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

Re: Rep:[f-cpu] vsim


here is a late answer but i guess you'll forgive,
i did not yet read the whole TLB thread anyway...

I believe you are speaking about Modelsim.
Riviera uses the same file names for its tools
"to ease user's migration from Modelsim".

I have run the tests on a PIII-550 (my crappy laptop).
it can't compare to the dual UIII when it comes
to bandwidth and cache (i have a cheap mobile Celeron...)

i think that Michael does not use imu64_test anymore,
rather the 2 and 3. i'll have to check.

I found ncsim to be fastest, with Riviera close (7% slower).
simili was 60x slower and Vanilla 2x slower than simili...
i did not run the tests on these two last, or i would have
to wait a whole day for a positive result...

But this does not qualify as a good testbench. The real one
will be the time it takes to boot the CPU, no ?

Nicolas Boulay wrote:
> Et voila : imu64_test :
> real    43m37.359s
> user    43m27.900s
> sys     0m0.400s
> so the others number ?
> nicO
> -----Message d'origine-----
> De: "Nicolas Boulay"
> A: <f-cpu@seul.org>
> Date: 08/07/02
> Objet: [f-cpu] vsim
> I have run vsim on the eu_imu. here is the result on a Ultra Blade 1650
> (bi UIII 650 Mhz 512 Mb thought NFS file system) the run use 9 Mb of
> memory. No special option are used for vcom nor vsim.
> imu64_test3 :
> real    0m12.671s
> user    0m4.830s
> sys     0m0.410s
> imu64_test2
> real    5m49.004s
> user    5m47.730s
> sys     0m0.130s
> the first test is a bit too long for today ! ;p Does somebody have the
> number for ncsim and simily ? To compare.
> nicO
To unsubscribe, send an e-mail to majordomo@seul.org with
unsubscribe f-cpu       in the body. http://f-cpu.seul.org/