One of Intel’s most infamous architectures, if not the most infamous, has to be Cannon Lake. This first generation Intel CPU is held in such contempt that even Intel tries to forget it exists by redefining Ice Lake as the first 10nm arch and by deleting the “products formerly known as Cannon Lake” page from the Ark database. Officially, only the dual core i3 8121U was ever released, without integrated graphics and with clock speeds and power worse than 14nm dual cores. But, it seems Intel actually did create some 6 and 8 core engineering samples, and even a dual core with graphics enabled.
For those who are unaware, Cannon Lake was Intel’s first 10nm CPU based on the Palm Cove core, Intel’s first departure from the Skylake core which debuted in 2015. While many expected Cannon Lake to succeed Skylake sometime in 2016, the year came and went without a new 7th generation of CPUs. In January 2017, a new architecture called Kaby Lake was launched instead, which was an “optimization” of Skylake with increased clock speeds and slightly reduced power draw. Around this time, the scale of the disaster unfolding with 10nm was becoming apparent and Intel eventually launched Cannon Lake in mid 2018 on a few laptops and NUCs. Cannon Lake was eventually discontinued in 2019. Cannon Lake had failed due to 10nm’s poor performance, efficiency, and yields, which made only a dual core with no graphics viable for even this low amount of volume. It came and went without so much as a whisper.
However, Intel seems to have intended to launch other Cannon Lake CPUs thanks to some digging done by leaker Rogame on the 3DMark database. He was able to find Cannon Lake CPUs with 6 cores, 8 cores, and enabled integrated graphics. These results were all obtained from engineering samples, of course, because Intel never bothered qualifying CPUs that were basically never going to make sense producing. The existence of these CPUs is probably not surprising to everyone, but to my knowledge this is the first time we have ever gotten proof that these SKUs really were supposed to exist.
It is really interesting that Intel did get as far as producing working samples of these high core count CPUs, however; it kind of lives up to the hype people had about Cannon Lake prior to 2016. Imagine that everything had gone to plan: Intel could have started 2016 off with new 6 and 8 core SKUs for the mainstream on their brand new 10nm node. AMD would have only had their Zen CPUs ready in 2017, and they would have faced CPUs with core parity, likely higher clock speeds, and higher IPC. AMD clearly got very lucky with Intel’s failure to complete 10nm and launch it in a working state.
Rogame appears to be leaking even more of these results to his Twitter, so we highly recommend you check him out.