This is the mail archive of the gdb-patches@sources.redhat.com mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: sim/mips/dv-tx3904tmr.c questions


Hi -

On Wed, Dec 13, 2000 at 07:53:04AM -0600, Joel Sherrill wrote:
: [...]
: What clock speed is the JMR3904 the board is simulating?  

The overall mips sim clock is not scaled to real-world units such as
Hertz.

: I see the comments in dv-tx3904tmr.c that it has the following properties:
: 
:    clock <ticks>
:    Rate of timer clock signal.  This number is the number of simulator
:    ticks per clock signal tick.  Default 1. [...]
:    ext <ticks>
:    Rate of "external input clock signal", the other clock input of the
:    timer.  It uses the same scale as above.  Default 100.

These two numbers are divisors for the timer's simulated clock inputs.
They divide the hypothetical CPU clock.  (There are two alternatives,
since the tx3904 on-board timer has two software-selectable clock
inputs.)


: In practical terms, what does this mean?  Is the simulator clock
: counting instructions or cycles?   Each count on the timer corresponds
: to what?

Most simulators in the gdb collection count instructions.  A few are
somewhat aware of pipelines and instruction latency effects, but they
are generally approximate.


: Also, on general terms, how accurate is this simulator? Functionally
: accurate, cycle accurate?

It varies.  Cygnus/Red Hat-authored simulators tend to be functionally
accurate, i.e., an "instruction set simulator".  More detailed
clock cycle counting tends to be an extra-cost option.


- FChE

PGP signature


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]