller news organisations in the digital age. The commission said, ‘The Publishers Support Service submitted that the transition to digital news consumption and advertising had resulted in all publishers seeing a massive decline in advertising revenue, and an increase in costs as newsrooms needed to devote resources to their digital presence. This transition leaves the publishers in a precarious financial position despite cutting costs.’ Additionally, the commission said the loss of advertising revenue for publishers meant they were unable to make the necessary investments. This in turn had the potential to reduce the quality and diversity of content. According to the commission, a "smooth transition" into digital distribution required investment in technology and highly skilled developers, which came at a price. A recent survey referred to by the Commission showed that 77% of people interviewed got their news from social media platforms like Facebook and TikTok. Those surveyed were attracted to these platforms because it of convenience and immediacy. However, 48% of respondents, still turned to traditional media because they saw it as dependable and in-depth. Whether traditional media organisations will have a happy ending and remain credible and relevant online remains to be seen. By Zandile KhumaloColor Reflective LCD Display Market Outlook and Future Projections for 2030

Slim-Llama reduces power needs using binary/ternary quantization Achieves 4.59x efficiency boost, consuming 4.69–82.07mW at scale Supports 3B-parameter models with 489ms latency, enabling efficiency Traditional large language models (LLMs) often suffer from excessive power demands due to frequent external memory access - however researchers at the Korea Advanced Institute of Science and Technology (KAIST), have now developed Slim-Llama, an ASIC designed to address this issue through clever quantization and data management. Slim-Llama employs binary/ternary quantization which reduces the precision of model weights to just 1 or 2 bits, significantly lowering the computational and memory requirements. To further improve efficiency, it integrates a Sparsity-aware Look-up Table, improving sparse data handling and reducing unnecessary computations. The design also incorporates an output reuse scheme and index vector reordering, minimizing redundant operations and improving data flow efficiency. Reduced dependency on external memory According to the team, the technology demonstrates a 4.59x improvement in benchmark energy efficiency compared to previous state-of-the-art solutions. Slim-Llama achieves system power consumption as low as 4.69mW at 25MHz and scales to 82.07mW at 200MHz, maintaining impressive energy efficiency even at higher frequencies. It is capable of delivering peak performance of up to 4.92 TOPS at 1.31 TOPS/W, further showcasing its efficiency. The chip features a total die area of 20.25mm2, utilizing Samsung ’s 28nm CMOS technology. With 500KB of on-chip SRAM, Slim-Llama reduces dependency on external memory, significantly cutting energy costs associated with data movement. The system supports external bandwidth of 1.6GB/s at 200MHz, promising smooth data handling. Slim-Llama supports models like Llama 1bit and Llama 1.5bit, with up to 3 billion parameters, and KAIST says it delivers benchmark performance that meets the demands of modern AI applications. With a latency of 489ms for the Llama 1bit model, Slim-Llama demonstrates both efficiency and performance, and making it the first ASIC to run billion-parameter models with such low power consumption. Are you a pro? Subscribe to our newsletter Sign up to the TechRadar Pro newsletter to get all the top news, opinion, features and guidance your business needs to succeed! Although it's early days, this breakthrough in energy-efficient computing could potentially pave the way for more sustainable and accessible AI hardware solutions, catering to the growing demand for efficient LLM deployment. The KAIST team is set to reveal more about Slim-Llama at the 2025 IEEE International Solid-State Circuits Conference in San Francisco on Wednesday, February 19. Almost everything you want to know about Large Language Models How organizations can make the most of LLMs AI chip built using ancient Samsung tech is as fast as Nvidia A100 GPU