xeon hyperthreading & GIL
Seb Bacon
seb.bacon at jamkit.com
Fri Feb 7 08:20:14 EST 2003
Aahz wrote:
> In article <1044530966.192487 at ananke.eclipse.net.uk>,
> Seb Bacon <seb.bacon at jamkit.com> wrote:
>>Can anyone give me the lowdown on how hyperthreading on the Xeon might
>>effect python performance?
>>
>>AFAICT the cost of context switching is very low so the impact of
>>releasing the GIL should be minimal.
>
>
> <shrug> Consider hyperthreading as functionally equivalent to multiple
> CPUs; from Python's perspective, there's no difference.
Yes, I realise that; my question was about performance WRT scheduling in
modern OSs & SMP/SMT systems, not about functionality in python.
Let me try again:
Currently python's global interpreter lock will often cause SMP system
performance to *decrease* due to the cost of context switching.
Is the same true for SMT systems, or is context switching less of an
issue due to shared cache, etc?
seb
More information about the Python-list
mailing list