- 资讯公开站Digitimes
TSMC mulls multi-fab Texas campus beyond Arizona, with Singapore also in talks
TSMC is reportedly weighing a new manufacturing campus in Texas that could add tens of billions of dollars to its US expansion, potentially creating a second major American production hub alongside Arizona as AI chip dem…
- 资讯公开站Digitimes
DIGITIMES Insight: Intel, TSMC, Samsung converge on backside power — packaging becomes the next divide
Data center AI accelerator package power has climbed from about 500W in the A100 generation to roughly 3,500–4,000W for Rubin Ultra-class products expected in 2027, pushing supply current into the thousands of ampe…
- 资讯公开站Digitimes
Commentary: Broadcom's US$42 billion Anthropic loan exposes AI's emerging chip-financing loop
When the underlying economics of infrastructure commitments meet the financing muscle of the companies supplying foundational compute, the traditional boundary between supplier and customer begins to blur.
- 资讯公开站Digitimes
Onsemi revises Synaptics takeover deal after rival bid
Onsemi has amended its agreement to buy Synaptics in an all-cash deal, a move that could reshape semiconductor competition and affect global technology supply chains. The revised cash transaction offers Synaptics shareho…
- 论文公开站arXiv
Cost-augmented Schrödinger bridges on graphs are exactly solvable: a Feynman-Kac tilt replaces learned control
The generalized Schrödinger bridge on a graph moves mass between two distributions while charging a cost for the states visited. It has been approached by learning the rates of a controlled continuous-time Markov chain, …
- 论文公开站arXiv
FERPO: Forward Entropy-Regularized Policy Optimization
Several state-of-the-art methods for online reinforcement learning in continuous control improve policies using action gradients of a learned critic. However, critics are typically trained to predict returns, and accurat…
- 论文公开站arXiv
TACO: Ternary Absolute-max Column-wise One-sparse Optimizer for LLM Fine-Tuning
Full-parameter fine-tuning of large language models (LLMs) incurs substantial optimizer state memory overhead, limiting the model sizes that fit on modern GPUs. Existing approaches either compress optimizer state, abando…
- 论文公开站arXiv
ScholarCatalyst: A Benchmark for Retrieving Papers That Inspire New Research
What makes great scientists great? Even as AI systems start to make progress on open problems, scientists remain far ahead of them at sensing which prior idea, buried in an ever-growing archive of research, a new problem…
- 论文公开站arXiv
Embedding Prediction Helps Image Generation
In diffusion transformers, a class label or a text prompt is embedded once, and the same condition is reused at every denoising step. We ask whether predicted embeddings can serve as this condition instead. Next-Embeddin…
- 资讯公开站CNX Software
ASRock Industrial NUC 300 series – Intel Core 3 304/Core 5 320 mini PCs and motherboards
ASRock Industrial has introduced the NUC Core 300 BOX series of mini PCs and the NUC Core 300 motherboards built around Intel Core Series 3 Wildcat Lake-U processors. The lineup includes the NUC BOX-320, with the Intel C…
- 资讯公开站Digitimes
Machvision expands early to chase NT$10B revenue
Machvision is expanding capacity earlier than planned to meet rising demand from semiconductors, substrates, and optical modules, while tight computer numerical control (CNC) and machining supply chains have emerged as i…
- 资讯公开站EE Times
Edge Computing and Security for Access Control Applications: Getting the Best of the Two Worlds
Join us where attendees will see how on-device processing improves resilience, reduces exposure of sensitive data, counters spoofing, and supports practical deployment across buildings, smart locks, vehicles and applianc…
- 资讯公开站rss_arxiv_cs_ai
ArgGYM: A Procedural, Engine-Verified Benchmark for Structured Defeasible Reasoning
arXiv:2609.38409v1 Announce Type: new Abstract: Recent progress in large language model reasoning has been driven by benchmarks and reinforcement learning environments with automatically verifiable rewards, particularly…
- 资讯公开站rss_arxiv_cs_ai
SimTrace: Grounded Multimodal User Trajectories Generation for Online User Modeling
arXiv:2609.38397v1 Announce Type: new Abstract: Virtual clients offer a cost-effective approach to support applications such as A/B testing, recommender system development, and interface evaluation. However, building th…
- 资讯公开站Semiconductor Engineering
The Hidden Challenges of Edge AI Design
- 资讯公开站Semiconductor Engineering
AI-Defined Vehicles Push Compute, Memory, And Validation Limits
- 资讯公开站Semiconductor Engineering
LLM Performance And Acceleration: Part 1
- 资讯公开站Digitimes
China's 'BEST' fusion reactor enters the engineering endgame
China's next-generation fusion programme has entered a new engineering phase with the formal opening of the campus for the Burning Plasma Experimental Superconducting Tokamak (BEST) and the start of critical main-device …
- 资讯公开站Digitimes
China OSAT share in cloud AI accelerators to hit 13% in 2026 as local ASIC volume ramps up
As China accelerates its push for homegrown AI computing chips, custom ASIC accelerators from domestic tech giants—including Huawei, Cambricon, and Alibaba—are entering volume production. These chips are bein…
- 资讯公开站Digitimes
Neweb Information targets late-October listing on Taipei Exchange
Neweb Information, a Taiwan-based IT systems integrator, is targeting a late-October listing on the Taipei Exchange. Long focused on enterprise core systems, the company has expanded in recent years into AI GPU computing…
- 论文公开站arXiv
Scaling Laws for Looped Mixture of Experts
Looped transformers and Mixture-of-Experts (MoE) offer complementary routes to efficient scaling: recurrence increases computational depth at fixed parameters, while MoE sparsity expands total capacity at fixed active co…
- 论文公开站arXiv
Cogentic：面向自动证明发现的多智能体编排
Cogentic: Multi-Agent Orchestration for Automated Proof Discovery
提出 Cogentic，一个用于开放研究问题自动证明发现的多智能体框架。前沿语言模型虽能单次生成优质数学想法，但面对需多路径探索的开放问题，单次生成往往不足。
- 论文公开站arXiv
DP-SGD 下权重绑定对纯解码器 LLM 是否仍有益？
Is Weight Tying Still Beneficial for Decoder-Only LLMs in Private Settings Under DP-SGD?
DP-SGD 是大模型隐私保护微调的主流方法。许多纯解码器 LLM 采用输入输出嵌入权重绑定，本文在差分隐私设定下重新审视这一设计是否仍然有利。
- 资讯公开站CNX Software
面向 CM5 的 30 美元树莓派智能显示模块，瞄准数字标牌与自助终端
$30 Raspberry Pi Smart Display Module for CM5 targets digital signage displays and kiosks
树莓派智能显示模块是兼容英特尔智能显示模块（SDM）规范的 CM5 载板，具体为 SDM-L 规格。英特尔 SDM 标准已存在多年。
- 资讯公开站Digitimes
谷歌将首次把TPU送入轨道，测试太空AI计算
Google to test AI computing in space with first orbital TPU launch
谷歌准备首次将其自研AI加速器TPU送入近地轨道，探索AI计算能否从地面数据中心扩展到太空。CEO皮查伊近日宣布了这一计划。
