Qualcomm’s ARM-based Centriq 2400 has higher all-core throughput and better power efficiency than Intel’s Skylake Xeon SP


The Skylake has 48 hardware threads vs. Centriq’s 46.  In terms of CPU cores, Skylake has 24 (the test was conducted on a dual-socket system) vs. Centriq’s 46 (single-socket).  So one may justify Centriq’s higher throughput by its higher core count.  But when power efficiency is taken into consideration, that justification is no longer sufficient:

On the NGINX web server, …the [Centriq] chip did 214 requests per second per watt, compared to 99 for the Skylake chip….

Of course, higher count of less powerful cores has always been a strategy of many of Intel’s competitors (witness the more recent SPARC processors), but they are suitable mostly for embarrassingly parallelizable software.  Unfortunately, many mission-critical software systems are not like that.  For example, banking, ERP, and CRM systems are ultimately bottlenecked by their need to maintain accurate state information (in databases).


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s