[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: etrax100lx bogomips



On Tue, 26 Jun 2001, Jerome Lefranc wrote:
> I have recompiled linux kernel 2.4.3 for etrax100lx devboard
> and I obtain 20 bogomips at boot start.
> 20 bogomips seems weak for that kind of processor: is it the
> normal bogomips rate ?

It's because the number of instructions in the udelay asm-loop is not the
same as the number in the i386 port (which the BogoMIPS presentation code
is made for). So if we have 4 times as many instructions in one
delay-loop, the bogomips value will be 4 times as low. But don't worry
everything is still running at 100 MIPS, it's just the calculation of MIPS
from delay loops per second which is broken.

In the 2.0 port, we changed init/main.c to report using the proper
multiplication value but in the 2.4 port we don't want to fiddle too much
with non-arch code.

Because of different loop behaviour the bogomips value does not accurately
represent the CPU speed in many other linux ports either, like Mikael
wrote.

/BW