Intel is reportedly launching its first 9th generation Core Whiskey Lake processors as early as mid August, 2018.

The Whiskey Lake processors will be the third refinement of the Skylake architecture following Kaby Lake and Coffee Lake. Whiskey Lake is expected to be a quad-core CPU with GT2 integrated graphics. The flagship Whiskey Lake CPU is expected to have 8 cores and 16 threads.

We have noticed a few Z370 motherboard makers indicating that the Whisky Lake processors will work on them in addition to the new Z390 motherboards. We suspect a BIOS update with updated device-id tables will be distributed close to launch day.

The latest reports suggest it may not be until next year in the fall before Intel can get the second refinement Ice Lake 10nm node yields up enough to be practical.

Intel only released one Core i3-8121U processor with Cannon Lake. The second generation 10nm  Ice Lake is expected to be designed from the ground up to deal with the Spectre and Meltdown problems. Intel suggested it will be about 12-15 months before Ice Lake is available in volume.

Redesigning the core logic for a new node is more challenging but the needs for a secure processor are very high. Intel, AMD, Qualcomm and others are all working to make their products more secure.

AMD is expected to launch their 7nm hardware this fall which will pressure Intel until they can get their ice Lake processor to the market. AMD is also making sure their new designs are more secure.

7NM V 10NM

Marketing is responsible for everyone in the semiconductor sector. Couple of technical details are needed: the minimum distance that the metal layers can handle and the limits of immersion lithography.

The 7nm and 10nm processes are essentially the same. They have the same wall of EUV limitations. Limitations of the design tools are another problem, it is hard to make the masks tolerant of increased power from the stepper lamps.

EUV use large CO2 lasers with a tin plasma to generate the designed EUV light. These machines weigh in excess of 180 tonnes and they do not last very long before they lose brightness. By comparison, the old style lamps would last over a year.

Right now silicon is at the very limits of what can be done. To make smaller devices will require new materials. Germanium has the shortcoming of thermal runaway. Gallium arsenide and other type III-V materials are not as well developed.

As the image shows, the imperfections in the substrate are rather apparent. Silicon atoms are about 0.2nm across. The number of atoms making up the crystal show that the they are not perfect and that occasional voids are present. Those atomic voids are the reason why 10nm has been so difficult.

Word on making silicon crystals better has been a long standing area of study.  A breakthrough in crystal manufacture is needed.