Re: Backporting DFly patches to FreeBSD?

[EM1897@xxxxxxx]

In a message dated 2/28/2005 12:00:18 PM Eastern Standard Time, Freddie Cash <fcash-ml@xxxxxxxxxx> writes:

>On February 26, 2005 11:12 am, EM1897@xxxxxxx wrote:
>> In a message dated 2/26/2005 1:40:20 PM Eastern Standard Time, Tom 
>Hummel <tom@xxxxxxxxxxxxx> writes:
>> >> Could you please phrase your comments in the form of a sentence,
>> >> so we can understand what you're attempting to say.
>> >
>> >I admit, there's one "be" too much, id didn't notice.
>> >Speak for yourself, maybe there's some creature out there, which
>> > understood.
>> >
>> >My opinion on this matter is, that nobody needs comments which
>> > say:"but it has a habit of doing so at the expense of things that
>> > others would consider unacceptable.".
>> >
>> >a) It would be practical if you'd explain which 'things' you are
>> > talking about and what makes them unacceptable in your opinion.
>> > Saying "few matters aren't good", is not helping anybody to
>> > understand your opinion, neither understand what maybe is wrong
>> > there.
>> >It is like saying:"Black is considered unacceptable".
>> >Reader:"What's so unacceptable about black?".
>> >
>> >b)If i would be aware of your concerns in detail, there's no need
>> > for me to read your statement. I wouldn't even be able to know if
>> > we are talking about the same thing. So the reader might
>> > think:"Yes, black is making my pets go upside down", while the
>> > sender infact means:"black is taking all the light from my rooms".
>>
>> I believe I did qualify my point, here's what I wrote:
>>
>> "Be careful how you define "better". Linux often appears
>> to the naked eye as being "faster", but it has a habit
>> of doing so at the expense of things that others would
>> consider unacceptable. Linux drops packets quite often
>> at 60% usage, while 'BSDs never do. Linux gives you the
>> negatives of device pplling even when you aren't device
>> polling, making it unsuitable, IMO, for high-end
>> networking. But its pretty good for gaming I hear."
>>
>> Linux dropping packets is "the expense". While FreeBSD
>> will show 70% usage at 250K pps, linux might show 60%,
>> except linux is only actually processing 240K packets,
>> and its dropping the rest, which probably isn't what
>> most would want. Sorry if I wasn't clear.
>
>[This is for Tom, who's message I deleted.]
>
>I don't have the reference handy at the moment, but there was an article 
>about a network test where FreeBSD was able to process and route 1 
>million IP packets per second.  Linux on the same hardware was only 
>able to process approx. one hundred thousand IP packets per second.

That doesn't sound even remotely credible. Maybe FreeBSD
was "routing" to the loopback device?

We're not to a million pps yet; half of that is ambitious.
And for performance to be THAT different there would have
to be something seriously wrong with settings. The raw
performance of the FreeBSD and Linux is similar, as the
limiting factors are generally external.

I thought this thread was dead?