lkml.org 
[lkml]   [1999]   [May]   [9]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Linux can't stay up for more than an hour?
Hi Steve.

>>>> Are you so sure? 2x SMP on intel means about 30% performance
>>>> increase, if you are lucky. SMP on intel can even _slow you
>>>> down_ for certain kind of loads, and web server may be exactly
>>>> this kind of load.

>>> That seems _highly_ unlikely, and my own experience is contrary
>>> to this. Are you sure about this?

>> It depends highly on the kind of website you're serving. If you
>> are mostly serving straight files (either html files or other
>> data) your cpu will be hardly used (relatively speaking) and the
>> bottleneck will be one of the network, your disks or the
>> throughput your machine can manage. Adding CPUs will have little
>> impact in such a case. If, however, the machine is used to run a
>> lot of scripts that do complicated things (and dont do too much
>> i/o, relatively speaking) the CPU might help.

> That's not what my response was about. The person I was
> responding to had stated (as you can see at the top) that there
> is a 30% performance increase with SMP. No stipulations about
> I/O-based bottlenecks or too-small RAM or anything. I want to
> know where he got those numbers from. Further, I would like to
> know of any real-world situations where he has actually seen
> performance decrease on a SMP system. In other words, I want
> evidence.

> I understand that when you increase the computing power of a
> system you often have to increase its available resources.
> Bigger power supply, faster disks, more RAM...no problem. But an
> example system gets only a 30% increase in capacity once you've
> added a CPU and this is a result of not enough RAM, you don't
> then say in a public forum it "means about [a] 30% performance
> increase."

I have seen a case of a system with 2xP2/400 CPU's and maxed out with
RAM perform faster on a UP kernel than an SMP one, so such is possible
with certain loads. However, general performance is near enough
impossible to measure, otherwise the marketing droids would've been
doing so for years now...

Best wishes from Riley.

+----------------------------------------------------------------------+
| There is something frustrating about the quality and speed of Linux |
| development, ie., the quality is too high and the speed is too high, |
| in other words, I can implement this XXXX feature, but I bet someone |
| else has already done so and is just about to release their patch. |
+----------------------------------------------------------------------+
* ftp://ftp.MemAlpha.cx/pub/rhw/Linux
* http://www.MemAlpha.cx/kernel.versions.html

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:51    [from the cache]
©2003-2011 Jasper Spaans. Advertise on this site