[GRLUG] virtual box/cpu speed"

Michael Mol mikemol at gmail.com
Thu Mar 24 07:46:56 EDT 2011


Benchmark results vs part price, based on a website Ben Rousch linked to on
IRC a month or two ago. I couldn't (and can't) easily grep my logs right
now, or I'd cite...
On Mar 24, 2011 12:49 AM, "Bob Kline" <bob.kline at gmail.com> wrote:
> Not sure what's meant by the sweet spot
> being at three processors. Why?
>
> A couple of observations. Processors with six
> CPUs have been available at popular
> prices for a while now. I take it if the sweet
> spot is three, then the incremental performance
> by having six is less than 2X - how much less?
>
> The taskset command sets an affinity for a
> CUP, which I take it is short of actually
> being able to assign a process to a CPU,
> and might be that way because one is competing
> with the Linux scheduler. But where physical
> CPUs are actually available, isn't taskset a
> way of taking advantage of an arbitrary number
> of CPUs? i.e., performance doesn't actually
> plateau as one adds CPUs? That would be
> true, if true at all, as long as there are always
> enough processes that need a CPU.
>
> So, again, what does a sweet spot of three
> CPUs mean?
>
> -- Bob
>
>
> On Wed, Mar 23, 2011 at 11:21 PM, Michael Mol <mikemol at gmail.com> wrote:
>
>> Sure, there's a speedup with quad-core. Your computer has dozens of
>> processes, and the ability to service more at the same time reduces
latency.
>>
>> Additionally, some server services scale very well to multiple cores.
>> HTTPd, for example. If you use soft RAID on Linux, and have the
appropriate
>> kernel option enabled, checksum calculations for RAID modes 4, 5, and 6
will
>> be split across your cores.
>>
>> Desktop apps are catching up, too. If you use Firefox, Flash is kept as a
>> separate process, a scenario which benifits from more cores (see my note
on
>> latency near the beginning). If you use Chrome or Chromium, each _tab_ is
a
>> separate process, which leverages multicore for app-wide performance
>> improvements.
>>
>> These days, hanging at two cores when looking at making a purchase
doesn't
>> make sense.
>>
>> Right now, I believe the price/performance sweet spot is at three cores.
It
>> will probably be at four by the end of summer.
>> On Mar 23, 2011 10:25 PM, "west mi" <west.mi420 at gmail.com> wrote:
>> > Do you think there is a significant speed difference between the dual
and
>> > quad core cpu's?
>> > I havent used anything but single and dual cores.
>> > I switched several years back to dual core, and did notice a
significant
>> > speed up.
>> >
>> > Do you think vbox can fully utilize a quad core?
>> > I hesitate on going to a quad core, because I don't know if today's
>> software
>> > can
>> > fully utilize 4 cores.
>> >
>> > thanks,
>> > Darrin
>> >
>> > On Wed, Mar 23, 2011 at 11:55 AM, Topher <topher at codeventure.net>
wrote:
>> >
>> >> On Wed, 23 Mar 2011, west mi wrote:
>> >>
>> >> Anyone have win7 working in vbox?
>> >>> And does it work good?
>> >>>
>> >>
>> >> I'll chime in.
>> >>
>> >> I have a 32bit win7 install on my 32bit Arch linux laptop. The host
has
>> 2G
>> >> of ram and I give one to the vm. I use Photoshop in it and it works
just
>> >> fine.
>> >>
>> >> I copied that vm to my 64bit arch host and that went flawlessly. Now
>> that
>> >> vm has 2G of its own, and 2 of my 4 processors. Still using photoshop,
>> but
>> >> it screams right along. It takes 5-7 seconds to go from power on to
>> login,
>> >> and maybe 12 seconds to reboot.
>> >>
>> >> I really like VirtualBox.
>> >>
>> >> topher
>> >>
>> >>
>> >> --
>> >> This message has been scanned for viruses and
>> >> dangerous content by MailScanner, and is
>> >> believed to be clean.
>> >>
>> >> _______________________________________________
>> >> grlug mailing list
>> >> grlug at grlug.org
>> >> http://shinobu.grlug.org/cgi-bin/mailman/listinfo/grlug
>> >>
>> >
>> > --
>> > This message has been scanned for viruses and
>> > dangerous content by MailScanner, and is
>> > believed to be clean.
>> >
>>
>> --
>> This message has been scanned for viruses and
>> dangerous content by *MailScanner* <http://www.mailscanner.info/>, and is
>> believed to be clean.
>> _______________________________________________
>> grlug mailing list
>> grlug at grlug.org
>> http://shinobu.grlug.org/cgi-bin/mailman/listinfo/grlug
>>
>
> --
> This message has been scanned for viruses and
> dangerous content by MailScanner, and is
> believed to be clean.
>

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

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://shinobu.grlug.org/pipermail/grlug/attachments/20110324/e081353c/attachment-0001.html>


More information about the grlug mailing list