Dead Ink Vinyl

The Adobe MAX conference is in my backyard this year. The lack of travel expenses and the steep discount provided to attendees of 360|Flex made it possible for me to attend.

After registering, I chatted with other developers in my company to determine what sessions would be of greatest benefit to the teams using or investigating Flex. I shuffled those in with some sessions I really wanted to attend and the resulting schedule is below.

Data Visualization with Flex

Customizing the Flex Framework

Continuous Integration with Flex, FlexUnit, and Ant

Optimizing ActionScript 3.0 Performance

Inspire Session: The Dawning of the Age of Experience

What Makes a Design Seem Intuitive?

Pixel Perfect Precision

XD: Prototyping Rich Internet Applications with Fireworks CS3

XD: Adobe’s Approach to Application Design

Optimizing Flex Applications

Designers and Developers CAN Love Each Other

Creating New Flex Components

Flex on Rails

I look forward to seeing some of you there!

The OpenLaszlo Project Blog gets some benchmarks of Java performance on the new MacBook, relative to the G5.

So the second thing that happened is that we got a shiny new MacBook. Today I compiled LaszloMail on a co-worker’s desktop G5, and the shiny Intel thing. Ready?

• G5 (dual 2ghz, 1 gig RAM), OpenLaszlo 3.2: 32 seconds
• MacBook Pro (dual core 2ghz, 2 gig RAM), OpenLaszlo 3.2: 17 seconds

Hot damn!

You said it.

In a previous post I wondered about the Java performance on the new Intel Macs. I now have a benchmark.

Okay, the term “benchmark” might be too strong. I ran one of my favorite applets on the 15″ MacBook Pro while at an Apple Store today. I took screen shots of the results to compare to the performance of my own PowerBook. The applet is a 3D visualization of a KohonenSelf-Organizing Map. The Kohonen Map is a type of neural network used to automatically group similar inputs and provide visiualization of large quantities of data. Another interesting application of the Kohonen Map is to solve very large TSP-type problems.
Anyway, this applet is interesting because it has simple 3D graphics, lots of floating point and integer math, and possibly a good amount of object allocation and garbage collection. I don’t know if the applet is multi-threaded, though, but by the results I would guess that it is not. Thus, this is a highly imperfect comparison between the PPC and Intel Macs.

System specs:
PPC: 17″ 1.33GHz PowerBook with 1.0GB RAM
Intel: 15″ 2.0GHz MacBook Pro with 1.5GB RAM

Raw Results:
PPC:

Intel:

Numbers:
PPC: 1107 learning cycles in 138.697 seconds (7.981 learning cycles / second)
Intel: 1100 learning cycles in 62.514 seconds (17.62 learning cycles / second)

X-Factor:
The MacBook Pro is 2.2x faster than my PowerBook on Java performance. Assuming that the applet is single-threaded, the MacBook’s JVM 1.47x more efficient per CPU clock cycle.

I’d like to also note that the applet’s 3D animation was much smoother on the MacBook than it was on my PowerBook.