April 25, 2007 | Comments: (0)
MacBook Pro health report, November through April
The MacBook Pro that Apple generously allows me to carry as my primary notebook is showing some signs of aging. It was sent to me new last November, and I've treated it well. It's been through 123 full charge cycles, higher than I expected because if I'm anywhere near an outlet, I'm plugged in. Still, it's fairly young in laptop years.
My battery life is down to about 2 hours per charge at the lowest possible energy settings, a bit early in my experience. It's also clear that the power management circuitry is dazed; a full charge reading that's displayed as "minutes to empty" starts at about 1:45. The machine takes forever to charge and discharges as if it's at full power when it sleeps. I have done a couple of deep discharge cycles and zapped the PRAM to no avail.
More annoyingly, the MBP will randomly report something like 60 minutes remaining on the battery with a comfy and linear ramp down, then suddenly snap into death sleep mode with the first charge status light on the battery pack doing its last gasp rapid flash. The MBP has become the first notebook for which I carry with its charger no matter where I am.
This MBP has developed a few other maladies that are beginning to interfere with my work. First is that pop-top Delete key I've mentioned before. That's no gag any more--I need to get it fixed. The second problem is one I've encountered in every Apple PowerBook and MBP since the last two PowerBooks. When the trackpad is set to tap and drag, the function works flawlessly on a new machine, then deteriorates such that moving the pointer often sweep-selects everything you fly over. Trackpads of yore had calibration routines to cover this circumstance. The case warps and bows with usage, and it's hard for the pad to tell there isn't constant pressure on the pad.
Lastly, OS X dealt me a once-in-a-lifetime catastrophic failure on the MBP. My primary drive was within about 250 megabytes of full. For whatever reason, I didn't get the pop-up from the OS. A file write failed for having too little space. The app closed and deleted that file, at which point I exited all apps and quickly wiped out about 5 GB of temporary files from the command line. This did not add to the system's free space, which is a bad sign. The system froze on shutdown, signaling that the final sync to disk never completed. Sure enough, after the reboot, when I opened Mail.App it had forgotten all of my accounts and layout preferences; oddly, rules remained. The shortcuts I had assigned in Butler were gone, but Butler still launched. The default font for text boxes and lists changed to a scrawny scrawl that shipped with Word.
It added up easily: OS X had destroyed every writable file that was open when the sync failed. After the boot, Software Update offered me an upgrade to the very OS version that I was using. The machine wouldn't boot at all after that upgrade, and it took a rare and delicate foray into single-user mode to clean things up. Before you ask, I do have backups, but I must apply them selectively. OS X software recreates configuration files when they go missing. I can't just do a "restore what's not there."
It's a small adventure in the scheme of things that leaves a faint sour taste that will persist for a day or so at most. My Mac switching study victim is still coming to terms with MacBook, which she finds more accommodating, and my loaned MacBook Pro will still probably see plenty of action before it gets any treatment.
Posted by Tom Yager on April 25, 2007 11:33 PM
April 22, 2007 | Comments: (0)
Apple TV is certainly no iPod, and it's nothing like a stripped-down embedded Linux box. I've concluded that Apple TV as close to a Mac as you can get for $299.
At a minimum, Apple TV runs Darwin (Apple's open source UNIX core), and there is enough Mac-grade kick in the presentation layer to support OpenGL, OpenEXR, xpdf, khtml and real-time decoding of video and music, including iTunes DRM-protected content. It has a 48 watt power supply and a measured draw of 20-21 watts, enough for an embedded PC. It is supposedly targeted at consumer HDTVs, yet Apple TV drives a 30-inch Cinema Display at its full native resolution. On the Cinema Display, the graphics and text generated by Apple TV are not scaled up from HD resolution, but are rendered and anti-aliased at the display's native scale. That calls for more graphics RAM (shared RAM, given that Apple TV almost certainly uses Intel's chipset integrated display controller) than a consumer electronics vendor would factor into its parts cost for a $300 component.
Apple TV's USB port, presently unused, has everybody talking. I believe that it's there for a keyboard and/or a game controller to be put to use for custom software and for streaming interactive TV. It could also accommodate a tuner or other means of bringing video into the box, but I consider that unlikely.
Did I say "custom software?" Yup, that was I, but don't get excited. I expect the development model to be like that of iPod, where only Apple and Apple-blessed third parties can code for it. During my briefing on Apple TV, I posed the question about an open Apple TV SDK and got a response that I took as "are you kidding?" Too bad. The idea of embedded OS X is awfully enticing and accounts for much of my interest in iPhone.
That's not to say that some enterprising hacker with a spare $300 can't get inside Apple TV purely for the sake of blogging about it.
If you're still unconvinced that Apple TV is a near-Mac, Apple's list of Apple TV open source and license acknowledgments, which I scraped from the screen, closes the case. Compare this list with OS X's list of acknowledgments and note the differences.
(How did efax get in there?)
3Dlabs, OpenGL2 Shader Language compiler
Various, BSD kernel
Apache, Apache and apache_mod_perl
AT&T, C library
Casas, efax
Clapper, poll
Corcoran, Smartcard Services
Creative Labs, OpenAL
Demetriou and Ustimenko, ntfs
Digital Equipment, bind and BSD kernel
Eaton, man
Elber, Raymond and Kuratomi, giflib
FSF, bash, gcc, gnutar, grep, libiconv, ncurses
Gailly and Adler, zlib
Gifford, SHA2
Giraud, smart card reader drivers
Gladman, AES, SHA2 message digest
Glyph & Cog, Xpdf JBIG2 decoder
Hinds, PC Card driver
Hipp, SQLite
IBM, Unicode
Indulstrial Light & Magic, OpenEXR
Johansen, Levenshein, Distance C++ code
Juniper Networks, PAM
Knoll, et al., khtml
Lane, JPEG library
Leffler and Silicon Graphics, TIFF library
Lucent, awk
MIT, Kerberos, WebDAV, install-sh
Matsumoto, Ruby
Miller, sudo
Mills, NTP
Moolenaar, gpt
Morgan, PAM
Mosier, cephes math library
Muffett, CrackLib
Netscape, arena files, parser
Netscape, network security services
Network Associates, et al., NET-SNMP
Nudelman, Less
OpenBSD, OpenSSH
OpenLDAP Foundation, OpenLDAP
OpenSSL Project, OpenSSL
OpenVision Technologies, Kerberos Administration System
Open Software Foundation, Mach
Percival, bspatch, bsdiff
Pixar, et al., tif_pixarfilm.c
Porten, et al., kjs
Randers-Pehrson, et al., png
RSS Data Security, MD4.H
Schlumberger, PKCS-11
Seward, bzip2
Solfrank, et al., msdosfs
Spencer, regular expressions
Steinberg, curl
Tsirigotis, xinetd
Ts'o, libuuid
Unicode, ConvertUTF
UC, Sun, Scriptics, Tcl
University of Cambridge, PCRE
UCAR, Unidata, wordexp(), wordfree()
van den Berg, procmail
Vaillard, libxml2, libxslt
Venema, TCP Wrappers
Vixie, cron
Wall, Perl Kit
Winning Strategies, asa
WWW Consortium, tidylib
Zimmerman, et al., kcanvas, kdom, ksvg2
Posted by Tom Yager on April 22, 2007 02:44 PM
April 09, 2007 | Comments: (0)
For Apple, 8-way Mac Pro is a stepping stone
Welcome back.
For about $4,000, you can now buy a Mac Pro fitted with a pair of 3 GHz Clovertown Xeon CPUs, creating an eight core OS X workstation and the meanest Intel Mac in Apple history. It's precisely the same box as the formerly-meanest Intel Mac in history, the four core Mac Pro, of which breed a well used and well appreciated exemplar sits at my right elbow, driving a 30-inch Cinema display.
The high initial cost of Intel's quad-core Xeon CPU, that price being buoyed by AMD's delayed arrival in that space, precluded Apple's usual tactic of transforming Mac Pro from four cores to eight overnight. I project that in Q3, when price pressure from AMD does come into play, Intel quad-core Xeon prices will fall to levels now occupied by dual-core parts, and the dual-core parts will begin to roll off the roadmap. As this happens, Apple will probably make the $3,298 3 GHz Mac Pro base model an eight-core machine. I'd be very surprised if Apple had a date set for that yet. Intel will squeeze maximum margins out of its quad core parts for as long as it has the field to itself (and it should).
Intel recently announced quad core CPUs with reduced power requirements, and the close timing of Apple's announcement might have caused some confusion. The 3 GHz Clovertown CPUs are not reduced-power devices compared to the dual-core Woodcrest. In a workstation configuration that includes 8+ GB of RAM, a dual-slot graphics accelerator, multiple 750 GB hard drives and Fibre Channel/10GbE/Infiniband bus adapters, a wattage drop affecting the CPUs alone would hardly be noticed. The reduced-power quad core CPUs will find their way into iMac as soon as Apple can lay hands on them.
There is not yet an eight core option for Xserve; the 3 GHz dual core Xeon is still Xserve's top CTO processor option. My hunch is that Xserve is in engineering right now and will re-emerge as a native eight core server. This will be part of a program that will see Apple gunning for enterprise business again now that the dust has settled on the consumer side.
What about an after-market upgrade to take your quad-core Mac Pro to octo? There is none in the offing from Apple, and it's a none-too-appealing option to tackle on your own at the moment. A manual chip swap to take an existing Mac Pro to an eight-core configuration will set you back at least $2K, and that's just for parts. As I say, that will get a lot cheaper.
I believe that $699 is the right price for the Clovertown CTO option. I expect--and this is based solely on experience with non-Mac x86 systems as I don't have an eight-way Mac Pro--that going from four to eight cores will reduce the time to completion for compute-intensive, non- or semi-interactive, threaded workstation task units (e.g. render, simulate, compile/link, analyze, transform, filter) by ten to thirty percent without code changes. (Sorry for the mumbo-jumbo. I'm putting a fair bit of energy into developing criteria for meaningful performance characterization lately)
Would I shell out the extra $699 for the extra four cores? When I look at the Mac Pro I use, I'm not wishing that it had more cores. A maximum of 16 GB of memory is very tight for an 8-way, 64-bit box. When I dream about the Mac Pro after this one, I see standards-based system management, larger memory capacity, better power efficiency and faster connectivity. Mac Pro feels just about balanced to me as a four core machine, but you can be sure that I don't work this Mac Pro the way you work yours. As a developer, I'm also living in Leopard, which, as any enterprise OS does, inevitably changes the way you look at x86 hardware. Leopard scales so bloody beautifully.
Posted by Tom Yager on April 9, 2007 04:28 PM
TOP STORIES
IBM boosts BlackBerry accessIntel to develop PC with Alibaba
Adobe refreshes Flash Player
Cybercriminals can rent a botnet
Comcast to buy Plaxo social network
Rootkit for Cisco routers
Leopard interface tweaks
Icahn to launch proxy fight
Office VBA and Mac IT
Test your Geek IQ
ADDITIONAL RESOURCES

- Virtualization: A Step by Step Approach to Success
- Dialing up Agility with Business Transformation
- 5 Things You Need to Know About Storage Virtualization

- Need to Secure Your Virtualized Environments?
- When you need backup will it be there?
- Is your storage capacity holding you back?


