Tech News
← Home  ·  All topics

Memory

269 GoKawiil briefs on this topic · RSS for memory

Developer details 'double headed stack allocator' technique from Surreal Software's cancelled game

A programmer who worked on animation systems for the cancelled open-world title 'This is Vegas' at Surreal Software (Midway Seattle) around 2010 describes a memory management technique called a stack allocator, also known as a bump allocator. The method pre-allocates a fixed block of memory and uses a simple integer pointer to track allocation and deallocation, replacing costly dynamic memory calls that had become a profiling bottleneck in per-frame animation blending.

Opinion piece questions corporate control and hype behind Rust's rapid adoption

A commentator argues that Rust's growing adoption carries risks, pointing out that the Rust Foundation was founded by Amazon, Google, Huawei, Microsoft and Mozilla, and that a board seat costs $325,000 per year. The piece also criticizes the trend of rewriting existing, mature software projects in Rust purely for the sake of using the language.

PyTorch gains knapsack solver option cutting memory-planning RAM use 20x

PyTorch's activation memory budget solver, which decides which operations in a computation graph to save versus recompute to fit a memory budget, has a new configurable algorithm called dp_knapsack_sliding_hirschberg. Switching the setting from the default dp to this new option reportedly reduces peak RAM usage during memory planning by roughly a factor of 20. The feature is described as a drop-in configuration change via torch._functorch.config.activation_memory_budget_solver.

Leaker: Apple supply-chain orders point to large iPhone 18 stockpile

Apple delayed the standard iPhone 18 until early next year while launching only the iPhone 18 Pro, Pro Max and iPhone Duo this cycle. Leaker Fixed Focus Digital reports that component inventory levels tied to the base iPhone 18, set to begin production in late November, suggest a substantial stockpile is being built.

Japanese retailers charge up to 41% more than Amazon US for Nvidia RTX 50-series GPUs

A Reddit user posted photos showing Japanese retail prices for Nvidia RTX 50-series graphics cards running 5.66% to 41.1% higher than equivalent Amazon US listings, with the RTX 5060 Ti 8GB showing the steepest markup. The comparison, converted at current exchange rates, found RTX 5080 cards averaging a 12.71% premium over US pre-tax prices, with the gap narrowing to under 4% when an 8.5% US sales tax is factored in. Notably, the RTX 5090 was absent from the Japanese listings entirely.

Leaker says AMD RDNA 5 GPUs due second half of 2027, flagship later

Leaker Kepler_L2 says AMD's RDNA 5 graphics card lineup will launch in the second half of 2027, with a possible unveiling at Computex 2027 in June. Unlike past generations, the mid-range AT2 chip may arrive first in 2027, while the flagship AT0 and lower-end AT3/AT4 chips could follow in 2028. The leaker also said upcoming 4GB GDDR7 memory chips likely won't be used in these consumer cards, being reserved instead for AI and enterprise products.

SK Hynix, Samsung, SoftBank shares drop sharply amid Asia chip sell-off

SK Hynix fell more than 8% and Samsung Electronics dropped 7.9%, both hitting their lowest intraday levels since August, as a US semiconductor sell-off spread to Asian markets. SoftBank Group shares sank over 8% in Japan, dragging down chip-related names including Tokyo Electron, Advantest, Lasertec, Disco and Renesas Electronics. The moves followed a 0.44% Nasdaq decline and a 3.9% drop in Micron Technology after China's ChangXin Memory Technologies reported progress on a new chip architecture.

Apple's smartphone share tops 20% in Q3 as global shipments fall 20%, says Counterpoint

Counterpoint Research reports Apple captured 22% of global smartphone shipments in Q3 2026, up from 18% a year earlier, even as total industry shipments fell 20% year-over-year to their lowest third-quarter level in 14 years. Samsung held the top spot with 23% share, while Xiaomi, OPPO and vivo all saw their shares decline. Apple's own shipment volume fell due to the lack of a base iPhone 18 model, though iPhone 18 Pro launched more strongly than its predecessor, especially in China and India.

Meta launches Muse Wonder AI models, claims frontier-level performance

Meta unveiled two new AI models, Muse Wonder and Muse Wonder Full, which it says match top-tier AI performance on coding, legal, financial and health tasks. The models add features called 'active memory' and 'active focus' meant to help AI agents retain context and persist through difficult tasks, and will roll out to Meta's Muse agent for paid subscribers. Muse Wonder is priced at $4 per million input tokens and $20 per million output tokens via API, while Muse Wonder Full requires a $50 or $200 monthly Muse Code subscription.

Meta unveils Muse Wonder and Muse Wonder Full AI models

Meta has released two new AI models, Muse Wonder and Muse Wonder Full, trained at its gigawatt-scale Prometheus data center. The company says the models build on earlier work shown with Muse Spark and post state-of-the-art results on benchmarks covering agents, memory, and coding, including domain-specific tests in law, finance and health such as Harvey's Legal Agent Benchmark.

Samsung foldable demand offsets memory shortage impact on shipments

Samsung is facing a broader memory chip shortage affecting device shipments industry-wide, but strong demand for its Galaxy Z Fold 8 foldable phone has helped the company maintain its position as a top smartphone shipper. The strength of this product lineup is credited with cushioning the company against the wider supply constraints.

Blog post details arena allocator technique for memory management in C

A developer-focused article walks through arena allocators as an alternative to standard malloc/free memory management in C. It explains how typical manual allocation leads to problems like forgotten frees, memory leaks, and slow performance from many small allocations, before introducing arena allocators as a solution and detailing their implementation.