The Core Ultra 7 366H has surfaced on Geekbench, confirming its 16-core configuration composed of 4P + 8E + 4 LP-E, paired with 18MB L3 cache, a 2.0 GHz base clock, and a 4.8 GHz boost clock. These specs align with earlier engineering samples, indicating this silicon is close to mass-production tuning. Achieving 4.8 GHz boosts within a 25W default power envelope also signals mid-to-high bin quality, reflecting strong voltage–leakage characteristics suitable for the H-series thermal profile.
🔍 CPU Architecture & Binning Insights #
The 366H’s ability to sustain a short burst to 4.8 GHz under 25W reveals ideal silicon characteristics. Lower-bin parts typically require more voltage for the same frequency, failing thermal validation in the constrained H-class envelope.
- 4P cores handle peak single-thread responsiveness
- 8E cores sustain background and multi-threaded workloads
- 4 LP-E cores manage idle and ultra-low-power tasks
This hybrid configuration is consistently seen across most Panther Lake-H SKUs, with variations mainly in GPU core count and frequency. The high structural reuse implies the architecture is fully finalized, and segmentation is performed post-fab through binning and defect mapping.
🎮 Xe3 iGPU Performance: 4 Cores Reaching GTX 1050 Ti Levels #
The integrated GPU includes 4 Xe3 cores—the lowest configuration in the Panther Lake lineup.
Benchmark results include:
- Vulkan score: 22,813
- Mobile GTX 1050 Ti score: 21,937
- Radeon 840M: ~26% slower
- 860M (8 Xe3 cores): ~37,552, significantly higher
Reaching 1050 Ti-class performance with just 4 Xe3 cores requires:
- aggressive frequency boosts
- tight cache-path tuning
- high-quality silicon to support elevated operating points
This matches the pattern seen in low-EU-count iGPUs, where reduced parallelism is offset by frequency optimization—possible only with strong die characteristics.
📉 Limitations of the 4-Core Xe3 Setup #
The 4-core Xe3 configuration is not designed for heavy graphics workloads. It serves as a baseline GPU for models without discrete graphics:
- strong short-term burst performance
- throttling under extended load due to 25W platform constraints
- limited texture/raster resources relative to 8-core variants
In sustained scenarios, performance will dip to remain within the platform’s power and thermal boundaries.
🧩 GPU Segmentation Strategy in Panther Lake #
Higher-end Panther Lake GPUs (10-core and 12-core Xe3 variants) have appeared in 3DMark Time Spy with results approaching the RTX 3050. These SKUs use fuller GPU Tiles and require higher turbo power budgets.
Intel’s segmentation reflects typical yield distribution:
- Low-core iGPUs come from partial-defect or edge-cut GPU Tiles
- High-core SKUs use clean, high-quality die sections
- Frequency and voltage tables are separately tuned for each segment
This efficient reuse increases yield while creating clear product tiers.
📊 What the 366H Benchmark Means for OEMs #
The Geekbench data confirms that even the smallest Xe3 configuration can deliver GTX 1050 Ti-class performance in short bursts—impressive within a 25W mobile envelope. For OEMs, this:
- sets a minimum graphics capability for ultraportables without dGPUs
- supports light 3D gaming and modern UI acceleration
- avoids reliance on discrete GPU components in cost-optimized builds
Panther Lake-H’s broad reuse of 4P+8E+4LP-E, with GPU-driven segmentation, gives OEMs a predictable performance baseline across multiple chassis designs.
Overall, the Ultra 7 366H leak validates Intel’s binning strategy and confirms the viability of 1050 Ti-level iGPU performance in next-generation thin-and-light devices.