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

[Dan Corkill]

Following up on the long real-time issue when running
portable-threads-test...

Antony Sequeira wrote:

> 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

I'm now able to eliminate x86_64 as the culprit, observing a similar
issue on a 32-bit Linux host running kernel 2.6.20-1.2320.fc5smp.  I
suspect the performance issue is with the kernel version (I normally run
2.6.22, and now 2.6.23) and/or related libraries.

Since the portable-threads test is stressing/timing thread support
operations, the observed real-time slowdown is unlikely to be a
significant factor in most multi-threaded application code.  However,
I've decided to leave the portable-threads-test as is, highlighting the
poorer OS-level performance with older kernels.

-- Dan


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