Intel’s 3rd Gen Scalable Ice Lake processors which represent the first new server architecture from the company in over four years are expected to see widespread adoption in the second half of the year. According to a report from TrendForce, increased demand in the post-pandemic era is expected to see rapid shipments of the Ice Lake platform, with the overall market share expected to cross 30% by the end of the year.
In comparison, AMD’s overall server share is slated to reach 15% sometime in the first half of 2022. Meanwhile, Intel’s next-gen Sapphire Rapids server processors are on track for volume production in the second quarter of 2022. This is a delay from the previously expected Q4 2021 ramp-up. Regardless, Intel is set to begin shipments to select clients in the first quarter of 2022. In hindsight, the Sapphire Rapids launch is looking much like that of Ice Lake.
Sapphire Rapids is going to be a big step up from the existing server solutions with a slew of new technologies and upgrades, most notably on-die HBM memory, DDR5, PCIe 5.0, and CXL 1.1. Meanwhile, AMD’s competing Genoa lineup is expected to enter mass production sometime in Q2-Q3 2022. It’ll feature similar technologies, albeit with denser compute capabilities. According to recently leaked documents, it looks like AMD is jumping onto the AVX512 bandwagon with its 4th Gen Epyc processors, one of the primary (only) advantages Intel still has over the underdog.
Like Sapphire Rapids, Genoa is expected to support DDR5 memory, PCIe 5.0, and instead of on-die HBM, we’re likely to see 3D stacked V-Cache. Trento which is a custom variant of Milan is expected to feature custom interconnect technology (much like CXL) in the form of Infinity Fabric 3.0, connecting each Epyc CPU to four Instinct GPUs.
AMD’s server share recently surpasses 10%m thanks to the influx of the newly launched Epyc Milan processors. By 2022, the company is expected to reach a total share of 2022%, with further advances expected with the launch of Genoa in 2023.