My next computer … Mac mini?

I’ve almost always had a powerful desktop machine at home. It started with gaming. By necessity, you needed a powerful machine. I typically was building my own machines so I could focus on getting the biggest bang for the buck. Once I got back into the Mac, I started with a PowerMac G4 Dual 1GHz. At the time, this was probably the middle of the PowerMac machines. I upgraded that to a PowerMac G5 Dual 2.7 a few years later.

I began to augment my main machine with a laptop starting with a Powerbook G4 15″ (which I still have). I had Windows laptops from work that were always pieces of garbage. The Powerbook was the first laptop that started to shake my desktop bias. Then I got a MacBook Pro last November. That’s when I started to really think seriously about going laptop all the time.

Earlier this year, I found a buyer for my G5. I decided to sell it for a couple reasons. The first was that the Intel transition was changing the Mac landscape much faster than I think anyone thought it would. Add in virtualization and the PowerPC machines quickly had little value. Given how much the G5 cost, I wanted to get something substantial for it. My intention was to buy a new MacPro. However, I felt at the time that the MacPro was due for an update (and rumors were pointing at that) so I decided to wait. In the interim, my MBP became my primary computer with only an old Dell PowerEdge as a second Linux machine.

It’s been a good and bad experience with the laptop as the primary. But I think I turned the corner and am mostly comfortable with it. The big problem was my work habits had to change a little. I treated my laptop in the past kind of like an iPod — it was always a snapshot of data from a master (in my case, my G4 or G5 desktop). Also, I have enormous iTunes and iPhoto libraries (200 GB and 20 GB). These don’t fit so good on a laptop.

I still want the MacPro as the ultimate multi-boot option (OS X/Linux/Windows). But I’m just not sure I can justify the cost. I have one at work and it is a monster (and I only have the 2.6). I will have a hard time bringing it to its knees for awhile. I run VMWare heavily on it. At home, I do more media processing but I do some development. What I think I really need is a Pro class single Core Duo 2. Something like the MacPro but with slightly less and lower cost. Like the lower end of the Dell XPS line.

However, Apple has never offered this machine. I had little hope that the event this last Tuesday would address what I wanted and I was mostly right. The iMac is close but I don’t want another screen. I upgrade machines and like to repurpose them. iMac’s are good as a desktop only. That leaves the Mac Mini or a non-Mac option.

Why do I want a second machine? Well, sometimes I want to run things that take a long time to do (video conversions usually). I also like Elgato’s EyeTV and would like to record video using that. My laptop is a poor choice for all of these. I’d either be chained to the desk when I’m running these or competing with some intensive processes for time to do anything else.

I found two options. The Mac Mini Intel Core Duo 2 2.0GHz or the Dell XPS 410(n). The mini loaded to the max (2GB RAM/160GB HD) comes in at $963. It’s as good as the machine will ever be. Intriguingly, the base specs are very similar to my MacBook Pro except that my MBP has a 2.3GHz and a much better video chip. Is it safe to assume the raw CPU performance of the Mini is on par with a MBP?

The Dell option highlights the real problem with the Apple hardware lineup. I found two XPS 410 configurations I liked - 410N with Ubuntu (2.4GHz/2GB/500GB) for $930 and a 410 with Vista (QuadCore 2.4GHz/2GB/500GB) for $1329. That means $400 buys another two cores and 0.4GHz per core. Vista of course would be nuked for Linux.

On the surface, the Dell is the clear winner. It is much more hardware for the money. Factoring in software capabilities, the Mini gets better. OS X is a big plus. I have and like EyeTV (Linux has the less refined MythTV). The Mini can be an iTunes server (although RhythmBox on GNOME supports DAAP serving except for DRM tracks). The Mini will use less power and space in my home office.

So, I’ll have to decide how bad I want to run the Mac software stack I guess. I wish Apple had a $1200 “Mini Expandable Pro” or entry MacPro in this price range. Instead, they have a gaping hole. For those of us that think an iMac is not a solution, there is a hole from the top end Mini of ~$900 to ~$2200. That’s insane. At least nail $1500. Apple is practically telling me to buy a Dell.

I’ve been hoping to see some reviews of the new Mini but so far nothing. I wish I could see some benchmarks showing me how fast (or slow) the Mini is. It seems like it would be an excellent workhorse for processing jobs like I described. I’d add firewire hard drives for storage of video. I could also find lots of other uses for a Mini even if I do get a MacPro eventually. The XPS as good as it is might not be enough better than my PowerEdge (P4 2.4/2GB). The PowerEdge makes a very good Linux machine. But I’d have similar problems sharing it with processing jobs when I use it for development.

Finished Summit To Surf

This post is really late but I finally got around to uploading my pictures to Flickr. I did indeed finish the Summit to Surf two weeks ago. Other than driving the long way, it went very well.I had never done the ride at Mt. Hood so I decided to go with the 48 mile. Unlike at Redmond, there was no official start times. You check-in and take off from your choice of starting points. There were some extras you could do like a 12 mile climb to Timberline but since I was unsure of the route, I stuck with the 48 mile route.I had been training pretty hard up to the week before. My pace was up and I felt pretty good. I knew that the first part of the ride would be the toughest with a long climb up the pass before the descent to Hood River.It was a tough climb but not as bad as I thought it might be. I should have started at Welches and done the full 56 mile. Once I hit the first rest stop, it was very easy. The next stage was the heavy descent and I easily averaged 30 MPH with speeds over that. I was on the tail end of a group of 4 drafting so we really cruised. I finished the full 48 miles in 2:21. That was ahead of the pace I thought I would do so I was very pleased.My wife’s company United Healthcare was a primary sponsor. She was there with my daughter to help out at the rest stops and with the event. I hope someday to do a ride like this with the whole family (my wife rides as well). My daughter is doing a good job riding around the neighborhood but it will be a little while before she goes barreling down mountain passes.I had a blast and am definitely hooked on doing the event rides. It helps me focus on the training. I plan to keep it up as much as I can. I did 13 miles this morning and have kept up at least 3 times a week with one extra on weekends when my wife lets me. And the Tour de Cure puts on first class events. Both Redmond and Hood were well managed with great support. I look forward to doing both again next year.Flickr photoset is here.

Mac OS X Officially UNIX

Apple has passed the Open Group requirements to officially be called UNIX. Not sure this is terribly important but seems like a nice marketing tool for Apple.

OSCON 2007 Wrapup

Although it is a few days late, I thought I would post something about the last two days at OSCON to wrap it up.Thursday was a big improvement from Wednesday. The talks were more in sync with what I was interested in. The keynote was good. Bill Hilff from Microsoft’s Open Source Lab talked about open source and Microsoft. He did a good job given the environment here. I think he is making progress on helping Microsoft understand open source and to even think about opening up their own work. Honestly, if Microsoft wanted to beat Linux, they’d open source Windows. But instead, it’s baby steps. I guess you can call that progress but they have a long way to go.I went to a series of Rails talks that all were very good. First talk was on Rails plugins. There are several projects I wasn’t aware of allowing you to freeze gems into your vendor tree. I’ve done subversion tricks in the past to do something like this. In fact, we usually would put snapshots of the source gem into the subversion repository and install from that. It is messy and hard to track. Freezing them lets you worry about installing gems on the development machine and then creating what almost amounts to a “build” to hold the deployable source and gems. Very nice.The next talk was on extending Rails with C. Most of this applied to Ruby in general but there was a bit of a Rails slant to it. Most of the grunt work is easy to do with some helpers (in particular extconf.rb). You can register and run static C code in Ruby or run C inline. The inline is a little crazy. It will compile C code on demand and use it. I can’t ever see using this in production but makes an interesting development tool. You can verify if performance tuning critical sections with C will work before committing to building a library.After lunch was another Ruby on Rails talk. This was presented by Jacob Harris from the New York Times. His talk was “Rails Under The Knife”. It dove into exposing how some of the magic in Rails works. In particular, he focused on how blocks, metaprogramming, and reflection all contribute to how Rails can do so much with so little code. The core premise is that code can and should be manipulated just like data. This is of course against everything we (I mean old school programmers like me) learned. In the C world, code and data are distinctly separate. In Ruby, it is much more fluid. It was a good talk and I recommend taking a look at the slides on his site.The other good talk for me on Thursday was on Prototype (the Javascript library). It is timely as the next phase of the project I’m working on is to bring in Javascript/AJAX goodness. I’ll readily admit I’m a Javascript bigot. I’ve never viewed it as a serious language. I still don’t think I’ll love it but the presentation gave some good basis for why things are the way they are in Javascript. Prototype as a library (especially in Rails) hides much of the Javascript ugliness so you can get the effects you want easily.Friday was only a half day. In the keynote, Philip Rosedale from Linden Lab talked about how all of Second Life is being open sourced. He made some very compelling arguments. Linden Lab is not positioning Second Life as a game. They believe that it is an evolution of the web to the next form of browsing and online interaction. He pointed out that there is simply no way Linden Lab can do this on their own. And that they would be better in the long run by being there first and helping to shape it. It is a very insightful point that many companies don’t get.The two talks I went to were basically repeats of talks I had seen at WWDC. One was on the LLVM project. Unlike the WWDC talk which was about how it fits with Xcode and Mac development, this version was about LLVM fitting in with the rest of the development world. The short version is that LLVM is a collection of compiler libraries including parsers, optimizers, and code generators. Each can be replaced as needed. In theory, you could write a parser for any language that generates LLVM intermediate language and get the rest of the compiler chain for free. There is a missing link between the scripting VM’s (Ruby/Python/Perl) that something like Parrot could solve. I wonder how widely LLVM will be adopted. I don’t know if the GCC community feels threatened by it or not. Much of LLVM seems to be about bring the GCC tool chain into the 21st century.The other talk was on developing for OS X using Ruby. Apple seems to be doing a great job of putting in support for Ruby as a first class development language. There was an example showing an irb-type application that you could dynamically make Cocoa framework calls and have it execute immediately. Even if you are going to write your app in Objective-C, I can see using this for testing out ideas or experimenting with the frameworks. Transitioning from Ruby to Objective-C is very straightforward and compatible.So there you have it. I got a little less out of this conference than previous ones but I’ve been immersed in the open source world for several years now. But it still cool to see such a wide range of topics covered in one place. I didn’t talk about several other talks I dropped in like one on Linux kernel performance testing. Too many topics, too little time. O’Reilly has posted most of the slides and I believe some videos up on the OSCON website.

Slingbox on the Road

I bought a SlingBox PRO earlier this summer. I primarily bought it for watching TV in my home office on my computer from my DirecTV HD-DVR. It performs fantastic on my fast home network.

This week, I’ve been watching it remotely in the hotel. I’m very impressed. I’m getting 800+ kbps. That’s about the upper limit of upload bandwidth from my DSL connection. The hotel has an excellent wireless connection which helps as well. It’s fully watchable fullscreen over the internet. Makes me want an iPhone SlingPlayer that much more.

  • Meta