I haven’t done a huge amount of digging, but I must say I’ve only had positive experiences with Microsoft Autoroute 2007. Honestly, while not being tremendously slick it is perfectly functional for my needs. My needs incidentally are for a reliable, well covered (in terms of maps) GPS Navigation solution that doesn’t require an Internet connection to be available. Important distinction, that last one, because more and more we’re seeing software that maps your location by streaming a relevant section of map that applies to your current location. I suppose this explanation is timely because Google Nav has just been announced, and while it seems on paper to be a nice technical solution, as soon as you step outside of cellular coverage or happen to be smack bang in the middle of a really slow GSM area you’re going to be in trouble. By far the safer bet is to use more conventional GPS software which reads your location, and interprets that on maps it stores in local storage.

1. Get a copy of Autoroute 2007 or above! I’m currently seeing it on eBay for £30 but it’s probably cheaper elsewhere. Not least because a lot of copies come bundled with a USB GPS dongle, which most people reading this will not need.

2. Boot up Autoroute

3. Open the dialog to configure the receiver.

4. You might want to scan for the COM port that your GPS receiver sits on if you don’t know it. Also, be sure to enable the receiver in whatever dashboard is provided.

5. I’m using the Dell 5530 (Ericsoon 3507) so I need to flick the switch in that.

6. The scan in Autoroute will then see the device.

7. Now start the GPS tracking.

8. Read and accept the obligatory safety notice.

9. A GPS data pane will open showing the output of the device. Wait for a lock.
*IMPORTANT* Some people have strangely high expectations of GPS. It’s not magic and it won’t work inside buildings unless you’re very lucky - a clear view of the sky is recommended. Also, it’ll take a while to get an initial (cold) fix, so be patient. If you have got this far and you’re unable to get a fix then I’m stumped. Could be weather, could be software.

10. After a while a fix will be made. I don’t like the mini pane so I choose to use the larger dedicated GPS pane as demonstrated here.

If the demand is there I may post more in the future about other GPS packages under Windows and (especially) Linux. Ideally I would like to get the GPS data passed through into the Android Emulator, but I’m not quite sure if there’s provision for that in the software stack. I had mixed success with a similar approach using WM6 emulation, so we’ll see if that ends up working!

Let’s cut to the chase. If you spec your Mini 9/10 with WWAN, GPS, or both, it will come with a mini PCI-E card installed called the Dell Wireless 5530 (aka Ericsson F3507). If you’re feeling confident you can buy one of these separately and chuck it try installing it in your laptop, compatibility permitting. The major stumbling block however is that the WWAN (and perhaps also GPS) components require antennae which are usually secreted in the bezel around the display. When you purchase a unit specified with this card it gets built from the ground up with the required chassis elements.

Many have complained that getting this adapter working is a right royal pain in the arse, and so I’m going to write a quick guide that worked for me under XP.

First of all you need to realise that there is a driver level and an application level problem, and both need to be fixed to enable you to use the GPS on your Mini.

Driver level

Dell and others in their infinite wisdom often provide drivers that do not expose the GPS element of the card. This is most likely because you wanted WWAN but not GPS, and thus they included the 5530, but have crippled it in this admittedly inventive and annoying way. Rather cheeky methinks, but that’s their prerogative.

A quick check under device manager will let you know if you have a unit with crippled drivers. Look under Ports (COM & LPT) and there should be an entry for a Dell Wireless 5530 HSPA Mobile Broadband Minicard GPS Port (COMx) – where x = a number dependent on the amount of other bits and pieces grabbing serial ports in your system. Mine was present from the start, but I still ended up faffing around with drivers simply because I like to know I’m using the most up to date code that supports my equipment – it can make the difference between a glitchy or flawless experience. If you can see a GPS device then you can probably skip the rest of this section and go straight to the software level issue, although I would always recommend being thorough and going through a little short term hassle to grab the latest files.

A quick look on Dell Support lists the latest drivers for the 5530 as being A06, although a quick Google search will reveal forum posts noting that A07 drivers are out in the wild. I’m not sure if Dell retracted these drivers for a reason (Maybe because they include GPS functionality outright?) but the links to them on the site were all dead.

However, using a combination of the driver revision and Dell identification inferred elsewhere I found that the A7 drivers could be downloaded HERE.

Alternatively, jeffijaffa at MyDellMini extracted the drivers from the installer and uploaded them to the forum. Now, I spent a good few hours using this driver pack and I can testify that installing the 8 DRIVERS manually is not a pretty experience. Now the A7 installer is available we’ll use that in preference the far more fiddly manual method. If anyone must use a manual install then leave a comment on the article if you need any pointers.

Flick back into device manager and check the driver versions of the currently installed devices. I must say that in haste I forgot to check the driver versions that ship with the admittedly bloated preinstalled copy of XP - I’m sure someone can comment below RE this. For my fresh install with A07 drivers the versions were as follows:

Modem >>> Dell Wireless 5530 HSPA Mobile Broadband Minicard Modem >>> 4.50.6.0 >>> 19/12/2008

Networking Adapters >>> Ports (COM & LPT) >>> Dell Wireless 5530 HSPA Mobile Broadband Minicard NetworkAdapter >>> 4.50.6.0 >>> 19/12/2008

Ports (COM & LPT) >>> Dell Wireless 5530 HSPA Mobile Broadband Minicard (COM5) >>> 4.50.6.0 >>> 19/12/2008

Ports (COM & LPT) >>> Dell Wireless 5530 HSPA Mobile Broadband Minicard GPS Port (COM6) >>> 1.0.0.29 >>> 07/01/2009

Smart Card Readers >>> Dell Wireless 5530 HSPA Mobile Broadband Minicard PC SC Port >>> 6.0.0.5 >>> 06/04/2009

Universal Serial Bus controllers >>> Dell Wireless 5530 HSPA Mobile Broadband Minicard Device >>> 4.50.6.0 >>> 19/12/2008

The first job therefore is to uninstall the current driver pack, which is listed under Add/Remove programs as something like “Dell 5530 Mobile Broadband Driver Package”. Give it a reboot after you’ve done this, and cancel any dialogs asking you to provide drivers. Then simply run the executable installer for the new A07 driver package.

Application level

I tried using Autoroute & Google Maps at this point but had no joy. Although the GPS port is there, it needs to be triggered by the software, otherwise it won’t output any data. Now, it is possible to enable the GPS data port output without using additional software by opening issuing some commands through a terminal, so bear that in mind if you’re somehow a devoted fan of Vodafone Mobile Connect (VMC). I however, won’t be going through this method, although it is explained HERE.

First uninstall VMC. It’s pretty resource heavy and only works with their services - instead we’ll use the Ericsson branded dashboard which also allows us to enable GPS functionality. I’m led to believe that Dell do rebrand this and offer it themselves, but Ericsson’s version is slightly newer at 5.3.2076.12, so it makes sense to use that.

Once the Ericsson software (Wireless Manager 5) is installed load it up and browse to View >>> Settings. Under Radio Power tick the combo box to Enable radio all the time. If this isn’t done then the WWAN card - and GPS by extension, will fail to function unless the software is running and set to enable the radio.

Dell provide a basic tray utility that allows the WWAN card to be disabled at any time, and I choose to use this in preference to having the Wireless Manager permanently resident in memory. To those of you who won’t be using GPS unless you’re online, or are particularly mindful about battery life, you may wish to set the options to enable the radio only when the software is running, and disable it when it closes. Accept the warning and your settings will be saved.

The last job is to head to View >>> GPS and turn the GPS on.

Mission accomplished. Your GPS software of choice (I use MS Autoroute) will function as expected once you point it to the relevant COM port.

Recently I received a new Option Icon 505 USB HSPA adapter through a project I am currently working on. I was quite excited (in a sad way) to get it working with my Linux setup as it is a much more competent piece of kit than the telco provided Huawei units that mobile broadband customers in the UK are given (up until this point a friend of mine had been lending me his Huawei E160G). As I plan on trialling various operators and their services it was clear that I needed to get my hands on a new adapter that was:

a) Unlocked

b) Had receive diversity

c) Had support for Class 10 HSPA and beyond

d) Was supported by Linux

At first glance the Option Icon 505 looked like the perfect choice and Option themselves list the device as Linux compatible. I soon learned that it isn’t quite as simple as that however…

So, I got the item home, installed a SIM & plugged it into my fully updated Ubuntu 9.04 (Kernel 2.6.28-15) machine. Having been spoilt by newer Linux releases and the fantastic updates with the Red Hat sponsored NetworkManager I was expecting a slick driver install, wizard popup, and subsequent quick dial-up to the net. All fine in theory, but things don’t always go to plan.

The first thing that you’ll know if you’re in any way acquainted with mobile broadband on Linux is that most adapters come with a delightful false CD or Mass Storage partition which, as you would expect, appears as a CD or USB drive in Windows Explorer. Yes, Windows. The logic behind this is relatively sane - to negate the need for an actual driver CD to install the modem, because the stick itself acts as a disk in the first instance. Also, early modems took the opportunity to include a microSD storage slot so you could (in theory) use the stick as a combo pendrive & modem but this idea didn’t really catch on.

• You plug the device in and an autorun popup appears
• You agree to run the setup program suggested
• The setup launches, installs the relevant software, and flicks the device over into modem mode
• Windows looks around for some drivers, and since it’s just had them installed it magically finds and installs the relevant modem ports

The initial problem here is that Linux doesn’t autorun CDs for security, and that the false CD or Mass Storage partition actually doesn’t include any installable driver package for Linux in any case, which is not surprising given that vendors would be expected to support every variant of the OS known to man. No problem, there are usually plenty of drivers in the kernel itself, and if not, we can probably locate some of our own and compile those. Unfortunately, the side effect of presenting a faux device to any operating system is that the device needs to be sent a signal to flip itself back into modem mode since Linux in particular will match up the device, find the relevant module, and claim it to act in whatever way it thinks it should. This of course prevents any other interaction occurring between the device and the correct driver. Enough of the theory, lets take a peek at what happens when we actually plug the Icon in. Here’s the output of dmesg:

[ 1538.245056] usb 2-1: new high speed USB device using ehci_hcd and address 8
[ 1538.377935] usb 2-1: configuration #1 chosen from 1 choice
[ 1538.393133] scsi15 : SCSI emulation for USB Mass Storage devices
[ 1538.393493] usb-storage: device found at 8
[ 1538.393499] usb-storage: waiting for device to settle before scanning
[ 1543.393577] usb-storage: device scan complete
[ 1543.394575] scsi 15:0:0:0: CD-ROM            ZCOPTION Icon 505         1.00 PQ: 0 ANSI: 4
[ 1543.400526] sr1: scsi-1 drive
[ 1543.400778] sr 15:0:0:0: Attached scsi CD-ROM sr1
[ 1543.400928] sr 15:0:0:0: Attached scsi generic sg2 type 5
[ 1543.613518] sr 15:0:0:0: [sr1] Unhandled error code
[ 1543.613528] sr 15:0:0:0: [sr1] Result: hostbyte=DID_ERROR driverbyte=DRIVER_OK
[ 1543.613537] end_request: I/O error, dev sr1, sector 32
[ 1543.613548] Buffer I/O error on device sr1, logical block 32
[ 1543.613558] Buffer I/O error on device sr1, logical block 33
[ 1543.613566] Buffer I/O error on device sr1, logical block 34
[ 1543.613573] Buffer I/O error on device sr1, logical block 35
[ 1543.613579] Buffer I/O error on device sr1, logical block 36
[ 1543.613586] Buffer I/O error on device sr1, logical block 37
[ 1543.613593] Buffer I/O error on device sr1, logical block 38
[ 1543.613600] Buffer I/O error on device sr1, logical block 39
[ 1543.613612] Buffer I/O error on device sr1, logical block 40
[ 1543.613619] Buffer I/O error on device sr1, logical block 41
[ 1543.784057] usb 2-1: reset high speed USB device using ehci_hcd and address 8
[ 1543.937639] sr1: CDROM (ioctl) error, command: Get event status notification 4a 01 00 00 10 00 00 00 08 00
[ 1543.937673] sr: Sense Key : No Sense [current]
[ 1543.937683] sr: Add. Sense: No additional sense information
[ 1543.991049] usb 2-1: USB disconnect, address 8
[ 1544.924082] scsi 15:0:0:0: rejecting I/O to dead device
[ 1550.324178] usb 2-1: new high speed USB device using ehci_hcd and address 9

Ugh. Great. So it gets picked up erroneously as a CD Drive, spits out a load of read errors because it won’t play nice with the CD device driver under Linux, resets itself, and the cycle continues (trust me, this will loop). At this point you may fancy reading the official Option response to this issue instead of my guide, which is described in full within the Ozerocdoff readme file, and is…

[[Visit blog to check out this spoiler]]

Job 1 - I’m a modem, honest

USB Modeswitch used to be the undisputed king of flipping devices back into modem mode (the computing equivalent of a slap & shake) but it doesn’t work at all reliably with Option devices. Instead, Option released their own tool for this called rezero. Unfortunately rezero turned out to be pretty unreliable and so Ozerocdoff was born. Catchy. This enumerated CD mode is often referred to as ZeroCD you see, as some ironic reference to the fact that under Windows it requires ‘Zero’ configuration.

Ozerocdoff jumps in when it sees the modem, and prevents the USB Mass Storage module from claiming the device. It recognises the modem by virtue of the vendor and product ID, which you can retrieve by issuing the command lsusb. Note that if you are using a Linux machine that doesn’t play nice with this particular adapter you may need to repeatedly issue to the command to catch the system before it resets the device. Alternatively you get get the required info from Device Manager in Windows.

Bus 002 Device 040: ID 0af0:d055 Option

The first job therefore is to download Ozerocdoff. Once downloaded, place the file somewhere you can easily access it, like the desktop.

1. Switch to the terminal

2. Change directories to the desktop - cd Desktop

3. Extract the file you just downloaded - tar zxf udev.tar.gz

4. Change to the newly extracted directory - cd udev

5. Compile - sudo make

6. Install - sudo install

I, and many others have had issues compiling.

a) You need of course to make sure you have a compiler installed - build-essential

b)You need the kernel headers appropriate to your system. Issue uname -r and make a note of the output. Replace that command in the following with the output given - sudo apt-get install linux-headers-`uname -r`

c) You also need the USB headers, which can be found in libusb-dev

If you see errors similar to warning: implicit declaration of function then you’re affected by step c in particular.

A successful install should look something like the following:

richard@holly-jaunty:~/Desktop/udev$ sudo make
cc -c ozerocdoff.c -Wall -O
ozerocdoff.c: In function ‘main’:
ozerocdoff.c:243: warning: ‘usb_dev’ may be used uninitialised in this function
cc -l usb -o ozerocdoff ozerocdoff.o
richard@holly-jaunty:~/Desktop/udev$ sudo make install
install -d /usr/sbin
install -d /etc/udev/rules.d
install ozerocdoff /usr/sbin
cp hso.udev /etc/udev/rules.d/51-hso-udev.rules
install -d /usr/share/hal/fdi/preprobe/20thirdparty
cp 10-wwan-hso-preprobe.fdi /usr/share/hal/fdi/preprobe/20thirdparty
install -d /usr/share/hal/fdi/information/20thirdparty
cp 10-wwan-quirk.fdi /usr/share/hal/fdi/information/20thirdparty
install -d /usr/lib/hal/scripts/
install hal-serial-hsotype /usr/lib/hal/scripts/
install -d /etc
install osetsuspend /usr/sbin
cp hso-suspend.conf /etc
richard@holly-jaunty:~/Desktop/udev$

Also, now is a good time to check that your user is in the dialout group - grep dialout /etc/group dialout

Now, the Icon 505 is actually too new to be included by default in the list of devices that Ozerocdoff will act on behalf of, so we do need to tell it is how to identify our adapter. This is where the info gathered with lsusb comes in.

In the current version of Ozerocdoff, the file that we need to edit is called /etc/udev/rules.d/51-hso-udev.rules but it looks like this is incremented for every new version of the software. An easy way to find out the filename is to take a peek at the output given when you issued sudo make install during the installation process.

1. Switch to the terminal

2. sudo gedit /etc/udev/rules.d/51-hso-udev.rules

3. Under “New Syntax” you will see ten or more almost identical lines of code each referencing a different device (idProduct), albeit from the same vendor (idVendor). Copy and paste the last statement and alter it to suit the product ID for you device. Mine looks like the following:
ATTRS{idVendor}==”0af0″, ATTRS{idProduct}==”d055″, RUN+=”/usr/sbin/ozerocdoff -wi 0x%s{idProduct}”
GOTO=”hso_end”

4. Under “Old Syntax” you will see a similar arrangement, the only difference being a more obfuscated set of commands. Again, copy and paste the last statement and alter the product ID. Mine looks like the following:
SUBSYSTEM==”usb_device”, SYSFS{idVendor}==”0af0″, SYSFS{idProduct}==”d055″, SYSFS{bDeviceClass}==”00″, RUN+=”/usr/sbin/ozerocdoff -wi 0x%s{idProduct}”

5. Save and exit the file

6. Refresh the udev rules, so  that the next time you plug in a device, it will read the freshly created rules and thus act upon them: sudo udevadm control –reload-rules

7. Plug in your Icon. Give the system ~20secs and then check the output of dmesg. You should notice that the device no longer gets claimed as scsi 15:0:0:0: CD-ROM            ZCOPTION Icon 505         1.00 PQ: 0 ANSI: 4

8. Although in all likelyhood nothing else will step in to fill the void if you’re using another member of the Option family it may be that the Option driver steps in to try and cater to the device, but since the Icon 505 and others use a new IP only architecture it simply won’t work properly. We need to blacklist the Option driver by editing a configuration file. Issue sudo gedit /etc/modprobe.d/blacklist.conf and append the following:
# Blacklist the option modem driver. This will allow the HSO driver (when installed) to step in and provide functionality to the Icon 505 amongst others
blacklist option

Job 2 - Drivers please

1. At this point Ozerocdoff should be working and preventing  the Icon from registering itself as a CD Device. However, it still needs some drivers in order to function. The drivers that cover these new WWAN devices are called HSO - High Speed Option, and have a new architecture to allow use of fast HSPA adapters. You can download the newest version (currently 1.12) here.

2. Download or move the archive to the desktop and extract it using tar zxf hso-1.12.tar.gz

3. Move into the new directory using cd hso_26-v1.12

4. Compile the driver using sudo make

5. Install the driver using sudo make install

6. Plug in your adapter and you should be up and running. As with any issues, I would heartily recommend a reboot or two, and a double-check of all steps before putting it down to a bad effort. As I was putting together the above steps I was trying it on my fresh 9.04 box and scratching my head as  to why I was having no luck. Turns out I was using d055 as the product ID instead of d050! A successful installation should push out a dmesg log like this:

[  111.852198] usb 2-2: new high speed USB device using ehci_hcd and address 4
[  111.986505] usb 2-2: configuration #1 chosen from 1 choice
[  113.194548] usb 2-2: USB disconnect, address 4
[  113.468210] usb 2-2: new high speed USB device using ehci_hcd and address 5
[  113.602299] usb 2-2: configuration #1 chosen from 1 choice
[  114.293835] Initializing USB Mass Storage driver…
[  114.294067] usbcore: registered new interface driver usb-storage
[  114.294074] USB Mass Storage support registered.
[  114.327312] hso: /home/richard/Desktop/hso_26-v1.12/hso.c: 1.12-Option Option Wireless
[  114.330282] hso0: Disabled Privacy Extensions
[  114.331628] usbcore: registered new interface driver hso

During the talk Ruiz mulls over the state of Internet connectivity in the world with some help from the great Gapminder data representations and reflects on how AMD technology can help to develop nations.

So what does this mean for UpStream, and why am I being so negative about the whole scheme? Well, for a start the project is interesting to us because to get more people connected you need to provide them with a means of doing so. For us this argument breaks down twofold into the terminals that a user will engage with day to day and the back-end services part of whose job is to provide an Internet connection which is sustainable financially.

We have already demonstrated swathes of research into low-power platforms, then put into place a method by which to use them in the most effective way for a mobile deployment. Equally, we also have research which is currently being considered into the concept of a Mobile ISP which would leverage some of the same hardware principles. A chip company like AMD offering alternatives in these streams is not competition, it’s just very good news as it should help us all achieve positive progress sooner.

Moving into more technical waters the negativity of my post becomes a little easier to understand. You see, Hector’s talk was back in June 2007, and the 50×15 project itself was founded in 2004. One of the major citations he makes during the presentation is that of the OLPC project which currently uses a low powered 433 MHz AMD Geode LX-700 CPU within its first generation design. Back in the technical research into AMD developed <15W platforms I slammed the Geode for providing woeful performance compared to Atom based systems but that’s perhaps not fair when we look at the context of an OLPC which is intended to fit into a ~5W average power envelope. However, the writing is on the wall for your product when the OLPC project looks towards ARM to provide the chips for their next hardware generation.

The natural question is to ask what AMD intend to do fend off the exciting ARM designs on the horizon or even compete with the slightly higher powered Poulsbo/Atom based solution we have praised. Well, for a start, the work on any further <5W Geode chips appears to have completely been abandoned post-2003, perhaps since Intel began to press home their advantage in more mainstream markets. The architecture simply has not been developed for years (arguably since Cyrix laid the foundations) and AMD have absolutely nothing waiting in the wings to offer improved performance with similar power characteristics. Now, that’s not conjecture, it’s fact. If AMD had such a part they would be producing millions of them and attempting to take on the Atom within the lucrative netbook market. Instead, they are targeting a space between netbooks and fully fledged PCs, and offering a CPU that trades blows in the much higher ~15W bracket.

Source: ExtremeTech

~15W does not compete with ~1-5W AMD - I’m sure you don’t need me to point this out. Your archaic offerings in that sector are vastly inferior compared principally to the new generation of ARM chips, and secondarily to more specialised iterations of the Atom. A simple glance at the 50×15 product portfolio bears this out without much analysis due to the slew of products aimed at purely thin-client deployments (ThinCan, Wyse), those which can’t even handle the most basic of media roles (Inveneo Computing Station, Linutop, Fic 603) or those which eschew power-consumption concerns in favour of a purely cost-cutting focus (HCL, DTK, Lenovo).

In short AMD, I think that you’re going to have to embrace the big blue or a certain British chip designer if you want to meet your aims in the best way possible, which for all your words is not something you will ever contemplate doing. If AMD are not bringing the CPUs then what’s their role or indeed business model? Because I think it leaves them bereft of one.

To finish I want to underline that when I write articles like this I do so with no bias, in the same state of mind as if I was building myself a new PC. I look at what technology is out there to fit my needs, and I look at key factors like price, performance and functionality which will ultimately sway my decision. As a fan of the underdog I would love AMD to compete in this space but they seem to have entirely relinquished their role in powering any computationally meaningful ultra low power devices. How can a company who have no products aimed at <15W TDP hope to achieve such a lofty ambition? Come back AMD, because Intel & ARM need you to keep them innovative, and Ruiz’s 50×15 initiative doesn’t deserve to slip away. Sadly, the longer AMD refrain from re-entering the market the more laughable their fringe schemes like 50×15 which depend on bleeding-edge products seem.

You see, Computer Aid are obviously came to the same sort of conclusions I did which is why they now will not accept donations of machines slower than Pentium 4 class hardware. A bold move, but one that ensures the equipment they send out is fit for purpose. However, the praise does not stop here. I actually stumbled on to their site while tripping over an email from last October my professor had sent me (Cheers Colin). What I actually came across was the brief testing ZDNet had done in conjunction with the company to look at more specialised low-cost and low-power platforms that could be used when modern desktop hardware simply didn’t seem appropriate. A quick hop over to the Computer Aid site informed me that since the date of that email back in October the charity had published their full report and it really is worth a read. It’s a brief yet concise nine page affair but what I find really brilliant are the quantitative results that they were able to attain due to their inherent connections to the regions being discussed. Not only does the paper have a close correlation to the benchmarking we did, but it also is looking at much the same hardware level questions I did during the x86 <15W platform research (Note you can skip to the last pages of those articles for the full PDF reports).

To summarise their report, the Asus eeePC was chosen as the lowest power and most feature rich platform:

“The Asus Eee PC is the overall ‘winner’ of the tests. It is the preferred solution by all partners. Despite   the   small   size   of   the   screen,   it   offers   the   best   compromise   between   power consumption, performance and portability in both Linux and Windows-equipped versions.  “

Also,  the Ncomputing X300 Windows based thin client system was highlighted as being especially suited towards lab deployments:

“The Ncomputing X300 is the preferred solution when setting up computer labs. Despite higher power consumption per each user and limited Linux compatibility, it was appreciated especially by   African   Universities   in   the   case   of   installations   not   requiring   portability.   Desktop virtualisation   can   be   a   viable   option   to   reduce   hardware   costs,   power   consumption   and required maintenance compared to the use of traditional desktop PCs.”

Now, these results excite me greatly because I wholeheartedly believe we can do EVEN BETTER by reflecting on the developments in the hardware market and combining this with our own research into the type of software deployment we favour.

Let’s take on the hardware first. According to ZDNet the eeePC used was an early 701 model which incorporated a rushed Celeron ULV processor running at 900MHz. This design was quickly phased out in favour of the 1.6GHz Atom N270 + 945GSE chipset combo designed to compete in this ‘netbook’ space. Despite the simpler in-order design, the increased clock speed, hyperthreading capabilities, and lower TDP caused the Atom solution to become the de-facto standard in the sector - and this of course is why we see several hundred designs based around these specifications. Now, these designs have been round a good long while and the market economics of the situation are very interesting if a bit long-winded (Intel being able to produce the chipset very easily and cheaply using existing R&D, old fab facilities et al). Needless to say that if you’re placed in the correct sector of the industry and have the buying power you can do even better than this woeful implementation.

You see, all Atoms are not created equally, and more to the point their chipset pairings vary dramatically. I’m not going to go into great detail (please see our research) but using a 1.6GHz Atom chip with something other than the stunted 9.3W 945GSE chipset and you’re onto a winner. A case in point: The Dell Mini 12 (now discontinued in blighty, but seemingly not in America - go figure) and now the Mini 10. Due (allegedly) to the rumours flying around that Intel did not want vendors to use the Atom Diamondville platform in anything other than <12″ lappies in order to avoid cannibalisation of their Core2 CULV market, Dell just went right on and used the Menlow Platform, intended for MIDs, which is essentially the pairing of a smaller, slightly more efficient albeit architecturally identical Atom with a chipset built from the ground up to be suitable for it - Poulsbo. It also helps them jump ahead of the competition by squeezing more run-time out of a 3 cell Li-ion battery and offers product subdivision in the case of the Mini 10 by pitching a cut-down “V” variant sporting the standard Diamondville kit. The result of Poulsbo is a complete platform drawing a mere 4.3W as opposed to the 11.8W that 99% of netbooks utilise, or the 29.5W one nettops are lumped with. Interesting stuff, especially when you consider the PowerVR influence over the GMA500 GPU in theory allows it decode 1080p H.264 in hardware.

The disadvantages to this platform? Silverthorne is damn expensive although we can’t get official figures from the Intel ARK, and the graphics drivers under Linux are a complete shambles apart from the custom 8.04 Ubuntu Builds Dell liaised with Canonical to develop. When I say expensive by the way I mean that it’s basically impossible for you and I to locate our local friendly consumer-embedded reseller and find anything using it. These boards are reserved for the target form factor (MIDs) and those organisations big enough to be able to go ahead and purchase huge volumes to make them financially viable like Dell. I did have one quote with an industrial embedded manufacturer who estimated a $500 cost per board which is frankly insanity when you consider the 1.6GHz Mini 10 is just £349. The maths don’t stack up in the cold light of day and this is hugely frustrating for consumers or people like me who like to keep an eye on the market. It’s also why we came to the conclusion that the trusty old 945GSE chipset, despite it’s failings was the best of the bunch, especially as mITX boards using it have recently reached you and I for ~£100.

This is where another company come into play - FitPC. FitPC have a single product at the moment, a low power self-titled PC running on the frankly antiquated Geode CPU (think cobwebs instead of thermal paste). Whilst this is fine for cost effective thin clients or other extremely undemanding applications it becomes a sticking point when you look into the kind of rich FOSS software implementations we have been discussing. Last week in the far east a sequel appeared to have tipped up powered by the 2W 1.6GHz Atom Z530 CPU and the 2.3W US15W chipset (Poulsbo). Sure enough, the company themselves are now advertising the device and I must say it looks like an absolute winner.

And so, to software. The proposition here is simple. The Ncomputing X300s are thin client machines which rely on proprietary Microsoft software, and are limited as all thin clients are when faced with any serious computational tasks such as the playback or rich media sources. In addition, the machines have severe range limitations (10M from the host PC according to their docs) and although fairly cheap (IRO £149) don’t scale particularly well in larger deployments. That said, the study still conveyed that technical staff were more than willing to work with such devices which really does bode well in terms of expected user experience.

Instead, my proposition is to build off the research done by enterprising coders like David Van Assche, and use the new generation of low powered hardware outlined above to perform all processing locally in the model we term CCL - Client Centric Processing. Jjust like thin clients the central kernel used is delivered using PXE and can be maintained just once but in this case it’s a much larger image designed to perform almost all tasks locally. Using this inverted approach a machine can utilise all its onboard resources whilst putting minimal pressure on a central server, whose only job is basically to send data using the NBD protocol. The huge bonus is also that the machines all benefit from having no local storage to support, and share a central point of authentication and management.

The dream therefore is a classroom (mobile or otherwise) supporting ten students and a teacher. Each student is equipped with a Poulsbo cored Atom machine which draws well under 15W including the VDU (which ironically becomes the biggest drain), and the teacher utilises the server, whose only other job is to deliver content quickly over a Gigabit Ethernet. The server itself could use the same equipment as the students, but ideally at this point we would utilise a richer set of core-logic to push data from a RAID1 array, as read speeds are really the defining burden. The result of all this is a fully functional, low maintenance, rugged, low power classroom that achieves all its aims for  the very minimum outlay possible. In addition, the seemingly positive intentions from Intel in supporting Linux and third-party IP blocks is the cherry on the top of ideas like this for research teams like ourselves.

All this and more is possible with the equipment today, and with more developments in the fertile netbook space such as the upcoming Pineview platform which places the GPU & memory controller onto the CPU die, the equipment we use to accomplish our goals can only become faster, cheaper, and lower in terms of electrical footprint.

I’ll cover Pineview (and Pinetrail - don’t ask) in the coming weeks for those interested. At the moment we’re simply working with conjecture as far as those platforms are concerned so it’s for the best if we wait for more solid details to leak. Till then there’s always l’inq, and good old Anand.

Whilst taking a stroll around the tech sites of the web I noticed Ryan Shrout at PCPer has got his hands on one of the new FitPC2 devices. You might want to check out the preview here.

You see, the premise is simple. When one compresses a bit-perfect original master of their favourite song down into mp3, they are using a lossy format and of course reaping the benefits of this. The user can then walk around with a ~5MB copy of their favourite song that’s indistinguishable against the original recording to all but the most ardent audiophile. Conversely they can choose a lossless codec that reproduces the original without any degradation whatsoever albeit at the cost of a >40MB file size. Well, the same is true with images. And where do we use images on our computer where file size may become a huge problem? That’s right - the Internet. Just as audio compression is a complex mathematical field, the same is true with its visual counterpart. We’re not going to attempt to delve into the science behind that because it’s not my field and I’m sure you can find umpteen sources to explain it better than I. However, what we will do is list a couple of further examples other than Lifehacker and offer up the original imagery for you to make up your mind.

First up - The official website of Leeds Metropolitan University. I happened to be browsing at postgraduate courses and noticed the lead image taking a second or so to load which (as pedantic as it sounds) is noticeable to a frequent web surfer.

Image saved directly off the site (please click for original full resolution copy - caption is not indicative of the statement I am making)

Clearly a pretty hefty 940px x 233px image weighing in at a grand total of 204.97 kB. Surely we can get that down without sacrificing much quality? Well, I fired up Irfanview, and in particular the Radical Image Optimisation Tool (RIOT) plugin then used the save for web option to produce a lossy JPG set to a quality level of 75%. You may find it easier to just use RIOT standalone but I like having Irfanview to do basic image editing if I need to change one or two things at this point. The settings you use here will vary wildly from image to image, and there are formats which specifically work better for less complex or images constructed in a certain way. For example, if my picture was a block of two colours arranged in a pattern I might choose to use PNG to get a more efficient end product. Again, delving into the logic here gets complex, but I would really encourage anyone in web design to read up. Some of the best articles can be found in the Yahoo developers section.

Using RIOT through Irfanview

Great stuff. Apparently without altering the dimensions of the image RIOT has managed to crush the file size down 174.3 kB to 30.67 kB - a brilliant effort and one which represents an 85% saving.

I then upload the image to this site through our WordPress frontend, at which point another piece of magic happens. You see, we run a small plugin called smush.it, designed by Stoyan Stefanov and Nicole Sullivan. Smush.it is a lossless image compression utility which will further strip the file size using techniques not designed to ever degrade quality (i.e. removing supplementary metadata). Because we’ve already optimised the image, smush.it has no success on this occasion, although it does achieve a respectable 5.66% reduction of the original image. I should point out at this point that the original image you see above was uploaded via a different method and has not been touched by the ’smusher’.

Checking what smush.it can do for us

Well, the proof is in the pudding, so please feel free to sample the difference. Again, please click the thumbnail to open up the full size version.

The Image after both RIOT & Smush.it

Probably the best way to compare the two is to right click the full size copy of each and save to your desktop. Then use your favourite image browsing tool (you should have a basic one in your OS) to flick rapidly back and forth between the two. I have to say I notice a very minor change in small sections of the skin tone, but really nothing whatsoever that would stop me using it on a site. In fact, If I used an image like that on this site in context I would probably push it down to perhaps the 50% quality zone, at which point the image becomes only 21.14 kB.

So what can you do? Well, it depends at what end of the spectrum you reside. If you’re a web designer you SHOULD know this stuff already. If not, learn… fast (for all our sakes). If you’re an end user you need to establish why you want to do something about it. At the end of the day, if you are connected to the world through an un-metered 10MB pipe there’s little drive to do anything about this minor travesty. Leave that to the people around you responsible for the content - you’re the market that has caused this slack practice but that’s not to say it’s your fault. However, perhaps you’re heavily metered (a ‘generous’ 1GB plan), use pay as you go, or simply have a slow link. Just because you’re not willing to operate your own gateway using RabbIT doesn’t mean you can’t slim down the footprint your web surfing has. It was pointed out to me that at least T-Mobile & Vodafone in this country operate server side software in a similar vein to RabbIT (see below) so you can perhaps leverage the client side configuration to encourage the proxy to be more vigorous with its compression tendencies. If this isn’t an option then by all means look into Opera Turbo.

A quick note to those of you covered by a wonderful HSDPA blanket and wanting to disable the in-line image compression on your links. You can either locate your operators software like T-Mobile’’s Web n Walk accelerator frontend or send explicit instructions to the proxy by modifying your HTTP headers.

On a more serious note - I presume Leeds Met have no problem with me using this imagery. With it being in the public domain I would expect there to be no backlash! Equally, I hope I’ve made it abundantly clear that this isn’t an attack on my university in any way, I simply happened to stumble across a good example to illustrate my point. To further underline this I hopped over to the Loughborough University site and managed to get one of their 70.63 kB images down to 19.72 kB using the simple RIOT lossy JPG compressions procedure outlined above

Update: I’ve just noticed that LMU have in fact resampled most of their imagery to between 50-140kB, which is much more reasonable. In fact, the image I dissected above in particular is now only 50.56 kB.

1.    You should first expand your virtual hard drive to at least 16GB to accommodate the new chroot.

2.    Open a terminal and run
sudo apt-get install portmap nfs-kernel-server nfs-common

3.    After installation has concluded type the following into the terminal:
echo “/home SERVERIP/NETMASK(rw,no_root_squash,async)” | sudo tee -a /etc/exports /dev/null
Take care to replace SERVERIP/NETMASK with the appropriate values. In my tutorials I’ve been using a server IP of 192.168.1.151 with a netmask of 255.255.255.0.

4.    One last command here – run:
sudo exportfs –a

5.    It’s vital that you’ve performed the above correctly. Failure to do so will cause the home filesystems not to be mounted when using your fat client, which will cause you not to be able to get to the desktop at all!

6.    Now download the fat client script from Nubae or from our mirror. I’ll presume you’re saving it to the desktop for the purposes of this guide. If like me you are using a single NIC setup you’ll want to edit this file to alter what it perceives to be which interface. I have included the excerpt below to help convey what I changed. You’ll notice we’re telling it that eth0 is actually eth0, and then asking it to use the IP address of that interface (the server IP) to mount the network file system. I must admit, I completely missed that the first time around, and was met by the system that booted but that wouldn’t mount the home directories, simply because I didn’t have an eth1, thus no IP was copied into nfsmounts.sh

eth0=$1
shift
eth1=$@
### Add nfs startup script for lts.conf
cat <<EOF >> /opt/ltsp/$CHROOT/etc/nfsmounts.sh
# nfs starts before portmap so mounts fail, so we restart portmap here
/etc/init.d/portmap restart
mount -t nfs $eth1:/home /home

7.    Place this edited file inside
/usr/share/ltsp/plugins/ltsp-build-client/Ubuntu/030-fatclient

This folder requires root permissions to manage, so it’s best achieved through the terminal by running sudo cp ./Desktop/030-fatclient /usr/share/ltsp/plugins/ltsp-build-client/Ubuntu

8.    A quick note here. When I installed the scripts at my most recent attempt (for this tutorial) the mozilla-acroread and acroread packages were broken. You can simply replace these with another open source PDF reader like FoxPDF..

9.    With the file copied, it is ready to be executed. We’ll be preparing a high-fat client here, so again at the terminal we execute sudo ltsp-build-client –chroot name-of-chroot –fatclient Ubuntu –highfat

Again, ensure you change the option that requires user input, in this case, the name of the chroot is italicised. The regular thin client chroot is simply named i386 so it would seem logical to prefix that with ‘fat’ or similar.

10.    If you’ve used the correct syntax you will see the terminal start to output lines signalling that software packages are being pulled from various repositories. This could take quite a while depending on your Internet connection, so put your feet up and relax!

11.    The next step is telling the DCHP server we worked with earlier in the documentation which machines need to be delivered the new chroot. To do this, we re-enter the dhcpd.conf file, accessible by running the following command from your terminal
sudo gedit /etc/ltsp/dhcpd.conf

12.    I’ve included my dhcpd.conf file below for reference. Remember that it’s setup for a single NIC environment and in this case a single thin client & two fat ones.

#
# Default LTSP dhcpd.conf config file.
#

authoritative;
subnet 192.168.1.0 netmask 255.255.255.0 {
option domain-name “edubuntu.localhost”;
option domain-name-servers 192.168.1.1;
option broadcast-address 192.168.1.255;
option routers 192.168.1.1;
#    next-server 192.168.1.254;
option subnet-mask 255.255.255.0;
pool {
allow unknown-clients;
range 192.168.1.160 192.168.1.200;
}
}

# regular thin clients
group {
deny unknown-clients;
get-lease-hostnames true; ### make sure you’re /etc/hosts contains all the hostnames of the clients
if substring( option vendor-class-identifier , 0 , 9 ) = “PXEClient” {
filename “/ltsp/i386/pxelinux.0″;
} else {
filename “/ltsp/i386/nbi.img”;
}
option root-path “/opt/ltsp/i386/”;
host thinclient1 {
hardware ethernet  00:FF:1E:00:00:00;
fixed-address 192.168.1.161;
}
}

# fatclients with highfat
group {
deny unknown-clients;
get-lease-hostnames true; ### make sure you’re /etc/hosts contains all the hostnames of the clients
if substring( option vendor-class-identifier , 0 , 9 ) = “PXEClient” {
filename “/ltsp/fati386/pxelinux.0″;
}
else {
filename “/ltsp/fati386/nbi.img”;
}
option root-path “/opt/ltsp/fati386/”;
host fatclient1 {
hardware ethernet  00:15:C5:00:00:00;
fixed-address 192.168.1.171;
}
host fatclient2 {
hardware ethernet  00:15:C5:00:00:00;
fixed-address 192.168.1.172;
}
}

13.    If you have made changes, remember to save and restart the DHCP server by issuing sudo invoke-rc.d dhcp3-server restart or sudo /etc/init.d/dhcp3-server start

14.    The final step is to alter your hosts file to match hostnames to IPs. The markup in Nubae’s sample DHCP script (above) prompts you to so this, but it’s easily forgotten. Use sudo gedit /etc/hosts to open the text file in question and use an identical syntax to the localhost entry; that is:
IP    HOSTNAME

15.    It was at this point that I booted a prospective fat client and finally got up and running. Just a couple of things were left to irk me.

a)    There was no sound
b)    Flash based videos weren’t playing. This is to say, flash was rendering the window replete with controls and progress bar, but not the content.

16.    To fix the sound issues, we need to ensure pulseaudio is using the local machine as its server (it’s running locally now remember). We use nano because it’s a text editor that doesn’t need a window manager.
nano /etc/pulse/client.conf
Add these lines:
default-driver=pulse
default-server=127.0.0.1

17.    To fix the flash issues, I first attempted to copy update the sources list on our server, and copy that over to the chroot in question. Then we enter the fati386 image, purge the flash package, clean up all sight that it ever existed, and then update our sources from the list we just copied. Armed with updated repositories we should be able to issue a simple install command for flash 10 to be added successfully. By the way, remember that all these commands are issued on the server. Also, for those wondering, the reason we’re so exhaustive with these steps is to ensure the install goes off without a hitch; because there were a catalogue of errors in older sources lists that caused the process to fail.
sudo cp /etc/apt/sources.list /opt/ltsp/fati386/etc/apt/
sudo chroot /opt/ltsp/fati386
sudo apt-get clean
sudo apt-get autoclean
sudo apt-get update
sudo apt-get –y –purge remove flashplugin-nonfree
wget
sudo apt-get install flashplugin-nonfree

However, it should be noted that I did not manage to get flash successfully working. By reinstalling flash I even managed to make firefox lock up completely every time it visited flash based sites.  I attempted downloading the .deb straight from Adobe and even compiling from source but still came up against the same problems. Obviously this is far from acceptable, and I will revisit thie segment once I have found a solution!

18.    At this point we can exit the chroot environment by issuing exit or hitting Control + D

19.    Lastly, we need to update the squashfs image that is eventually delivered to our fat-clients, otherwise none of these changed will make it onto our machines (which will need rebooting).
sudo ltsp-update-image -a fati386

20.    And there you have it. Fat-clients with local processing. As I have said, once I find a fix to the flash woes, I’ll amend this piece.

[[Visit blog to check out this spoiler]]

You could also use synaptic or the GUI frontend of apt-get for this, in case you’re not confident with the terminal.

Other common configuration steps

• Installing Macromedia Flash (note, from 7.04 the second step should not be needed)
• Installing x11vnc to allow screen-sharing in the thin client manager
• More features of the thin client manager
• Customising Thin Client behaviour
• General Maintenance of the LTSP server
• Installing Sugar (The OLPC XO Window Manager)
• Other FAQs & Tricks

Further sources of documentation:

• The Edubuntu Handbook

• The Ubuntu LTSP Wiki

• The LTSP Manual

• The Gentoo LTSP Guide

With enterprise level virtualisation now within the hands of any user who cares to take the effort to download and install it, there has never been a better time to experiment with operating systems in general, and in particular Edubuntu.

Perhaps you have some older machines in your house, or just want to experiment with the great possibilities available due to the Linux Terminal Server Project. Either way, I have compiled a guide to installing the Edubuntu 8.10 LTSP Server, replete with images. It takes you from the end of the last tutorial straight through to getting a working thin client, and after that, you’re more than welcome to revert to the aforementioned community documentation for more extensive configuration.

We’ll pick back up at the point of installing a netbooted fat client.

By this point you should have a functioning ESXi server. If you’re experiencing problems, then the main thing you’ll want to ensure is hardware compatibility. With ESXi being more of an enterprise product, built from a stripped back kernel, you will find it does not have support for many pieces of consumer hardware. That said, I am using an old Dell Optiplex GX20 as my demo rig, and have had no issues with it thus far.