A Mid-September Teaser

As I sit here typing away on my Windows-powered laptop, I've got three Macs (a dual-G4 Power Mac, a dual-G5 Power Mac and a G4 PowerBook) around me, fully occupied running the SPECINT2000 benchmark suite. By early next week, a Mac mini, this laptop and a Hyper-Threaded 3.2 GHz Pentium 4-based Windows desktop PC will join them. What am I up to? Well, now that a short blurb on my September 15 cover story has been published, I thought I'd share a bit more information on this hands-on project with all of you, EDN's readers.

The paragraphs below come from a heads-up memo that I recently sent to PR contacts at various companies who are related to the article topic. For more background on the motivation for this project, see my past blog entries 'Fast-Track the SAF' and 'Open-Source Bias Benefits No One'. Feedback, as always, is welcomed; preferably by posting a comment to this blog entry but also, if you insist, via private email.

Three recent events greatly influenced the status of the PowerPC CPU family in embedded applications and the broader electronics market:
1) Beginning in mid-March, IBM published a series of articles discussing the Apple Mac mini as an embedded hardware and software development platform, both under the Mac OS and with various iterations of Linux and BSD Unix.
2) In mid-May, all three next-generation consoles (Microsoft's Xbox 360, Nintendo's Dreamcast and Sony's Playstation 3) were unveiled at the E3 Expo, containing various spins of the PowerPC architecture, and
3) In early June, Steve Jobs announced that Apple would begin a phased transition away from the PowerPC, to Intel x86 CPUs.

Is the Mac a valid option as an embedded development platform for PowerPC-based system designs? What does the embrace of the PowerPC architecture by next-generation game console developers indicate about its cost, performance, power consumption and other advantages? And conversely, does Apple's embrace of x86 suggest that the PowerPC is at a crossroads versus AMD-, Intel- and Via-supplied CPU alternatives?

In this hands-on article, Brian plans to explore all of these questions, both in print and online. He'll first test the premise of IBM's article series, evaluating the applicability of the Mac mini as an embedded development vehicle under both the Mac OS and Yellow Dog Linux. He'll then compile the SPEC benchmarks under GCC (the GNU Compiler Collection), some of which is included within Apple's Xcode development toolset, and run both them and the Xbench benchmark suite on a variety of single- and dual-CPU, G4- and G5-based hardware, with varying combinations of DRAM main memory capacty, HDD rotational speed and (where appropriate) graphics subsystem. Finally, he'll compare the PowerPC benchmark results not only against each other, but against SPEC benchmarks run on various AMD, Intel and Via CPUs, again compiled both under GCC and with CPU vendor-provided compilers.