Last week I began delivering benchmarks of the low-power yet massively scalable Calxeda EnergyCore ECX-1000 ARM Server and followed the initial tests with some ARM compiler benchmarks and other benchmarks from this 5-Watt Linux Server. In this article is what many Phoronix readers have been waiting for: comparing Calxeda's quad-core Cortex-A9 ARMv7 performance against a dual-core Texas Instruments OMAP4460 PandaBoard ES and then an Intel Atom processor.
When all of the systems were running Ubuntu 12.10 with the Linux 3.5 kernel, the 1.1GHz and 1.4GHz Calxeda ECX-1000 EnergyCore nodes were compared to a PandaBoard ES and an Intel Atom NetTop. The PandaBoard ES is the development board for the Texas Instruments OMAP4460 SoC, which has a dual-core ARM Cortex-A9 processor with each core operating at 1.20GHz. The Intel Atom NetTop was the ALUSA Atom Desktop with an Intel Atom D525. The Intel Atom D525 has two physical cores plus Hyper Threading, thus four logical cores, and a 1.8GHz operating frequency. The TDP on the Atom D525 is 13 Watts.
With the comparison hardware being used for these published tests today all being 2~4 cores, benchmarks for this article are only from single Calxeda cards -- both the quad-core 1.1GHz and 1.4GHz variants.
The benchmarks results have shown that the average power consumption of just the Calxeda quad-core ARM SoC is 4.0 Watts for the 1.1GHz version and 6.4 Watts for the 1.4GHz version. However, for this article there are unfortunately not any performance-per-Watt results to share.
Calxeda exposes the SoC power consumption via IPMI while neither the PandaBoard ES or Intel Atom system expose their individual SoC/CPU power consumption via IPMI or any other monitor-able interface. The AC power consumption of the PandaBoard ES or Intel Atom could be monitored via the WattsUp USB power meter that integrates nicely with the Phoronix Test Suite too, for getting the overall system power consumption. However, that isn't comparable to just the Calxeda SoC power consumption. The Boston Viridis server used for this testing could also be monitored via the WattsUp power meter on the AC end, but then that would include all of the EnergyCore cards and just not a single card. There's also different disk drives, memory, etc between systems. So there isn't an adequate way to properly compare the power consumption in this article, but for what it's worth, the Intel Atom D525 has a 13-Watt TDP.
All x86 and ARM Linux benchmarking was carried out in a fully automated and reproducible way using the Phoronix Test Suite software. You can also compare your system's performance to the Calxeda quad-core results, PandaBoard ES, and Atom D525 in an automated and side-by-side manner simply by executing phoronix-test-suite benchmark 1210247-RA-CALXEDAAT08.