Hello,
I am running 2 NAS parallel benchmarks (more specifically EP and LU) on two 2-socket servers (Intel Xeon processor E5-2630 v4 and also on low-power Intel Xeon processor E5-2630L v4) [1] and measuring the power consumption using sysfs interface. I am noticing that the average power consumption for these benchmarks on low-power Intel Xeons is more than the standard Xeon. Please see the plot below. I am unable to wrap my mind as to what could be the possible reasons for this? Any ideas? Is it related to Performance and Energy Bias Hint support in them?
[1] https://www.intel.com/content/dam/www/public/us/en/documents/product-briefs/xeon-e5-brief.pdf
Link Copied