A few weeks ago I held a Raptor Lake wafer in my own two hands, however, it looked absolutely nothing like the Raptor Lake wafer just spotted at Intel Innovation by Paul Alcorn for our sister site Tom
Audio player loading…
A few weeks ago I held a Raptor Lake wafer in my own two hands, however, it looked absolutely nothing like the Raptor Lake wafer just spotted at Intel Innovation by Paul Alcorn for our sister site Tom’s Hardware (opens in new tab). So what gives?
This new wafer shows off a fundamentally different chip design to the ‘standard’ Raptor Lake-S wafers we’ve seen so far. Rather than two rows of P-cores butting up against four clusters of E-cores—for a maximum of 24 cores in total, a la the Core i9 13900K—what we’re seeing on this unannounced wafer is an interconnected grid of what appears to be solely P-cores. 34 of them.
This sort of die layout is more expected of Intel’s server-grade processors, starting with those based on the Skylake architecture. It works by increasing interconnectivity by having more cores connected directly to one another, reducing the bottlenecks that could happen with high core count chips on a ring bus architecture.
Intel had previously brought these sorts of remixed server chips to the enthusiast and workstation market under the X-series branding, though that all stopped when desktop core counts skyrocketed.
