We'll have a shipping update down the page a bit but first I want to tell you about Fallon's project for his school Science Fair. Fallon is just finishing fourth grade at Austin Creek Elementary School here in Santa Rosa, CA. Next year he'll go to a new school so this was Fallon's last Science Fair for awhile (the new school doesn't have one).

The question Fallon wanted to ask was when you are playing Minecraft against your friends and enemies in Survival Mode is it better to invest in a faster game PC or a Mineserver? Which purchase will give the greatest advantage if you are playing locally?

So we did an experiment with identical Mineservers here at the house in California and at Clinton Collins's place in Dripping Springs, TX, 1,781 miles away. Note the two Mineserver cases glued to Fallon's trifold presentation board. The client machines we tested were: 1) Cole's 5 GHz 8-core gaming machine with two GPUs and 16 gigs of DDR3 and; 2) the crappy Acer notebook we use to run the laser cutter with a two-core 1.4 GHz Celeron processor, Intel integrated HD graphics and four gigs of RAM. Using the Valley benchmark Cole's machine scored 2240 (53.5 fps) and the notebook's benchmark was 279 (6.7 fps). Just for reference Cole's PC cost around $2000 for him to build while the notebook came from Amazon for $300.

In the local test the Mineserver was only two hops from the players with an average ping of just under 10 milliseconds for both machines. The ping would have been faster but we've been doing all local testing over WiFi to make sure that interface works well. The Mineserver in Texas was 20 hops away with an average ping time of 125 milliseconds. While Cole's PC was obviously faster, one interesting result was that each machine was able to maintain the same frame rate on either server. Cole could do 60 fps (the Minecraft max) in California or Texas while the notebook could do 30 fps on either server. 

Where the distance really came into effect was in that ping score and it showed in game play. Server proximity is a great leveler, which is why Fallon called his project Location, location, location. Playing locally gave the $300 notebook a huge advantage, so if you want to smoosh your opponents it helps a lot to run the game on your server.

Let's be clear here: Mineservers are plenty fast. Even on the Texas server the game was totally playable from California. But it is still way better to be playing two hops away than 20.   

Even against players in the same town there was a significant difference. Running a traceroute to our Mineserver bank down at Sonic.net we discovered that our signals were bouncing all over the Bay Area (an average 25 millisecond ping and 10 hops total going as far south as San Jose!) just to get the eight miles to the data center in town. 

If your town is anything like our town, then, it's a real advantage to be running the server at your house.

Now to our beta test results from Sonic.net. We tried everything we could and were never able to solve the transient Ethernet bug described in an earlier update. We even waited for an Ubuntu upgrade hoping that would solve the problem. Our board vendor in Asia kinda/sorta acknowledges that there is an Ubuntu-specific problem interfering with his Ethernet driver, but he has yet to come up with a solution.

The pernicious bug sometimes doesn't allow the Ethernet interface to boot at all and sometimes it boots but then at some later point simply stops. This may well be two bugs, not one, but a solution doesn't seem to be in sight so we did the only thing we could think of which was to abandon Ubuntu Server and go back to Arch Linux. We probably never should have left Arch, but Ciinton is something of an Ubuntu Whisperer so we thought that switch would save time, not lose it.

All six test Mineservers have been running Arch for three weeks now without a hiccup. Solid as a rock. So now we're finalizing our custom Arch distro, packaging it with a Mumble voice chat server, and making sure the WiFi interface works right. Just a few more days and all that will be ready.

But there remains one final problem: AMP, our server management utility that runs on a desktop or mobile, doesn't yet support Cuberite, our preferred Minecraft server.

Vanilla and Spigot, our other two servers, are written in Java and have been supported by AMP from the beginning. But Cuberite is different, written in C++, and is screaming fast as a result. We love Cuberite, which is a better server in every regard, running faster and supporting more users than the others. Spigot is fine and can achieve our performance goals. We had 10 players running on a Spigot Mineserver the other day with no problem at all. But Cuberite is even better.

We can ship with Spigot or wait for Cuberite, which is promised for the next full release of AMP. We were inclned to go with Spigot and try slipstreaming the Cuberite upgrade later, but Mike, the AMP developer at Cube Coders in England, said just tonight that his next AMP release will be coming in "early June."

So we've decided to wait for that release. Once we have it and have done 2-3 days of hard testing, we'll finally start shipping Mineservers. That looks to happen around the middle of June.

We're sorry it has taken so long but the product will be much more powerful and reliable as a result.

And in case it wasn't obvious, the result of Fallon's science project was a clear win for putting $99 into a Mineserver instead of that $100 gaming mouse that keeps you awake at night flashing on your computer desk.