Archive
AI 晶片、GPU、資料中心、邊緣裝置與硬體安全。
硬體
The MobiCom 2026 study divides model weights and the KV cache into shared buffers, allowing the GPU to begin prefill while the CPU restores evicted data. The work targets recovery latency after multitasking on smartphones, rather than improving the model’s generation speed itself.
醫療 AI 與物理導向學習
New research combines GPU pencil-beam calculations, two frozen neural priors, and a 1.36-million-parameter 3D U-Net in a shared CT/MRI workflow. On internal data, the 1%/1 mm gamma pass rate reached as high as 99.12%, but the results do not yet constitute cross-institutional or clinical-use validation.
推論系統
`@huggingface/kernels` can dynamically load GPU operators from the Hub, complete with interfaces, tests, and WGSL implementations. In operator-level tests on an Apple M4, Hugging Face reported a 2.57× geometric-mean speedup, but the results exclude model loading, compilation, and data transfer costs.
推論系統
Unsloth now enables multi-token prediction by default for Qwen3.8-Flash-Next and GLM-5.3-Flash, while improving its MLX, multi-GPU, and ROCm paths. The official claim of up to 2× higher speed is not accompanied by a complete benchmark matrix, while Qwen’s technical documentation reports gains of 1.3× to 1.7× at low concurrency.
AI 安全
CrowdStrike and NVIDIA have unveiled SafeMind, which uses an offensive agent to generate telemetry and Nemotron defensive agents to write, validate, and replay detection rules. Closed testing indicates that specialized models outperform general-purpose frontier models, but model weights, datasets, and the full evaluation protocol have not yet been released.
最新模型
Atlas places text, images, camera poses, and depth data in a single autoregressive diffusion architecture, enabling controlled camera motion and outputting point clouds or 3D Gaussian splats. Vendor evaluations show that it outperforms specialized models on sparse-view reconstruction, but with no paper, weights, or physics-simulation validation available, it remains limited to early access.
推論系統
LMDeploy 0.17.0 adds the DeepEPv2 expert-parallel backend to TurboMind and connects its KV connector to Mooncake Store. The release also supports Kimi K2.6, non-power-of-two page sizes, and structured responses, but provides no end-to-end performance data across hardware configurations.
代理工具
OpenAI has added output token limits for individual MCP tools and now allows app-server shell commands to run for more than an hour. The release also fixes an issue where user instructions and authorizations could be lost after conversation history was compacted, while the built-in planning tool is now disabled by default.
AI 安全
F5 has added anomaly detection and agentic threat intelligence to Distributed Cloud WAF, while connecting vulnerability scan findings to runtime blocking workflows. The new capabilities could shorten the exposure window, but detection rates, false-positive rates, and latency figures are still based solely on vendor testing.
AI 基礎設施
HUMAIN is now offering production-grade GPU services in Saudi Arabia built with AMD MI355X, EPYC, and Cisco Silicon One networking. This moves the partnership beyond the “planned capacity” stage, but 250 MW and 1 GW remain targets for deployment beginning in 2027 and by 2030, respectively.
AI 基礎設施
EuroHPC and Bull have signed a €387.8 million contract to deploy LUMI-AI in Finland, built around AMD MI430X GPUs and sixth-generation EPYC CPUs. The system is expected to come online in 2027, but the projected tenfold increase in AI capacity remains a planning estimate, while node counts, interconnects, and storage configurations have yet to be fully disclosed.
推論系統
NVIDIA Triton Inference Server 2.72.0 fixes an issue that could cause the scheduler to stop dispatching work while requests were still pending and restores the completion-order semantics of `preserve_ordering`. The release also updates its CUDA, TensorRT, and gRPC dependencies, but no TensorRT-LLM image is available, while security warnings remain for vLLM and Ray deployments.