Bang for the Buck, comparing Celerons to Xeons (and PIIIs?)

[Alan DuBoff]

George Bonser wrote:

> You are seeing the benefit of SMP. By specifying -j4 I am able to use both
> processors.

This has been very educational for me, since you really can't specify that
type of parameter on Solaris. Solaris really controls the processors itself,
but this would imply that Linux doesn't have the same amount of control. I've
been getting mixed results with -j2, -j4, and -j6, but as you mention, -j4
seems to be a decent happy medium with no benifit really seen with 6.

> There are still a few bugs in 2.3.18 ... one SCSI controller would not compile
> for me but I will see how it runs. Look for Alan Cox to put out some 2.3.18ac
> kernels while Linus is away. They will likely sync up in the pre-2.3.19 series
> when he gets back.

I don't hang out much on the kernel list, just happened to notice the post on
Freshmeat this past friday.

> I have additional case fans as well. One is controlled by the SuperMicro mobo
> when the CPU temp reaches the threshold. Simple rule of thumb ... convection is
> your friend. Draw air in at the bottom and exhaust it at the top. Case temp
> should always be measured at the top center of the case if you are monitoring
> the temp. Most mulitmeter manufacturers sell temperature probes now.

I like the design of the SuperMicro cases, they use this architecture, with
the 2 fans pre-installed in the case, the exaust on top, and one over the CPUs
on the bottom.

The ABit mobo has a lack of fan connectors, the SuperMicro S2DGE mobo I have
(dual Xeons) has about 6 fan connectors...of course Xeons have 4 mounted on
the sides of the processors, 2 on each side. But with the ABit, after using
the 2 for the processors, there is only one left which I use for the one fan
over the CPUs.

I have some code I've run on Solaris x86, a program called health and it will
tell you the temp. of the CPUs. I was trying to find that for Linux, I know it
originally came from Linux, but have been a hard time finding it. I'm not sure
if the ABit mobo has the proper chipset on it to give the temps for the CPUs.
Have you heard of health?

> Use make -j4 or maybe even higher and compare the numbers. I must say the
> latest kernels are much more stable on SMP. I could kill <=2.2.10 at will with
> high network activity. I can't anymore with 2.2.12 and pre2.2.13

My take is that we really need to move to the 2.3/2.4 kernel to get the better
SMP advantage. 2.2.x seems to be a stop gap in between, but 2.3/2.4 should be
able to compete with some of the other SMP systems, like Solaris x86 and/or
even FreeBSD.

Do you know of a program that will monitor both CPUs? I would like something
like top, but that supports both processors. Do you know of anything
available?

BTW, I looked into the distributed network client last night. Where I work
they run that on a bunch of machines. I would like to join you guys in a
couple weeks, but I'm making some changes and my machine hasn't been up all
the time just yet.

I'm watching my numbers still on -j6, it's definitely slower than -j4. -j2 is
barely negligable from -j4 and at times better. Of course I'm sure it matter
what else is happening on the machine when it's running...

-- 

Alan DuBoff
Software Orchestration, Inc.