Problem Compiling Bogomips

Problem Compiling Bogomips

Post by mike » Thu, 04 Nov 1999 04:00:00



Hi,
       I am fairly new to Linux and C.  I wanted to use the bogomip
program to benchmark various kernels and systems. In the Bogomips-
HOWTO the describe the program. I downloaded it from
ftp.metalab.unc.edu. I have Redhat 5.1 and can't  seem to compile
bogomips. I got the gziped file and unzipped it. I did "make" and got
the following answer:

', needed by `bogomips'.  Stop.t `bogomips.c

  I tried using the gcc compiler and did "gcc bogomips.c " and got:

/usr/include/delay.h: In function `udelay':
In file included from bogomips.c:17:
/usr/include/delay.h:47: `loops_per_sec' undeclared (first use this
function)
/usr/include/delay.h:47: (Each undeclared identifier is reported only
once
/usr/include/delay.h:47: for each function it appears in.)

    I have heard that old programs might have problems due to
different compilers and libraries. This program, I believe, is
from 1994.
        Could you please help solve this problem as I would
like to use the bogomip benchmark program.

                        Thanks
                                Mike

 
 
 

Problem Compiling Bogomips

Post by Shawn D'Alimont » Thu, 04 Nov 1999 04:00:00



> bogomips. I got the gziped file and unzipped it. I did "make" and got
> the following answer:
> ', needed by `bogomips'.  Stop.t `bogomips.c

Is the Makefile in MSDOS format?  I remember make giving me an error
like that when the Makefile had CRLF line endings.  It is telling you it
can't find the file ^M.

 
 
 

Problem Compiling Bogomips

Post by J.Blan » Sat, 06 Nov 1999 04:00:00




Quote:> Hi,
>        I am fairly new to Linux and C.  I wanted to use the bogomip
> program to benchmark various kernels and systems. In the Bogomips-
> HOWTO the describe the program. I downloaded it from
> ftp.metalab.unc.edu. I have Redhat 5.1 and can't  seem to compile
> bogomips. I got the gziped file and unzipped it. I did "make" and got
> the following answer:

> ', needed by `bogomips'.  Stop.t `bogomips.c

>   I tried using the gcc compiler and did "gcc bogomips.c " and got:

> /usr/include/delay.h: In function `udelay':
> In file included from bogomips.c:17:
> /usr/include/delay.h:47: `loops_per_sec' undeclared (first use this
> function)
> /usr/include/delay.h:47: (Each undeclared identifier is reported only
> once
> /usr/include/delay.h:47: for each function it appears in.)

>     I have heard that old programs might have problems due to
> different compilers and libraries. This program, I believe, is
> from 1994.
>         Could you please help solve this problem as I would
> like to use the bogomip benchmark program.

First, did the distribution contain a configure script? If so, run that
before compiling. Or read any README or INSTALL files that are in there.

Second, don't trust BogoMips as an absolute indication of system speed. In
fact, be quite suspicious of all benchmarks unless you know exactly what
they're doing. BogoMips is a quite small program, that these days tends to
sit inside the L1 cache of cpus and only uses integer instructions. So, it
doesn't give you any indication of how the: bus, harddiscs, RAM, FPU etc
affect the overall performance of your system.

Also, it should already be on your system. My 3.0.3 RedHat system (yes, yes,
well it is ARM you know...) has it. If you have a file /proc/cpuinfo it
should list in that the BogoMips rating amongst other things.

As an indication of how a small simple program will run on your box it's
fine. Otherwise it's just misleading. The best benchmark is to just use the
different machines with the applications you will be running and see which
performs best or quickest.

Shrike

--
 John Bland                            Webmaster and Ph.D. Research Student,
 M.Phys (Hons) Grad.Inst.P              Condensed Matter Physics Department,

 http://www.sliced.uk.eu.org/~shrike  http://www.liv.ac.uk/~olmsg01/physics/
"OK! We're young, rich and full of sugar! What do we do?" - Simpsons

 
 
 

1. BogoMips absurdly low -- is this a problem?

: I read the BogoMips faq, which lead me to believe my DX4-100 (intel) should
: get a bogomips of about 50.  Instead it's 3.10!  Is this a problem?  The
: system seems to run just fine, but I basically use it as an X terminal these
: days.  It's plenty fast under DOS, so I'm not overly worried, but it is
: strange!

Hmmmm.  That does sound a tad on the low side.  How fast can you
compile your kernal?  My system, a 486-33 w/ 8MB ram, an IDE drive
on an ISA controler and 16MB swap can do a bare bones kernal (1.2.3) in
about 30min at the console.  My system gets a tad over 16bm with turbo
on and around 4.5 w/ turbo off (I have the turbo switch disabled
anyway so I don't recal what the actual value is, its been a long
time).  Sps1, a 486dx66 gets around 32 or so.  I would take it as a
sign that something is wrong somewhere.

        Rob

--




2. Apache authentication

3. Bogus BogoMips - udelay problems

4. SIOCADDR problem

5. Cx486DX4/100 problems ( 3.02 Bogomips, NO WAY!)

6. WARNING: serial: Garbage

7. problems with low BogoMIPS / can't run with turbo

8. Anybody using Realaudio Player?

9. Problem with Bogomips

10. How to solve BogoMips problem

11. "--mode compile" instead of "--mode=compile" problem with libtool

12. Qt compile problems and compiling in general

13. Problems compiling perl5 to compile satan to test my security