35 results for “Fusion”
Broadcom released VMware Workstation and Fusion 26H1u1 to remediate two VM-escape flaws, led by a critical CVSS 9.3 integer overflow in VMXNET3. Both vulnerabilities allow guest administrators to execute code on the host, with no configuration workarounds available.
NVIDIA unveiled NVLink Fusion, letting hyperscalers plug custom XPUs into its 72‑XPU NVLink domain and MGX rack architecture. The program cuts non‑differentiated engineering so silicon teams can ship custom accelerators faster while reusing proven power, cooling, and cluster software.
NVIDIA launched NVLink Fusion with NVHBM on August 26, 2026, giving hyperscalers a standardized path to deploy custom XPUs alongside NVIDIA GPUs. The co-designed memory and interconnect stack promises 30% higher bandwidth, 25% more compute area, and 15% lower HBM power versus standard HBM4e.
Tencent has released AngelSpec, a unified training framework designed to tackle the persistent inference efficiency challenges that accompany large. Inference cost remains one of the most significant barriers to widespread LLM adoption.
On 29 July 2026, Broadcom published security advisory VMSA-2026-0006 addressing five vulnerabilities across VMware ESXi, vCenter Server, Workstation, and Fusion. vCenter Server is the central management plane for vSphere environments.
Hugging Face released @huggingface/kernels with 207 versioned WebGPU kernels and the Fleet benchmarking tool for browser-based AI. The kernels beat ORT WebGPU by 2.57× geometric mean on Apple M4, creating a shared performance foundation for local inference.
MiniMax released Music 3.0, an open-weight model that generates complete songs up to five minutes with structured vocals and evolving arrangements. The model combines an 8B Global LLM with a continuous Flow Matching synthesis pipeline for 32 kHz stereo output.
On 2 September 2026 the PyTorch Foundation released 2.14: nccl2 and NVGEMM land in-tree, fault tolerance becomes a first-class c10d feature, and Apple Silicon gains native linear algebra plus Metal kernels that close an 8.5× single-token decode gap versus the prior MPS path.
The Zvec team open-sourced zg (zvec-grep), a local-first CLI and MCP layer that unifies ripgrep, BM25, and vector search for humans and coding agents. Qwen Developers amplified the launch on 2 September with benchmark claims on agent token use.
NVIDIA validated Alibaba's Qwen3.8-Flash-Next 176B MoE model on its GB300 NVL72 rack, achieving over 16K tokens per second per GPU. The hybrid GDN/QSA architecture keeps memory and compute bounded for million-token agentic coding workloads.
NVIDIA declared Scale-In the fifth pillar of AI networking infrastructure, powered by BlueField-4 DPUs, DOCA software, and Spectrum-X Ethernet. The architecture offloads security, storage, and telemetry from host CPUs at up to 800 Gb/s to keep pace with agentic AI factories.
NVIDIA detailed how Spectrum-X Ethernet uses hardware-accelerated adaptive routing, congestion control, and plane load balancing to solve traditional Ethernet's failures at giga-scale AI training. The platform delivers 1.6x performance over off-the-shelf Ethernet, near-perfect multi-tenant isolation, and 2.68 ms failover versus 1.08 seconds for standard Ethernet.