University of York Neuroimaging Centre chooses ‘faster and cheaper’ Apple Mac over Windows, Linux

“The University of York’s Neuroimaging Centre has made a £300,000 investment in Apple technology, including the Mac OS X operating system. Jobs which once took days now take minutes, improving productivity and lowering costs,” Lindsay Clark reports for ComputerWeekly.

“A review by the university’s computer science department found the operating system to be faster and cheaper than Windows and Linux for the scientific imagining applications,” Clark reports.

“‘We wanted a high-performance computing system that could cope with many potential users and fast turnarounds,’ professor Gary Green, director of the Centre, said. ‘In the past, processing jobs could take weeks, and yet much of the analysis is fine grained and very repetitive, so we were looking for a system that could handle large data sets in parallel,'” Clark reports. “Green wanted to deploy the system on a grid system, which could allow it to scale and evolve when necessary. He also wanted off-the-shelf software at a good price.”

“Having decided on the specification, an independent survey was conducted by a team of computer scientists. Windows, Linux, Mac OS X and Sun technology were all assessed with a fixed set of criteria,” Clark reports. “A long list of standard brain-imaging software was tested on them, including Matlab, Atlas, Maple, Brainstorm, Neurolens, Brainvoyager and MPI. More general software programs were tested such as those for firewalls, databases and e-mail. Technology was also assessed for security, back-up and support, future-proofing and upgrade paths, contingency planning and cost.”

“‘Apple won hands down,” said Green. ‘It was the fastest and easily the cheapest option in storage, back-up and computing,'” Clark reports.

“The Centre’s overall investment in Apple technology amounts to around £300,000. This includes 52 Dual Processor Xserve G5 cluster nodes, 8 Xserve G5 servers, 10 Power Mac G5s, 2 5TB and 2 3TB Xserve RAIDs (mirrored in another building for 32TB of total storage), 34 iMac G5s, 2 Opteron Linux Firewalls, 2 HP 128-port Ethernet switches, a 32-port Fibre channel switch, a HP tape library and various printers, faxes, photocopiers and digital cameras,” Clark reports.

Clark reports, “Green is particularly pleased with the performance of the Xserve cluster. He said: ‘We’ve already adapted software to the cluster so that it can perform proper parallel computing. Jobs that previously took days now take minutes, even seconds. This obviously makes a huge difference in terms of workflow and productivity.'”

More in the full article here.

MacDailyNews Take: Excellent choice, obviously, but: G5 Xserves, iMacs, and Power Macs? Sheesh. What year is this? No wonder they were cheaper.

That said, this shows the excellent raw performance those PowerPC G5-based Macs offerred.


  1. The fact is, though, they COULD choose older G5 technology and still get an increase in speed. The same CANNOT be said about the trash CPU known as the Pentium 4.

  2. I was at JPL a couple weeks ago for an open house, and there were Macs everywhere. All the computers set up to give kids a “virtual tour of the Solar System” and such were all iMacs. There were Mac Pros w/Cinema Displays in the rooms where people work as well…

  3. These are all PowerPCs, even the iMacs. I’m sure they got a very good deal. I don’t think Apple’s still building these, so this was probably inventory. Worked out well on both ends.

  4. Hey!!!

    If the G5 excels in one area, it’s in number crunching.

    Please note that the fastest computers on the planet are utilizing PPC derivatives, not descendants of the 8086.

    Face it- If IBM had been able to create a mobile variant of the G5, we *might* still be based in PPC.

    Apple’s G4 laptops were getting slaughtered by Intel, however, and that alone might have forced Apple’s hand.

  5. I used to have a iMac g4 (the pretty one with the floating monitor) running @ 800 Mhz with 1Gb of ram, and ir performs special effects in iMovie as fast as my new iMac Core2Duo @ 2Ghz with 1 Gb of ram. in a single task, the G4 was as fast as the core2duo at higher speeds. I never compared with G5s but I am pretty sure that will out perform by far the Intels ones.

  6. I still say what would totally rock is Leopard and XGrid (I think that’s the name of their shared computing system) on a Cell processor.

    The Cell is PPC based (both for the primary core and for the cell cores), so that much isn’t a problem, and I’m sure XGrid and Leopard can be adapted to use the Cells without too much code change. That would simply rock!

Reader Feedback

This site uses Akismet to reduce spam. Learn how your comment data is processed.