Hardware

PogoLinux: Taking tight budgets to the limits

PogoLinux is offering a new configuration for its Altura spec that allows users on a tight budget to enjoy amazingly high speeds, large-capacity hard drives, and dual-booting, among other features. Here, Jack Wallen describes what you'll find.


The title of this article may seem a bit misleading, so let me explain. PogoLinux is offering a new configuration for its Altura spec that allows users on a tight budget to enjoy amazingly high speeds, large-capacity hard drives, dual-booting, Ethernet, sound, solid video, a removable drive bay, a standard three-button mouse, the PogoLinux version of RedHat 6.2 (and other software), and a simple users manual.

Although the majority of this review (including any testing) will take place in the Linux environment, please note that PogoLinux will install Windows Me (for $90 more) or Windows 2000 Pro (for $152 more). By having PogoLinux do the installation, the newer Linux users can begin their journey into Linux-land much, much faster!
The particular machine I used for this review included a dual boot of Linux and Windows Me.
The lowdown
So what makes the above package, and the $999 price tag, so enticing? Well, for one you're getting a fairly hefty workhorse of a machine. The Altura comes packed with an Athlon Thunderbird running at 1000 MHz! That's a gig, people... and that's fast! Along with this phat processor comes the following:
  • ·        Global Win CPU fan
  • ·        ASUS A7V Motherboard with UDMA-100
  • ·        128-MB PC133 SDRAM (Micron Brand)
  • ·        IBM 20-GB 7200-RPM UDMA-100 HD
  • ·        Mobile hard drive rack
  • ·        SoundBlaster 16PCI
  • ·        ASUS 7100 GeForce2 32-MB video card with TV out
  • ·        10/100 Ethernet card
  • ·        300W power supply
  • ·        A really cool case with insulated side panels

All this for $999 (plus shipping and handling)!

Add to that mix the friendly gang at PogoLinux, and you might have the makings for one of the best deals in town. But does it stand up to the specs? Let's put it through the mill and find out.

Unpacking and plugging in
As with any new computer, the first moments are critical. We all hope for a painless first encounter with our computer, and the PogoLinux Altura nearly hit that mark with perfect stride.

The first problem came about when I tried to shave a bit of time and effort (I'm working with a broken arm here... give me a break) by using my current mouse and keyboard. The keyboard gave me no problems; however, the mouse did. The computer was set up to use the supplied three-button mouse (Logitech). At installation, the Red Hat kudzu tool caught that a different mouse had been plugged in and asked if I wanted to remove the configuration. I did, assuming that I could easily run the mouseconfig tool after startup. Well, as most sellers of computers today would assume, PogoLinux had this baby set up for the average-to-less-than-average computer user, so the run level was set to 5!

Say what? you ask. With the run level set to 5, the computer would immediately boot to the graphical login screen. This is fine if there are no graphical problems. However, should there be a problem (such as my mouse issue), you’ll need to use the old linux single boot mode so you can change the section:
id:5:initdefault:

of the /etc/inittab file to:
id:3:initdefault:

That’s a fairly simple procedure but not one that the average-to-less-than-average computer user would know. And if you’re new to Linux? Forget it. Of course, this would be a non-issue had I simply used the supplied mouse. So much for shortcuts.

Once the mouse issue was over, it was only a matter of setting up the usual accounts, changing root’s password, loading up the necessary programs, testing utilities, and hammering on this baby!

The first test: The CPU/memory benchmark
The first test will be a benchmark handled by the lmbench benchmarking application. The lmbench-2alpha11.tgz file can be downloaded from the BitMover site and is a bit tricky to figure out. Obviously, the first step in using this tool is installation.

Installation of the lmbench tool is simple. To begin, untar and unzip the package with one command:
tar xvzf lmbench-2alpha11.tgz

Once the file is untarred and unzipped, cd into the new directory with:
cd lmbench-2alpha11

and run the following commands (as root):
make

which installs the code, and:
make rerun

which actually runs the software on that particular machine.

During the run of this benchmark, read the instructions that pass by. You’ll be asked some questions regarding memory, CPU, device name, fast memory, and so on. These results are fairly extensive and, to most, gibberish. However, there are some results that are quite telling for this machine, and here they are for your eyes.

Here are the most valuable numbers for the PogoLinux Altura:
Simple syscall: 0.3417 microseconds
Simple read: 0.6199 microseconds
Simple write: 0.3328 microseconds
Simple stat: 3.5659 microseconds
Simple fstat: 0.4996 microseconds
Simple open/close: 4.4233 microseconds
Select on 10 fd's: 4.9161 microseconds
Select on 100 fd's: 43.6667 microseconds
Select on 250 fd's: 100.3818 microseconds
Select on 500 fd's: 122.7778 microseconds
Signal handler installation: 0.769 microseconds
Signal handler overhead: 0.912 microseconds
Protection fault: 0.602 microseconds
Pipe latency: 5.0437 microseconds
AF_UNIX sock stream latency: 10.4165 microseconds
Process fork+exit: 136.7073 microseconds
Process fork+execve: 734.1250 microseconds
Process fork+/bin/sh -c: 4054.0000 microseconds
File /usr/tmp/XXX write bandwidth: 20186 KB/sec
Pagefaults on /usr/tmp/XXX: 1303 usecs


If you don't quite understand the above, let me give you a point of reference. Another machine I use—a Dell that houses a Pentium III 550 with 128 MB of RAM—came up with these numbers (using lmbench):
Simple syscall: 0.5515 microseconds
Simple read: 0.8673 microseconds
Simple write: 0.7283 microseconds
Simple stat: 5.2748 microseconds
Simple fstat: 1.1162 microseconds
Simple open/close: 6.5548 microseconds
Select on 10 fd's: 5.9946 microseconds
Select on 100 fd's: 41.3684 microseconds
Select on 250 fd's: 93.9273 microseconds
Select on 500 fd's: 198.3704 microseconds
Signal handler installation: 1.633 microseconds
Signal handler overhead: 2.074 microseconds
Protection fault: 1.243 microseconds
Pipe latency: 5.8566 microseconds
AF_UNIX sock stream latency: 13.2929 microseconds
Process fork+exit: 270.8095 microseconds
Process fork+execve: 1359.0000 microseconds
Process fork+/bin/sh -c: 5693.0000 microseconds
File /usr/tmp/XXX write bandwidth: 3873 KB/sec
Pagefaults on /usr/tmp/XXX: 475 usecs


As you can see, there are moments when the Altura more than doubles the speed of the PIII 550. As well it should... we're looking at a 1,000-MHz versus a 550-MHz. The reason for the comparison was to show you that the Altura from PogoLinux is right on spec with that from a much larger company.

The second test: Networking benchmarks
We're focusing on benchmarking the hardware and not the server software. Apache is the obvious choice for Linux users (since we're currently working on the Linux platform), and so we don't have to install third-party software, we'll run the server benchmarks from there.

Before I ran the server-benchmarking software, I decided to do a little real-world benchmarking. On two different machines, the PogoLinux Altura and the Dell PIII 550, I downloaded a 97-MB file. On the Dell side of the world, the download took 21 minutes, 15 seconds; on the PogoLinux side, the download took a mere 7 minutes, 34 seconds. That's nearly three times as speedy as the Dell! Is this a fair test?

The next step is the actual server-benchmarking tool. I used the tool netperf, which can be snagged from the official Netperf site.

I ran the test on two machines: the Altura and the above-mentioned Dell. The test sends and receives packets and then calculates the time and throughput.

The command I ran on the Altura (from the remote Dell) was:
./netperf -a 16384,16384 -A 16384,16384 -l 60 -H willow —
-s 2097152,2097152 -S 2097152,2097152
Enter the above command as one continuous line.
The results were:
TCP STREAM TEST to willow
Recv Send Send
Socket Socket Message Elapsed
Size Size Size Time Throughput
bytes bytes bytes secs. 10^6bits/sec

131070 131070 131070 60.66 1.44


The same command run on the Dell (from the remote Altura) gave me the following results:
TCP STREAM TEST to giles
Recv Send Send
Socket Socket Message Elapsed
Size Size Size Time Throughput
bytes bytes bytes secs. 10^6bits/sec

131070 131070 131070 69.20 0.12


Searching the netperf database (go to the Netperf Numbers link and then click Search The Netperf Database), I found the best results (from the same command) were from an SGI O2000 with 8 CPUs, which looked like:
TCP STREAM TEST to dozer3-gsn0
Recv Send Send
Socket Socket Message Elapsed
Size Size Size Time Throughput
bytes bytes bytes secs. 10^6bits/sec

2097152 2097152 2097152 59.99 4470.82


Note that the first two tests were actually run over the @home cable modem service. Because of the network variable, the comparison to the SGI is really a bit skewed (besides the fact that we're sparring single-CPU machines against an 8-CPU machine). A fair test? Not really, but it's always nice to see how you stack up against the big guns.

The third and final test: Booting into Windows and running Wintune
I have to first say that I was amazed at how well the Windows side of things was set up. All I had to do was enter the computer machine name (in order for @home DHCP to work) and all was great.

Wintune is an online benchmark application that can benchmark your machine and compare it to other, similar benchmarked machines. I'll cut to the chase and show you the top two results (from the comparison):
 CPU MHz Name Secondary Memory MBPS Cached MBPS Uncached MBP
Athlon1009 Altura 2653 140 6.96
Athlon1009 Unnamed 2556 154 5.37


As you can see, the PogoLinux Athlon ranked first in its category. The numbers are close (and this is not a complete listing of the benchmark numbers), so let the ranking speak for itself.

Conclusion
With the Altura, applications snap open with amazing speed. The rumors of the Gig Athlon CPUs being unstable are obviously unfounded (at least not with the Altura Thunderbird), and the heat issue is obviously not a problem with the well-designed case of the Altura.

I've had a number of dealings with PogoLinux. This is a company that honestly cares about what it’s shipping and to whom it’s shipping. As I've executed this review, the CEO, Tim Lee, has asked for my input on how this new configuration can be improved. To that question, I can honestly say the only way to improve this machine would be to ship it ASAP!

The numbers don't lie. The benchmarks show that this machine is well up to snuff with its competition. But why buy from PogoLinux? Simple: price and service. There are few places to find a 1,000-MHz equipped machine (one that can work as a low- to mid-level server) for under a grand.
The authors and editors have taken care in preparation of the content contained herein but make no expressed or implied warranty of any kind and assume no responsibility for errors or omissions. No liability is assumed for any damages. Always have a verified backup before making any changes.

About Jack Wallen

Jack Wallen is an award-winning writer for TechRepublic and Linux.com. He’s an avid promoter of open source and the voice of The Android Expert. For more news about Jack Wallen, visit his website jackwallen.com.

Editor's Picks

Free Newsletters, In your Inbox