Revue de presse générée automatiquement avec
Tech Enthusiast
Une personne férue de technologie, passionnée par les dernières innovations et avancées, qui recherche des informations approfondies sur les tendances et les percées du secteur, et qui s'intéresse également aux découvertes scientifiques.
Vous souhaitez recevoir chaque jour la revue de presse de ce profil ?
Your brief - 24h (10 articles)
Samedi 13 décembre 2025 à 19:50
AI Model Surge, DRAM Crunch, and Kernel Milestones...
AI Breakthroughs & Enterprise Models
OpenAI rolls out GPT‑5.2 for professional workloads
OpenAI announced GPT‑5.2, touting a 70.9 % score on the GDPval benchmark for knowledge‑work tasks, more than double the previous version. The upgrade adds longer context windows, advanced tool‑calling, and vision capabilities, and is being released first to paid ChatGPT plans and the OpenAI API.
SD Times
Google upgrades Gemini with audio‑first models and live speech translation
Google unveiled Gemini 2.5 Flash Native Audio, a model that handles complex audio workflows and supports real‑time speech translation in over 70 languages via the Translate app. The company also opened a beta Interactions API to let developers stitch together Gemini agents, signalling a push toward more interactive AI services.
SD Times
Intel pivots to ASIC‑centric AI strategy
Intel’s new AI roadmap favors a Broadcom‑style ASIC model, emphasizing custom silicon and edge AI over the prevailing large‑scale training approach. The plan includes offering foundry and packaging services to partners, aiming to close the gap with rivals NVIDIA and AMD in the AI accelerator market.
Wccftech
NVIDIA’s Hopper‑based H200 GPUs see soaring demand in China
Following the U.S. easing of restrictions on legacy Hopper GPUs, NVIDIA’s H200 AI GPUs are experiencing “massive” demand in the Chinese market, prompting the company to expand capacity while assuring U.S. customers of uninterrupted supply. Analysts view the surge as a short‑term boost before newer Blackwell chips dominate.
Wccftech
Hardware Market Pressures
Global DRAM shortage ripples through PCs, servers, and smartphones
SK Hynix warned that commodity DRAM supply will lag demand through 2028, a trend echoed by Dell’s upcoming price hikes on RAM, storage, and GPUs as memory costs remain “out of our control.” Wccftech adds that the shortage may force smartphone makers to re‑introduce 4 GB RAM devices and slow the rollout of 16 GB flagship models, underscoring a systemic bottleneck across the ecosystem.
Wccftech
Tom's Hardware
Wccftech
Open‑Source & Kernel Innovations
Oracle’s bpftune brings BPF‑based auto‑tuning to Linux systems
Oracle released an updated bpftune tool, enabling automatic performance tuning of Linux kernels via eBPF programs. The open‑source project, now available on GitHub and Oracle Linux, promises dynamic adaptation to workload changes without manual intervention.
Phoronix
Rust solidifies its place in the Linux kernel
The Linux kernel maintainers announced that the Rust experiment is concluded, with the language officially accepted for ongoing kernel development. Miguel Ojeda highlighted the success of early Rust modules and the commitment to keep Rust as a first‑class language alongside C.
Phoronix
NFS gains initial directory delegation support in Linux 6.19
Linux 6.19’s kernel merge introduced basic directory delegations for the Network File System (NFS) client, enhancing cache consistency and reducing round‑trip latency for distributed workloads. This feature marks a step toward richer NFS capabilities in future releases.
Phoronix
LoongArch32 support lands in Linux 6.19 and GCC 16
The LoongArch architecture, a Chinese RISC‑V‑inspired CPU line, received its first 32‑bit support in the Linux 6.19 kernel and GCC 16 toolchain, broadening the platform’s software ecosystem and paving the way for legacy application compatibility.
Phoronix
Aller aux sources
10 sources citées