Technologic

I have a thing for experimenting with operating systems.  I believe it’s healthy to see what else is out there, and to learn what kinds of transitions have to be made to move from one OS to another.  Easy translations from one OS to another are comparable to cognates in human languages.  Some may argue that it’s more logical to just stick to what you know, but I suspect that it’s more logical to maintain enough flexibility to function in different software and OS “countries” with their distinct interfaces and dialects.  It’s not wise to remain linguistically or geographically insular, lest the country next door suddenly decides to colonize.

I can move between editing audio in GarageBand on a Mac to ProTools on Windows; I can survive in Linux with Open Office, Firefox, and that initially baffling image editing software, GIMP, as a replacement for Photoshop.  That doesn’t exactly make me multilingual in these OS “countries”, but I am able to successfully cross the border and get a little business done without hurting myself…usually.

Today I decided it was time to take a tour in undiscovered territory, so I packed my Parallels Desktop up and set off for the land of the Sun.  Sun’s Solaris homepage welcomed me with: “Be Brilliant Faster.”  That sounded encouraging.  But from this point they immediately divide you into two camps: “Already Using Solaris?” and “New to Solaris?”  Easy choice – I’m new, so off I wandered into the land of “OpenSolaris” and I downloaded the LiveCD ISO (687 MB) to install via Parallels Desktop onto one of my Macs.

I’ve installed a number of OSs via Parallels before: Windows XP, Windows 7 RC, SUSE Linux, and Ubuntu Linux, and it was getting things going has usually been reasonably straightforward.  Not so with OpenSolaris.  This is migraine territory unless you really know what you’re doing.

I followed various pieces of advice to set up the virtual machine in bridged networking mode in Parallels, and I also needed to link an additional disc image with network drivers to the OS.  Then I wound up executing a shell script in the Terminal… and then I started to realize that I just didn’t have a visa to enter OpenSolarisland with a Macintosh passport.  OpenSolaris was closed.

I was not feeling at all like I was getting brilliant faster, until it occurred to me to return to the Parallels website and recheck the supported OSs for the current build of Parallels Desktop.  Wait a minute – it doesn’t mention anything about OpenSolaris – it only states that it supports versions 9 & 10 of Solaris.

So back to Sun I went to take another look at their enterprise version of Solaris, which also happens to be a free download if you’re willing to part with your vital statistics.  In the process they offered me membership to the Sun Developer Connection, so I grabbed that badge along the way.  The enterprise version of Solaris, however, only comes in full-strength enterprise size, so I’m still downloading the 3 GB or so that I need just to get started.

So there you go.  I just spent hours trying to install Solaris, but I’m basically back to square one.  Despite the fact that I haven’t even been to Solarisland, I’m now a confirmed member of the Sun Developer Network.  It all feels so logical and natural.  Strangely enough, I think I did get a good idea of what Solaris is like today:

A psychiatrist still dealing with the loss of his wife, Chris Kelvin receives a disturbing video message from a friend and scientist, Gibarian, asking for Chris’ help and that he come to the enigmatic ocean world, Solaris. He agrees to go on the mission to Solaris as a last attempt to recover the crew. Kelvin, arriving at the space station, quickly learns that members of the crew have died (or even disappeared) under mysterious circumstances with the only two surviving members reluctant to explain the cause. After shockingly encountering his dead wife alive again, Chris discovers that Solaris has been creating physical replications of people familiar to each crew member. Up until the end, Chris struggles with the questions of Solaris’s motivation, his beliefs and memories, and reconciling what was lost with an opportunity for a second chance.

NOT MY NEXT POST: “Why did it take me nearly all day to realize that Safari 4 beta causing my blog admin to hang nearly every time I tried to embed links or pictures?”

This blog post was kindly rescued by Firefox.

It’s somehow appropriate that the Wordpress admin pages on this new blog are failing me for my very first post, and that I’m entering the HTML manually as a vanilla text document with manual markup. This is how far we’ve come, but we have so much further to go! The ‘logic’ in ‘technological’ is sometimes the elusive white rabbit who entices us into Wonderland, and then fleetingly appears in baffling guises that force us to solve puzzles before we can get back to the plot.

This morning I received an email from the SETI@home project to let me know that I haven’t processed any work units for 60 days. I joined SETI in 2003, and I have to admit that one of the most inspiring emails I’ve ever received came in the form of mass mail from the late Arthur C. Clarke, who was a vocal ambassador for SETI and other distributed computing projects supported by BOINC, Berkeley’s Open Infrastructure for Network Computing.

After opening BOINC on several Macs and refreshing my work unit load, I proceeded to attempt the same on a new Mac Mini, which is full of migrated apps from an old G5 PowerMac. Lo and behold, BOINC wouldn’t launch, and I needed to return to to Berkeley’s BOINC site to get a fresh copy of the BOINC client. This was all executed without any problems, but I noticed a new message on the download site that claims, “Note: if your computer is equipped with an NVIDIA Graphics Processing Unit (GPU), you may be able to use it to compute faster.” Further information is provided on how NVIDIA’s GPUs use “CUDA” to leverage parallel processing in the GPU to achieve much higher speeds than are achievable in the CPU. Bravo!

What a great opportunity to logically combine all of the resources that are now featured in a standard computer. However, I searched for support and information for the GeForce 9400 on my MacMini, and the GeForce 9400M on my Al MacBook, and there’s nothing there for Mac. My current iMac has an ATI Radeon GPU, and my other MacBooks have Intel’s integrated GMA system, so I was unable to find a way to leverage CUDA for scientific/distributed computing on any of my Macs. Just to be sure, I checked the messages log in BOINC on all of these machines. Each reported back: “No coprocessors”.

So can we expect any updates on CUDA from Apple in the near future, other ways to leverage parallel processing in GPUs? I’ve previously looked at other options that take the strain of heavy processing off the CPU, such as Elgato’s TurboH264HD, which is a USB plug-in processor that frees up the CPU and speeds up processing for video compression. But do we need more accessories when we have underutilized power that’s just sitting under the hood?

Hopefully some of these answers will be forthcoming with the release of Snow Leopard, but I’m left scratching my head as to why this process would be presently easier to construct with a custom-assembled computer and a garden variety of Linux. How logical is that? As for the WiFi networking that supports my networked computing, I’m still finding that Apple’s firmware update 7.4.1 is crashing my Airport Extreme network, and I need to keep rolling back the firmware drivers to version 7.3.2. I have tried to upgrade my Airport firmware 4 or 5 times now, but each time the newer/recommended firmware is installed, my WiFi reliability bombs. As for BOINC, despite the fact that its network activity should be extremely low (ie. brief download periods to grab new work units, and brief upload periods to report back), its presence seems to also have a crippling effect on my network.

Therefore, I have underutilized hardware due to system and driver architecture, on top of which I have wasted processor cycles, as the BOINC app seems to bring regular WiFi IO to new lows. As a result, a group of five computers that could be efficiently contributing to scientific computing are essentially left doing nothing for most of the time just so I can support light web activity. This is not logical.

Metaphorically, I’m sitting in an idling Hummer in the middle of peak hour traffic, when I could be carpooling on the information superhighway. This is something that needs to be solved. Anyone?

NEXT POST: Opening Solaris