Intel has lifted the veil on its upcoming Meteor Lake platform, the largest architectural shift Intel has had in over 40 years, and their most power efficient client chip ever.
has finally lifted the lid on its upcoming Meteor Lake architecture after teasing it for some months now. The first to be built on their own Intel 4 process node, Intel claims Meteor Lake will be their most power efficient client processor ever, in no small part thanks to a major shift in how its designed.
Starting off with the compute tile, as its name suggests it’s home to the actual CPU of it all, containing Intel’s latest generation P-core and E-core, each of which are themselves based on newer microarchitectures dubbed Redwood Cove and Crestmont respectively. There’s little in terms ofhow much better it is than previous generations, but Intel does at least say that the new P-cores have some improved performance efficiency while the new E-cores have IPC gains over previous E-cores.
This SOC tile meanwhile is where you’ll find an integrated Neural Processing Unit too that will take care of all the AI-related tasks such as Windows Studio Effects or Stable Diffusion more efficiently than if it was carried out by the CPU, with compatibility for OpenVINO. The SOC tile is also home to the integrated memory controller and DDR, support for up WiFi 7 and Bluetooth, and takes care of display output like HDMI 2.1 and DP 2.
On top of the up to eight Xe-LPG graphics cores, the will be eight ray tracing units too, a first for Intel’s integrated graphics. Intel says that the hardware ray tracing won’t just be beneficial in gaming, but for creation and research purposes too. There’s also support for Intel XeSS, its own take on upsampling similar to NVIDIA DLSS or AMD FSR.