Will Beazley wrote:
> Folks,
>
> I had been working on m5-2.0b6 with 4 cores/2threads and what I found is
> that it would never complete even after months (runaway).
>
> I downloaded and compiled the later versions and then for some reason
> any attempt at running the benchmarks wouldn't work.
>
> So my questions are:
>
> 1. Is there a resolution to the runaway problem?
>
We'd need a lot more information to be able to tell what the actual
problem is. We (to my knowledge) haven't seen this happen. What
benchmarks? What ISA? What mode? Did you make any changes to M5? Do you
know if M5 itself is stuck, or if it's the simulated code? What is the
simulated code doing if/when it's stuck? Does the output look ok
otherwise? Do the statistics, if any, look strange in any way, for
instance a huge number of, say, cache misses, and only a modest number
of completed instructions?
> 2. What are the steps to transition from m5-2.0b6 to newer versions?
>
If you haven't made any changes to M5, you'll just need to download
the new version using mercurial. If you're copy is managed by mercurial,
you could try pulling in the changes from the development repository. I
would suggest backing up your tree first just in case something goes
wrong or the merge is too complicated. If you're not using mercurial,
you could create a patch and apply it to an updated repository.
> 3. A colleague said we need to recompile the benchmarks for the newer
> version, is that so?
>
I don't know of any reason you'd need to, but there might be something
I'm not thinking of.
> (also does anyone have a copy handy of the config/alpha.cfg ?)
>
> http://www.m5sim.org/wiki/index.php/SPEC2006_benchmarks#Build_the_SPEC2006_alpha_binaries
>
>
> _______________________________________________
> m5-users mailing list
> [email protected]
> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
>
_______________________________________________
m5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
