How does speed of QEMU emulation compare to a real Raspberry Pi board?

Question

I've managed to following Jivings answer to set up QEMU on my Ubuntu machine with the default Debian image, though I have had to use arm1136-r2 instead of arm1176.

What is the best way to compare the speed the emulated processor is running at?

Doing a cat /proc/cpuinfo gives me BogoMIPS = 135.57. This was on a Intel Atom @1.6 GHz, no KVM.

How does this relate to an actual Raspberry Pi?

Answer 1

On my Pi, cat /proc/cpuinfo gives me:

pi@raspberrypi:~$ cat /proc/cpuinfo
Processor       : ARMv6-compatible processor rev 7 (v6l)
BogoMIPS        : 697.95
Features        : swp half thumb fastmult vfp edsp java tls
CPU implementer : 0x41
CPU architecture: 7
CPU variant     : 0x0
CPU part        : 0xb76
CPU revision    : 7

Answer 2

Mathematica, which is available now for the Rpi, has a benchmarking tool. One example is to calculate a discrete fourier transform. I timed the operation five times each on an RPi and Qemu-RPi on a Xeon 6-core Windows 64-bit system and obtained the following:

• RPi 5.55s (0.7% variability)
• Qemu 12.9 (1.4% variability)

So it looks like the emulator is taking a roughly 50% speed hit under these conditions.

For those interested, here is the Mathematica code I ran:

Table[Module[{data},Timing[SeedRandom[1];data=RandomReal[{},{120000}];Do
[Fourier[data],{11}]]],{5}]