For my system, I have updated it a little since the previous post. It now has a TV tuner card, 2GB of DDR2-6400 memory, and is clocked at 2.76GHz, a bit higher than previously. I have also scored a UPS, which the PC, modem, router and the backup server are now plugged into.

Some updated readings:

My Core 2 Duo PC:

On, running folding@home, screen on: 120W
On, idle, screen on: ~110W

These readings are really just estimates, as the power meter’s readings just keep fluctuating anywhere from 104 to 129W. It is also clouded by the fact that the UPS is plugged into the power meter, playing havoc with the results. when the PC is off, it can still show a power draw of around 70W, with just a modem, router and a standby server running off it.

Athlon System:

On, folding@home running: 150-160W
On, idle: 148W
On, idle, screen off: 130W

Core 2 Quad Q9300 PC, nvidia 7200GS, all at stock speed:

On, folding@home running: 130W
On, folding@home running, screen off: 106W
On, idle, screen off: 64W

You can see what a difference a few years and processor generations makes. I also recently purchased an nvidia 8600GT graphics card for the Quad box, to boost its folding output with the GPU2 client. I chose a relatively low-power drawing card, as I am conscious of how much juice it uses.

Core 2 Quad Q9300, nvidia 8600GT, all at stock speed:

On, folding@home SMP (CPU) client only running: 133W
On, folding@home SMP and GPU2 clients running: 143-151W

I was pleasantly surprised. The faster video card, running the folding client, consumed only around 20 Watts more power than an idle 7200GS. Quite a good result for an extra 1400 points per day in folding production, in terms of performance per watt.

While I was in the mood, I though I’d re-test the home theatre setup. The only difference here was that the video player has been unplugged, since it is no longer used.

Home Theatre:

All on standby: 16W
All on standby, HD Set Top Box on: 31W
TV and STB on: 146W*
TV,STB, receiver on: 120W

* The figure for the TV and STB on was curious. since there was no signal to the TV, it was displaying the blue background and power usage went up. Once a signal went to it from the receiver, usage dropped.

Ad departments. Gotta love them. Somebody probably got a bonus for coming up with that campaign, too.

If you feel like joining, you are more than welcome. Just enter Team 24 when you are setting it up!

Aaaanyway, after updating my system earlier in the year, I noticed my points production had dropped off. What had happened, is that the CPU was not running at full speed: it was throttled back to its slower idle speed. A little fishing around the web turned up this forum thread at techreport.com. It turns out that the CPU speedstep function on was on, with the multiplier dropped back to 6, not 7 (in my case with an intel E6300). The operating system’s frequency scaling control was not treating the folding client as an ‘important’ enough process - it was ignoring the ‘nice’ processes by default. To find what speed the system is running at, open a terminal prompt and enter:

cat /proc/cpuinfo

A bunch of info will come up - look for the cpu MHz field and see if it is the speed it should be. The other way is to install the CPU frequency scaling toolbar applet in Gnome - I have done this and can see at a glance if the cores are running at full speed or not.

To let the frequency governor include ‘nice’ processes, a flag has to be set, as the root user:

sudo su
echo 0 > /sys/devices/system/cpu/cpu0/cpufreq/ondemand/ignore_nice_load
exit

For a dual- or quad-core machine, you may have to do this for all cpus - cpu0, cpu1, cpu2 and so on. An example for the second core:

echo 0 > /sys/devices/system/cpu/cpu1/cpufreq/ondemand/ignore_nice_load

Note that I only had to set it for the first CPU (cpu0) and the other one changed as well.

To make this change permanent, instead of having to re-enter the above commands after each boot, there are two methods, only one of which worked for me. The first method, which did not work, is to add the echo 0 > … line to the /etc/rc.local file, before the line “exit 0″. This did not work for me, as I found out after rebooting.

Instead, I used the second method, found in this thread at the ubuntu forums, involving changing a setting in the Gnome Configuration editor (found under the Applications -> System Tools menu):

Open gconf-editor, go to /apps/gnome-power-manager/cpufreq
and check box consider_nice.
Takes effect immediately.

I believe that the setting in gconf-editor was overriding the setting in rc.local, so it never took effect. After changing that, my system has been happily folding away ever since.

Turns out that most of the problems stemmed from that. In a thread in the ubuntu forums, it is stated that the current flash release is still pretty buggy. Kilz from the ubuntu forums has created a script that installs the plugin with the nspluginwrapper, for running in 64-bit installations. The thread in question is here. I tried it, first uninstalling the current plugin using the synaptic package manager, then ran the script. I now have the 9.0 r48 version, and it everything is running fine again.

Damn Adobe Flash - the majority of my linux usability problems have centred around getting the damn plugin to work in browsers at all, then to actually work properly. I can’t wait for the Gnash project to have a viable replacement up and running so I can dump Adobe’s product for good.

Game show (or is it just a comedy sketch?) set in a library:

Spa Prank

Treasure hunt…

This operating system was GEOS, standing for Graphical Environment Operating System. An excellent article about it is here. I felt like having a bit of a tinker with it, relive some old memories. One problem though was I no longer had a C64 to load it on. However, emulators came to the rescue. For linux, I installed VICE, which is in the ubuntu repositories. To make it work though, you have to download the version from viceteam.org, that contains the ROM files that are not included in the ubuntu version. There are two ways of putting the ROM files in.

First, the way I tried, was to unzip the version 1.22 file from viceteam, and copy the ROM files into the installation directories (in ubuntu, it is /usr/lib/vice/). The ROM files are the ones without file extensions that live in the data directory in the zip file. They are named ‘kernal’, ‘basic’, ‘chargen’, and so on, and are in different subdirectories for different commodore models the program emulates. Once it is installed and the ROM files are in the right spots, you can start the C64 emulator by typing ‘x64′ at the command prompt. If all goes well, you will see the C64 screen come up in a window.

The second way, which I did not try, is to follow the instructions here for compiling the latest version of VICE.

Once you have the emulator running, you can download a copy of GEOS. Put the *.D64 files in a directory somewhere, and using VICE, choose the File -> Attach a disk image -> Unit #8 menu. Point it to the GEOS64.D64 file, and it should start loading. You can then marvel at the mid-1980’s computing experience. It is similar to the first Mac desktop, but was available for a fraction of the price.

There are other productive things you can do with the Commodore 64 emulator, too…

It was always my thought that driver training would be the answer - I attended a driver training course a number of years ago, and it was an excellent learning experience. In fact, I still highly recommend anyone to attend one of these. But I found a link to a report (in pdf format, briefer version here) prepared for the RACV that noted that it may not have much effect. Its author found that what was learned during the training would drop off in time - people would forget most of what they were taught. It also noted the ‘low individual crash risk’ of drivers - I believe it refers to the actual chances of being in a position, where a crash is imminent, being so remote for the average driver. The other factor, widely expressed, is that the increased level of skill following training would lead to an increase in confidence of the driver, therefore encouraging more risk-taking behaviour. An example of this is “Hey, I learned how to recover from a slide, lets try one around this corner.”

This latter opinion is what concerned me - surely having knowledge of how to control a car is better than none? I wouldn’t get into a plane whose pilots were not told how to recover from a stall, for instance, but rather told “Just don’t get into one in the first place.” A vague analogy, but hopefully you follow my thoughts.

I still believe there is a place for advanced, or defensive, driver training. Not the sort of higher level stuff that was mentioned in the above paragraph, with stuff taught on skidpans and the like, but just the basic stuff like what I was taught, like how to brake in an emergency, to not look at what you are about to hit, but rather where you want to go to avoid the obstacle. To check your tyre pressures regularly. To look further down the road for any dangers, rather than the bumper of the car in front.

Experience is the main thing the report suggested, particularly supervised training while a learner driver. This I can understand. But some of that training should also include just some basic emergency skills, like trying out the ABS brakes on a quiet wet road, to know what it feels like when they activate. The report also suggested a graduated licence scheme, with further tests to be passed to get less and less restricted licenses. I agree with this. But the main suggestion of the report, in my opinion, is buying more crashworthy cars - cars that can better survive an accident. Even better is a car that can avoid an accident with better active safety.

I’m finally getting around to my main point. Active safety in cars, such as better brakes, ABS, traction control, and in particular, Electronic Stability Control. My next car will definitely be fitted with ESC. Basically, the common factor in crashes is overwhelmingly human error. The more ways you can take human fallibility out of the equation, the safer the roads will become.

If you take that last point to the extreme, the only ways you would end up with zero road deaths would be to limit them to 20km/h, or remove humans from the control of cars altogether. Have each car computer controlled, with sensors taking in information about the surroundings, lane markings, radars picking up obstacles to the front and side. Communication between other vehicles, using a common standard, that lets each car know what others in its vicinity are doing, so they don’t hit.

Even then, there are still factors that will cause possible deaths. Mechanical failure, or errors in programming; after all, the human factor is still in the design and construction stages. Pedestrians, particularly drunk ones, or suicidal ones, will still find a way to jump in front of a car.

In the end, is this a situation we want? It would be a massive cost to replace all vehicles on the road with automatic travel pod-type things (I don’t know if you could call them cars). There would be massive opposition. I happen to like driving - so do millions of others. People would not trust these automatic drivers, even if they could be feasibly developed - there is no chance at today’s level of technology. I guess the moral of the story is, you make something foolproof, they will just make better fools…

http://bort.blogsome.com/crossing-the-nullarbor-index/

Today I bought a Mains Power Meter, model MS6115 from Jaycar Electronics. They finally arrived in stock - I’d been waiting for one of these since before Christmas of 2006. The principle of the thing is that it plugs into the power outlet, and you then plug an appliance (or appliances, if you use a powerboard) into the socket on the meter. Dan’s Data has reviewed one of these just recently - I suggest you have a look there for more detail than I will be able to provide.

I started testing a few things around the house to see how much juice they draw, and got a few surprises. First of all, the voltage it read was 290V - from a 230ish volt outlet. This was a bit curious. Plugging in a kettle and starting it off then gave me a maximum reading of 2984 Watts. Hmm. Switching over to the Amp reading screen showed it was pulling just over 10 Amps, which makes more sense.

Next to try was the home theatre setup, consisting of a 76cm widescreen CRT TV, high Definition set top box, dvd player, video recorder, receiver and 5 speakers plus a subwoofer (more info about the setup can be read in an earlier post). The results are as follows:

Home Theatre Power Consumption
All items on standby: 31 Watts
All items on standby, except set top box on: 43W
TV and set top box on: 172W
TV, set top box and receiver on: 197W

Wife’s PC: Athlon XP 2500+, running at 9x200MHz (1800MHz), Abit NF7-S, Radeon 9700, 19″ Samsung 930BF

PC and screen in standby: 37W
PC and screen on, idle at Windows desktop: 168W
As above, folding@home client running: 178W
As above, folding@home and 3DMark2001SE demo running at 1024x768, speakers on: 181W

Server: Pentium 233MMX
On, idle: 40W

My PC: Core 2 Duo E6300 @ 7x350 (2450MHz), Abit AB9, GeForce 7300GT, 19″ Viewsonic VX912

PC standby, server off, router (Netgear RP614) & Modem (Standard Optus Cable) on: 22W
PC on and idle, server on: 99W
PC on, idle, screen off, server on: 82W
All on, Folding@Home running: Max 99W (0.34A)
All on, folding, mythtv watching live TV, stereo on: 96W

That was a shock to see how much more power the older system used than my new dual core box, considering how much more powerful the Core 2 Duo is. It really shows the progress made in making new systems more energy efficient. I think the time may have come to upgrade the Athlon box…