隨著代理式AI（Agentic AI）快速發展，輝達認為資料中心CPU的設計邏輯正在出現反轉。過去雲端時代追求的是「塞進更多核心」，但AI代理需要CPU與GPU之間高速往返，因此單執行緒效能、記憶體頻寬與低延遲重新變得重要。
輝達 Vera CPU 團隊接受 Wccftech 獨家專訪，坦言雲端時代為了堆核心數而犧牲單執行緒效能的 chiplet 路線正在被 AI 代理反轉：當 GPU 是最昂貴的算力資產，CPU 必須以最短延遲完成順序任務，輝達索性改回單一晶片設計，專為 ...
NVIDIA Vera promet 50 % de performances en plus que Grace grâce à une puce monolithique pensée pour les agents IA et les centres de données.
On, NVIDIA's Vera team told Wccftech that cloud-era chiplet designs traded single-threaded performance for higher core counts ...
NVIDIA's Vera team argues chiplet layouts leave little room for focusing on per-core speed, a trade-off that Vera's design is ...
How-To Geek on MSN
Free-threaded Python is no longer an experiment. Here's what it unlocked in my workload
It's been in the works for a while.
Hello, this is k-chan.There are times when you need to download and collect thousands of data points at once from a target ...
Explore a atualização Python 3.15 com JIT aprimorado, carregamento lazy e UTF-8 padrão para mais desempenho e compatibilidade.
정적 필드는 다중 스레드 빌드로 마이그레이션할 때 신중하게 처리해야 합니다. 다중 프로세스 모델에서도 단일 프로세스가 여러 프로젝트를 빌드할 수 있으므로 정적 상태는 공유되지만 ...
Tokenization is the one part of the language modeling stack that almost nobody profiles. Gigatoken, released by Marcel Rød (a PhD student from Stanford) under an MIT license, argues that this was a ...
A new Python-based malware called SolyxImmortal has been found quietly stealing browser passwords, cookies, sensitive files, and keystrokes from infected Windows systems. The malware uses well-known ...
Nvidia has shared a list of global AI companies that have adopted the VERA systems, confirming that Anthropic, Open AI and SpaceXAI are the early adopters of this CPU system. Source: NVIDIA As part of ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results