<html>
<head>
<meta content="text/html; charset=UTF-8" http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#FFFFFF">
On 05/05/2012 04:09 PM, dave taht wrote:
<blockquote cite="mid:4FA5B320.8050207@gmail.com" type="cite">On
05/05/2012 03:48 PM, dave taht wrote:
<br>
<blockquote type="cite">On 05/05/2012 03:39 PM, Eric Dumazet
wrote:
<br>
<blockquote type="cite">On Sat, 2012-05-05 at 15:34 -0700, dave
taht wrote:
<br>
<blockquote type="cite">On 05/05/2012 03:09 PM, Eric Dumazet
wrote:
<br>
<blockquote type="cite">On Sat, 2012-05-05 at 15:03 -0700,
dave taht wrote:
<br>
<br>
<blockquote type="cite">Maybe on your arch, but highly
doubtful on a 680Mhz mips that isn't even
<br>
superscalar.
<br>
<br>
</blockquote>
CPU are fast, memory is slow.
<br>
<br>
<blockquote type="cite">I'd prefer to leave it in and be
able to compile it out, and actually
<br>
measure the difference.
<br>
</blockquote>
You optimize the case where there is no need to optimize
(small queue)
<br>
<br>
I can see count bigger than 100000 with 20 concurrent
netperf
<br>
<br>
This makes no sense to have a cache so big.
<br>
<br>
Or there is a bug in codel
<br>
</blockquote>
The original reciprocol approximation test code rapidly goes
AWOL after
<br>
exceeding 2^8.
<br>
<br>
I went looking for butterflies and didn't see any in the
scaled code in
<br>
the range 0-100000,
<br>
and they would only take flight briefly, so...
<br>
<br>
However I have not corrected it for BITS_PER_LONG as per our
4AM
<br>
discussion.
<br>
</blockquote>
You should use the exact code in kernel. (using BITS_PER_LONG)
<br>
<br>
<blockquote type="cite">...
<br>
<br>
interval/sqrt(99999)=316229 approx :6250190 19.76475908
interval/scaled:
<br>
316236 1.00002214
<br>
<br>
<blockquote type="cite">
<br>
</blockquote>
</blockquote>
If you read the code , there is no possible overflow, even
with very
<br>
large 'u32 count'
<br>
<br>
anyway the problem is q->count keeps increasing under load.
<br>
<br>
Only when load is stopped for a while, count is reset to 1
<br>
<br>
</blockquote>
Stalking butterflies. (
<a class="moz-txt-link-freetext" href="http://en.wikipedia.org/wiki/File:Lorenz_attractor_yb.svg">http://en.wikipedia.org/wiki/File:Lorenz_attractor_yb.svg</a> )
<br>
<br>
I suspected also we would have issues as we hit some natural
quantums (clock rate/interrupt rate/bql estimator etc) but for
all I know it's just a plain bug. I need a reboot. Goin to
dinner.
<br>
<br>
</blockquote>
<br>
If we just compare us to us rather than ns to ns you get chunkier
drops, by a lot...
<br>
</blockquote>
truncate at 10 * <b><a
href="http://en.wikipedia.org/wiki/Mu_%28letter%29" title="Mu
(letter)">µs</a> </b>or .1ms<br>
<br>
seriously going to dinner now<br>
<blockquote cite="mid:4FA5B320.8050207@gmail.com" type="cite">interval/sqrt(370)=5198752
approx :6250190 1.20224815 interval/scaled: 5198761 1.00000173
<br>
interval/sqrt(371)=5191741 approx :6250190 1.20387169
interval/scaled: 5191776 1.00000674
<br>
<br>
<br>
secondly adding a fudge factor to the calculation would bring it
closer to inline with an actual sqrt.
<br>
<br>
<br>
<blockquote type="cite">
<blockquote type="cite">
<br>
<br>
<br>
<br>
<br>
</blockquote>
<br>
</blockquote>
<br>
</blockquote>
<br>
</body>
</html>