🤖 AI Summary
Intel unveiled Panther Lake, its next-generation mobile “Core Ultra” platform due at CES 2026. The chip family builds on Lunar Lake’s efficiency while reintroducing beefier performance cores, a fifth‑generation NPU (claimed 50 TOPS), and an AI-capable Image Processing Unit (IPU 7.5). Significantly, Panther Lake’s Xe3 GPU tile adds aggressive multi-frame graphics generation (Intel suggests 3–4x frame-rate boosts in demos), positioning Intel to compete with AMD’s Ryzen AI Max and to power a new class of AI‑aware laptops and handheld “AI workstations.” Intel also touted process gains from its 18A node (compute tiles on 18A), claiming 10% higher single‑thread performance over Lunar Lake, >50% better multithreaded throughput, and overall system power reduced by ~10%.
Technically Panther Lake is a disaggregated, tiled SoC using Foveros 2.5D and a 2nd‑gen scalable I/O fabric. Three SKUs are shown: an 8‑core (4 P, 4 LP E) with 4 Xe3 cores; a 16‑core (4 P, 8 E, 4 LP E) with 4 Xe3 cores; and a 16‑core variant with 12 Xe3 cores and higher memory bandwidth (up to 9600 MT/s LPDDR5x). PCIe lane counts differ (12 vs 20), hinting which SKUs can attach discrete GPUs. Panther Lake also exposes Thread Director scheduling across P/E/LP E cores and an Intelligent Experience Optimizer for AI-driven power/performance tuning. Mixed manufacturing (some tiles at TSMC, others on Intel processes) and the separable GPU tile underline modular upgrade paths — and strategic tradeoffs — for future AI/graphics differentiation.
Loading comments...
login to comment
loading comments...
no comments yet