The battle between Intel and AMD for the consumer chips market has been going on for quite some time, and AMD also came up with a new lineup for the server chips. As of today, Google offers the possibility to use a new range of AMD chips for virtual machines in the Google cloud.
The 7nm chips come from the Epycome series and will go through life as the N2D series. These are Google‘s fastest to date, and, in their own words, more efficient than instances using Intel chips.
N2D series instances will run on special chips from the Epyc-Rome branch, with a standard clockspeed of 2.25 Ghz, including boost (all cores) to 2.7 Ghz. A single-core boost to 3.3 would also be possible according to Google. The amount of actually available cores depends on how much is made available for a specific instance: this in turn depends on the amount of virtual CPUs a user wants to have at his disposal, but can increase significantly with N2D. Google offers the possibility to go up to 224 vCPUs in the new line-up.
Intel vs. AMD
Instances with Intel chips are already overtaken on that aspect by AMD, since those instances (offered by Google) did not go beyond 96 vCPUs. Also, the maximum amount of memory to be allocated was considerably less than with AMD.
Google expects that the performance with just under forty percent would be better than the current N Family instances (running on Intel) and will be about thirteen percent more efficient in costs.
The use of N2D instances is now in the beta phase.