Back in April when I was a fool and still figuring out how this whole Bitcoin economy worked (hey I might still be a fool, but today’s price at ~$830 / 1BTC is just crazy right?), I put in a pre-order for a BFL Labs Jalapeno 7GH/s miner. Supposedly delivered in Two Weeks ™.
Later I found out the company was full of shit and people are trying to get refunds having ordered theirs in 2012 and still not delivered. I put the purchase down to n00b stupidity and moved on. Well today my BFL miner finally showed up.
Back in May when I put together my mining rig with a bunch of GPUs, it mined at ~2000MH/s and used about 700W of power. The new ASICs are much faster and much more efficient than GPU miners, but I didn’t have funds to drop on a big ASIC and I wanted general computing power of a GPU for other projects. I’ve since shut down my measly 2GH/s miners as being unprofitable.
With the mining network now on the order of 5 PH/s (yep, PETA hashes per second), the tiny BFL Jalapeno is a drop in the ocean and would earn … Continue reading
As I woke up this morning, I found that my messing with IP policy routing on Linux had actually worked. There must have been a cache somewhere that I hadn’t flushed because last night I was still around 800ms ping times. This morning… 380ms.
But I’m getting ahead of myself…
Set up a backup 3G connection to complement my Satellite connection. 3G has lower latency than Satellite but less data allowance (and bandwidth in my location far from the tower). Move specific streams such as ICMP, DNS, SSH, IPSec, VoIP traffic to use the 3G connection.
I might not live in the middle of nowhere, but you can see it from here. There’s no mobile phone coverage, which means I am eligible for NBNCo Satellite Internet. However.. Satellite may have an OK bandwidth (6Mbps/1Mbps)* it has horrible latency, with minimum ping times around 650ms due to the speed of light (signal has to go up.. and back down, and there’s some packet overhead too). VoIP was like talking to another planet.. maybe a small moon with the lag. Multiplayer gaming.. forget it. SSH to administer remote machines was slow work. Continue reading
Now that uni exams are over, I’ve had time to work on some of the unfinished projects that have been piling up around the place. Today I made some progress finishing the tabletop CNC router. The unit is based on the Shapeoko for the mechanics, but with my own electronics sourced from EBay using the rather common TB6560 stepper driver controllers. The Shapeoko project had their own electronics based on an Arduino, so this was a little more challenging than just following build instructions.
Today I fixed up the nagging electrical issues (ie swapped cable pairs!) and parallel port pinouts, to get the unit going. The CNC is connected to an old Celeron PC running LinuxCNC which is why the CNC is printing the default print of the LinuxCNC logo. According to the Internet, there are many variants of the TB6560 boards which is why they are such a pain to configure, but a slow methodical process (trial and error) finally got it going.
Looks like the unit needs some more calibration rather than just rough eyeballing, as well as adding some home limit switches so it can home itself rather than me resetting it manually … Continue reading
One of the first Kickstarter projects I backed was the KickSat picosat project. Well it is getting closer to launch date so last week an email was sent out asking for our code. I was interstate at a conference at the time and did not have my code handy, and I’ve been busy with my other projects, so my code is pretty boring.
I wasn’t sure how much I could transmit, but I thought it was a good idea to include at least one byte for a frame check sequence, ie CRC-8. I also wanted to send out some static code, so I chose part of my amateur callsign, which is VK7LXX, and so prepended ‘LXX’ to the start of each transmit message.
There were two sensors on the picosat – a magnetic field sensor and a gyro measuring Angular velocity. To cut down on the amount of transmitted bytes, I chose only to use the magnetometer, which provides 3 floating point values for field strength in X, Y, and Z directions. These are 4 byte floats, so I packed them in, and my transmit frame is 16 bytes long in total. Here’s the code:
/* AlkaloidSprite - Sprite Kicksat … Continue reading
So I decided to enter a programming competition to flex my coding muscles. This weekend it was GovHack 2013. The first time the even had come to Tasmania. I teamed up with a bunch of others who we had not met before and came up with a simple mapping app that overlaid environmental data, allowing users to add problems such as damaged stormwater pipes, or toxic discharges, or tag any photo to the site.
The project website is at http://tiger.scriptforge.org which has links to the project source, blog and demo. I ran up an instance on the Nectar Research cloud, so if I need another instance I will still archive the site permanently on my Linode VPS servers.
I got my 3D Printer a week before I left Sydney to move to Tasmania, and in the past 18 months life has gotten in the way of sitting down and fixing nagging troubles with my build. Most of the prints were warped or fails due to shortcomings with the design, or my build. Last week, with the help of my local DIY friends who came over to tinker with my toys, we tackled each problem one by one:
- I had a jar of ‘ABS Juice’ – adhesive slurry of ABS plastic and acetone. Painting this on the build platform made the models stick such that I could even print raftless.
- New versions of ReplicatorG and Skeinforge started producing accelerated GCODE but my printer was running old firmware. We had to open up the electronics to get to the extruder controller
- The extruder would often become blocked more often than not when changing filament colours.
- Many of my failures were using this roll of blue filament that is not up to spec – probably too thin – for the stepper to grab at the … Continue reading
So there seems to be a bunch of primates going bananas about these emerging cryptographic, Peer-to-Peer digital currencies. The most widely used one is Bitcoin. I’m not going to go into an “introduction to bitcoin” or “mining for BTC” type post as there are plenty of those out there. Go Google and find out about it. These are just some thoughts I’ve had. I’m not an expert in currency trading, economics or anything financial – my expertise is in technology.
Trading (and buying Bitcoin)
The price of bitcoin has been rising steadily, but also fluctuating wildly, due in part by speculation, but also by panic selling and some DDOS attacks by market manipulators trying to look for short term profit. The main exchange for bitcoin is MtGox which is based in Japan, but there are also others worldwide that deal with a variety of currencies. Being stuck in Australia did limit the options somewhat. MtGox does now accept funding from Australian banks and also withdrawing to them, but it took over two weeks for them to verify my identification so I could start to trade. Since that time the value of BTC essentially doubled, but I took … Continue reading
So this server was upgraded from RHEL 6.3 to 6.4 via yum upgrade, rebooted and it hung during the bootup with:
task: swapper 1 blocked for 120 seconds
I switched off quiet booting so I could get some other errors and the last thing the kernel did was
switching clocksource to tsc
After wasting my time with trying a bunch of kernel boot parameters to change the clocksource, I left the terminal and came back 10 minutes later to find a kernel panic stack trace:
So it looked like the IPMI driver was causing the problem. I couldn’t boot to an earlier kernel since Redcrap seemed to uninstall old kernels leaving them not available to me unlike FREE AND OPEN SOURCE SYSTEMS such as Debian. However, booting from the RHEL 6.3 boot DVD still worked, so it looked like a problem that occured at the last kernel update. Since RH also put their binary packages behind the Redhat Network subscription paywall, it was not easy to find a link to download an earlier kernel package. In the end, I booted into rescue mode with the install DVD, did a chroot /mnt/sysimage … Continue reading
For those who wanted my slides for today’s talks at the Hobart TasLUG meeting, here they are in all their Google Docs (*) glory:
(*) – until Google decides to arbitrarily shut down Google Docs.
So as you can tell from my previous post, I’ve been tinkering with this KickSat development kit. The Arduino-like code is pretty trivial to write, so most of my effort will actually be to put together my ground station to receive the signals when the satellite is launched.
So right now I have the Sprite (what each individual pico-sat is called) transmitting the magnetometer sensor reads every half-second. Since I don’t carry my USRP with me I also have a FunCube Pro dongle. And since I could not get the serial monitor to work properly on Linux, I thought I would test the Funcube receiving my packets.
So it looks like it is working! Those red sections in the 3D plot are incoming packets. Now to bring out the Python to put together my groundstation software… Stay tuned! (*groan – radio pun*)