[Gbbopen-developer] portable threads on sbcl 64bit linux tests fail with pristine source

[Antony Sequeira]

sbcl --version  shows
SBCL 1.0.9

uname -a shows
Linux xxxxxxx 2.6.20-1.3002.fc6xen #1 SMP Mon Aug 13 14:21:21 EDT 2007
x86_64 x86_64 x86_64 GNU/Linux
-Antony

On 10/25/07, Dan Corkill <corkill at gbbopen.org> wrote:
> Following up more specifically on your long real-time post...
>
> Antony Sequeira wrote:
> > So, I modified the code portable-threads-test.lisp a bit ...
> > and I get
>
> > ;; Performing basic thread tests...
> > ;;   Timing (sleep 0), run time should be zero seconds... 0.0 seconds
> > 0.004 real seconds
> > ;;   Timing (sleep 10), run time should also be zero seconds... 0.0
> > seconds 10.004 real seconds
> > ;;   Timing 100000 (sleep 0)s... 0.0 seconds 400.168 real seconds
> > WARNING: 100000 (sleep 0)s took 0.0 seconds
> > ;;   Timing 100000 throwable (sleep 0)s... 0.0 seconds 400.159 real seconds
> > WARNING: 100000 throwable (sleep 0)s took 0.0 seconds
> > ;; Basic thread completed
>
> These real-time values indicate a high cost of calling sleep on SBCL
> 64-bit linux, which is not good.  I'm very surprised by the huge
> difference from SBCL 32-bit linux.  I guess it's time to look again at
> the SBCL sources to see what might be going on with SBCL's 64-bit OS
> interface code.
>
> What version of SBCL are you using (lisp-implementation-version)?
>
> -- Dan
>
> > P.S.: should I stop cc'ng the mailing list ?
>
> No, please continue to CC the list!
>
>
>

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.