1

I want to report performance statistics about an algorithm that runs on a virtual server instance with Ubuntu on it. I measured runtime and maximum memory usage but I want to give a statement about the CPU speed as I feel the runtime is meaningless without. However, lshw just gives me the following information about the CPU:

description: CPU
      product: QEMU Virtual CPU version 1.0
      vendor: Hynix Semiconductor (Hyundai Electronics)
      physical id: 401
      bus info: cpu@0
      slot: CPU 1
      size: 2GHz
      capacity: 2GHz
      width: 64 bits
      capabilities: fpu fpu_exception wp de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 syscall nx x86-64 nopl pni cx16 popcnt hypervisor lahf_lm svm abm sse4a

How shall I summarize my environment? "Tested on a 2 GHz virtual server instance."? I feel like clock speed is not sufficient in modern environments as it is stagnating for years.

Falcon Momot
  • 24,975
  • 13
  • 61
  • 92
Konrad Höffner
  • 195
  • 1
  • 9
  • Measuring performance on a vm will vary depending on what other vms are doing at the time. If you need repeatable benchmarks I would run it on a physical machine, and include the exact processor model. – Grant May 08 '14 at 15:15
  • I would compare bottlenecks of your algorithm on your vm vs your other benchmark system and state what systems would've caused the bottleneck. For example, on the VM it ran slower because there was more iowait time due to the VM only having access to a single hard drive SAN over ethernet whereas my other test environment had direct access to a SSD. – Jason Zhu May 08 '14 at 23:09

0 Answers0