On Sat, Feb 28, 2009 at 2:27 AM, Jim Van Meggelen <[email protected]> wrote:
> Dave Donovan wrote:
>>
>> Personally, I'd be less interested in which processors do what than
>> seeing how things scale with processor power.
>>
> You mean how things behave at the limit?

No, I'm thinking about if 2GB ram and 4 Ghz gives me 150 conference
channels, does 4GB and 8Ghz give me 300 channels or more, or less?
The advantage of using a VM host is that you could adjust processor
and memory with a few key clicks and re-run your automated test.

>>
>> Are there still significant timing issues with Vmware?  If that can be
>> overcome there are some serious advantages to using it as a test
>> platform.  You can dial the CPU and memory up and down on demand.
>> Would that be beneficial or would it just muddy the waters?
>>
>
> It would both muddy the waters and be beneficial :-p

It would certainly be interesting, but the more I think of it, the
more I think that we're as likely to to be limited by some quirk of
the virtualization platform than of the software we're actually
testing.

It looks like Bill has a good platform that would get to the heart of
the original question which I understand as "Is it currently possible
to use a general purpose processor in a large scale conferencing
bridge approaching 1000 channels, or is hardware DSP required?"

On that note, there should be some agreement on exactly what premise
or configurations are to be tested/proven.

Dave

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to