Has anyone run the ALPBenchmarks on M5 System Emulator mode?
I have built the statically linked binaries for ALPBench using a cross
compiler.
First I got the error of: 'unimplemented system call' (_systcl: which i
changed to ignore system call).
Is that a type-o? I assyme that was supposed to be sysctl? That syscall
is comparable to reading a variable in /proc, and depending on the
variable, values could be hardcoded for that.
then I get the error : fatal: Syscall 427 out of range
What version of m5 are you running? Is this M5 1.1? Syscall 427 is
fstat64 which is implemented in 2.0b2. If you want to stick with 1.1, you
could backport the syscall.
I also had another fundamental question, if in my run.py script for the SE
mode, I specify 4 CPUs, and my application has 4 threads, then it would
schedule a thread on each of the CPUs,right?
I'm no expert on the threading stuff in SE mode, but it's not just a
matter of running a threaded program in SE mode. We don't fully implement
the user-level threading stuff since you can just run in FS mode and get
everything. [ Steve, Kevin: what exactly can we do here? ]
Also, can anyone tell me the significance of the code in the encumbered
directory?
That code is stuff with a more restrictive license than our own. You can
use it provided that you are happy with the license.
Nate
_______________________________________________
m5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
