Ill_take_Power
Junior Member
- Apr 29, 2014
- 7
- 0
- 0
It is an issue when the platform is deficient. x86 has to be mindful of it for thermal and packaging reasons. For that matter, so does Power. However, within the thermal envelope which Power servers run, they are demonstrating up to and sometimes over 2X the performance per core of Ivy Bridge EP and EX on various workloads. That is just the foot race. Factor in the efficiency provided by the Power Hypervisor where it will service the workloads interlacing available resources what you have is true efficiency that drives utilization. As such, you can scale up vs out. Power8 is aptly now called Scale out servers because it offers customers both the ability to scale out with price parity based solutions while still scaling up. Perf / watt is not my a Power server problem so don't try to lay that on it. Fine if you want to measure the x86 servers with it. Use it to compare which one is more efficient over the other. At the end of every month, whether one server is using a 90W or a 135W chip would surprise me if it made up for the cost of the software licenses like Oracle that ran on it vs running on Power.
