The Intel multi-threaded performance metrics can be difficult to understand.
The Intel multi-threaded performance metrics can be difficult to understand.
It seems Intel has finally acknowledged Linus for their usual promotional content, and they’ve released a video after previously overlooking any mention of MTL. While reviewing the material, I noticed some interesting points. They highlighted an IPC drop in MTL, which made me think about the 155H model—the same chip used for the 1370P? Deciding was tricky since both now use different naming conventions. I considered comparing their power consumption; yes, I7 might outperform i5 at the same wattage, but throttling does affect this.
What surprised me was the maximum turbo output: the 155H reaches 115 watts, while the 1370P tops out at 64 watts. Both have six cores plus eight E-cores, and those two LPE cores seem largely ignored. Yet the 155H only hits 4.8 watts compared to the 5.2 for the 1370P. What could explain this extra power? It might be due to a more powerful integrated GPU and NPU. Also, checking the 165H specs shows a 5 GHz turbo, closer to the 1370P than the 155H, so why not benchmark with a laptop? If we trusted the numbers, it would have been useful to see a graph of clock speed versus power use.
I also questioned their shift away from the tile-based modular design they favored. The monolithic layout lets cores communicate faster, reducing delays. In contrast, AMD has struggled with communication issues between its components. This likely hurt AMD’s performance but allowed them to build very fast cores that still compete. Intel’s approach avoids routing connections through the PCB, using a silicon interposer to cut latency—though some residual delay remains. If you watch their videos, it might help clarify these differences.
Noticing this trend, I wonder if they’re overlooking the downsides of modularity. Most people appreciate Intel’s monolithic chips for their speed and efficiency. The standby power savings matter too; keeping the laptop on standby wastes energy. Still, the real focus should be on actual performance metrics rather than just design philosophy.
It seems Intel hasn’t fully embraced these new technologies yet, mainly because they didn’t meet desktop performance goals. @Gat Pelsinger is just one of many voices now. But the new GPU and NPU features are definitely worth noting.