| RSS | Software | Links | Email Policy | About | ||
|
MacBook Pro external drive speed with Firewire and USB I’ve added a new section to my Apple MacBook Pro Experience Report on the speed of external drives using Firewire and USB. See the DiskTester speed tests—Firewire and USB section. I’ve added a section on power usage to my Apple MacBook Pro Experience Report. At least one user is concerned that airline power plugs max out at 75 watts. Apple axes Aperture engineering team? Rumors like this can really kill a product–especially if they’re true! I take the rumor very seriously, and I suspect many others do as well. Apple might not like to comment on rumors, but I certainly won’t be committing to Aperture until I hear appropriate non-weaselized statements from Apple about its future (including an alternative to the troublesome single-artifact file-storage approach). Aperture shows tremendous potential, so let’s hope that Apple stays in the game in a serious way so that other vendors are forced to bring their products up to a higher quality level. Traveling to the Thelon wilderness, part one. If you’ve ever thought about traveling to a truly remote area of the world, you might be interested in my experiences in the Thelon Wilderness in late August, 2005. Read the first part of my travelogue Traveling to the Thelon Wilderness. As time allows, I’ll cover the remainder of the trip. Windows API on Macs in Mac OS X 10.5? Speculation has begun that Apple might implement the Windows API directly in Mac OS X, making it the ultimate cross-platform computer. See OS X to run Windows XP apps natively?. This would be a stunning and useful development, assuming safeguards are in place to prevent destruction of MacOS by the many rampant Windows viruses. MacBook Pro 2.16GHz 17" cheaper than 15"! Perhaps Apple has yet to adjust their pricing on the 15" MacBook Pro. Configuring the 15" model with the 2.16GHz option and 7200rpm hard drive, and the 17" model (2.16Ghz standard) with the 7200rpm hard drive yields: MacBook Pro 15" 2.00/1GB/100GB@7200: $2599 I triple-checked the pricing—the 17" is $100 less expensive in the 2.16/1GB/7200rpm configuration . (Perhaps this means a price adjustment is coming for the 15" model; it might be good to wait a week or so if you’re about to buy one). At 2.0GHz, the 15" model, is $200 less than the 17" model (7.7%) more. Considering that the 17" model offers lots of extras, it’s a no-brainer to get the 17":
The Firewire 800 port is much faster if you use an external hard drive—its presence on the 17" model and the lack of it on my 15" model adds insult to injury—artificial limitations to position the 17" model? This is customer disservice, and a good reason to wait a few months before buying a new Apple product. As discussed in my Apple MacBook Pro Experience Report, the current non-availability of fast external hard-disk expansion is a major weakness of the MacBook Pro 15". The Firewire 800 port goes a long way towards addressing that weakness, assuming its write performance isn’t crippled as it is on the G5 PowerMacs. I’ve added a new article on the accuracy of manual focus. Please see Manual Focus Accuracy, but I recommend starting at Focus Accuracy. Photoshop drag-scrolling problem Recently, on MacOS X 10.4.6, Adobe Photoshop CS2 has developed a screen redraw problem while drag-scrolling (pressing the space bar and dragging the image around). It occurs in either full-screen mode (with or without menu bar), but does not occur when using a normal window with scroll bars. Here is what it looks like:  I’ve seen no other instability or problem. The most significant things I can recall about this development, most recent first, for the past month or so, are:
I’ve added a Buying More Memory section to my Apple MacBook Pro Experience Report, as well as a MacOS X section. Here’s a nice link for various utilities for MacOS X. See also the Apple Troubleshooting page. From the “look what crawled out from under a rock” department… A 3rd-grader offers a friendly, handwritten list of tips for improving the iPod to Steve Jobs (CEO of Apple Computer), and gets a response that makes her cry. No wonder lawyers get a bad name. Maybe Steve should give her a new iPod. I’ve reorganized the Reviews Page, moving the list of free articles onto their own separate page. As the list of articles has grown, too much scrolling has been required to see what’s available. I’ve added a Buyer’s Guide to configuring your MacBook Pro section to my Apple MacBook Pro Experience Report. Learn what the worthwhile options are, and how to save some money. The more I use the MacBook Pro, the more I like it. I have never been able to say that about any Apple laptop before—they were always sluggish compared to desktop machines. But the MacBook Pro blows away previous Apple laptops, and, performance-monger that I am, it is actually a machine that ought to suffice for 90% of the user base, especially after adding a mouse, full-sized keyboard and Apple 23" Cinema Display (or the awesome 30" Cinema Display—yes the MacBook can drive the 30" display!). I am continually impressed each time I compare my PowerMac G5 Quad to the MacBook Pro. See my evolving Apple MacBook Pro Experience Report. So I’ll say it plainly: Buying a G4 PowerBook is nuts—the MacBook Pro blows it away, even using Rosetta, and it will only get better as more Universal Binary applications emerge. Buying a G5 desktop other than the Quad is questionable, unless you must have the additional expansion capabilities it offers—the MacBook Pro 2.16Ghz is simply amazing with Universal Binaries. There are a few caveats for some users:
I’ve written up a brief test of autofocus accuracy with the Nikon D2X and 17-55DX—please see Autofocus Accuracy. The test is short and simple, but interesting. I’ll likely expand the article to include Canon EOS equipment at a later date. The D2X, along with the 17-55DX and 17-35AFS are now back from Nikon service for the lens mount problem I detected on my D2X. Thank you Nikon for keeping the repair time under 3 months, (though some portion of that time was the local dealer’s error). Nikon determined there was “impact damage” to the D2X, the 17-55/f2.8 and the 17-35/f2.8 (which I had sent in initially, believing that the lens was at fault. It was “interesting” to see “impact damage” for the 17-35/f2.8 also, since it had sat in a drawer (except for the test shots) since its last visit to Nikon not long before, where it had received a good bill of health. Erroneous notes only? I can’t say for certain, but my D2X itself never hit a hard surface. However, the Really Right Stuff “L” bracket on the D2X did fracture the largest bone in my middle finger when I took a nasty fall on slippery granite last September. The 17-55/f.28DX was attached to the D2X at the time. So “impact damage” is at least plausible, though the camera body itself had nary a mark on it, and never actually hit anything hard. Bone and flesh impact are apparently enough to warp a lens mount, even with the modestly-sized 17-55DX.). So far I’ve taken Nikon at their word “working to specification” and haven’t yet reshot the multi-lens test described in Lens Mount Misalignment—but I will soon. Are there actually a fair number of camera bodies out there with warped lens mounts, and their owners bemoaning the “soft” optics? The best way to check your own camera is to shoot at least two lenses on a tripod at the same focal length [see Lens Mount Misalignment], and see what kind of results emerge. Even a single lens on an appropriate subject will reveal a problem; the question then becomes whether it’s the lens or camera—so shoot at least two lenses if possible. Here’s the uneven sharpness you should look for:
Aperture’s speed on MacBook Pro with Canon CR2 raw files I’ve just started using the Universal Binary Aperture 1.1 on my MacBook Pro, and it seems quite speedy. I’ve done a brief test and added the results to my Apple MacBook Pro Experience Report. In short, performance is impressive, but memory use temporarily approaches 700MB real memory while batch-exporting—so plan on installing 2GB of RAM into you MacBook Pro if you intend to use Aperture. Aperture checks its serial number on the network I inadvertantly started Aperture 1.1 on my Quad while I had left it idle on my MacBook Pro—it apparently checks the network for other copies with the same serial number because it reports that the other copy is running, then quits. Not a problem for me, as I have no need to run both at the same time or to use it improperly, but it’s worth knowing. MacBook Pro 2.16 GHz beats dual 2.5GHz G5 for software development [See Apple MacBook Pro Experience Report for other MacBook comparisons] Macintosh developers will be interested to learn that the MacBookPro 2.16GHz easily bests two 2.5GHz PowerPC cores for builds. The Quad is faster than the MacBook Pro only when all 4 cores are used; otherwise, on a per-core basis, the Quad was 16%-30% slower in my tests. With such impressive CPU speed, it’s unfortunate that the MacBook is hobbled by a hard disk that is 1/2 the speed of a fast desktop hard disk, with no way to add something faster, at least not yet. The ExpressCard slot on the MacBook may yet yield a 3rd-party solution for fast external storage (the Firewire 400 port is even slower than the internal hard disk). This latest test adds one more data point suggesting that the 2.16 GHz Intel Core Duo cores will often match or outperform two 2.5GHz PowerPC G5 cores, in spite of the 340 megahertz clock rate handicap—which means the Macintosh users can envision considerably faster PowerMacs when Apple builds nearly 4 GHz Intel-based machines. MacBook Pro 2.16 GHz beats dual Opteron 2.4GHz And the PowerMac Quad (with WindowsXP-SP2, Digital Photo Professional, single worker). Forget the boring MacWorld magazine Boot Camp benchmarks noted yesterday— photographers who insist on running Windows will find the MacBook Pro (see my experience report) offers outstanding performance with Digital Photo Professional. See the numbers for yourself in my Digital Photo Professional Batch Processing Performance Tip article. Nikon Capture users should be equally impressed. MacWorld magazine Boot Camp benchmarks with Windows XP Macworld magazine has some interesting benchmarks running Windows XP on the MacBook Pro, comparing it to various other PCs. Even forgetting the demonstrable performance superiority of the MacBook in the MacWorld benchmarks, why buy a Windows PC when you can buy a MacBook and run both Mac OS (vastly superior to XP in multiple ways) and Windows XP (and Linux and Windows 2000 and many other operating systems using software such as Parallels and probably VMWare before long). A foolish choice indeed, to buy an inflexible plain vanilla PC, unless price is the primary issue. Then again, running more than on OS is confusing for most casual users. I’ve gotten around to installing Aperture, now that version 1.1 is available, and will be updating my Raw-file Converters article as time allows. Apple’s claim of greatly improved raw-file rendering sounds promising. The revised article will be free to prior customers, but the price will go up by $1 after the update. I’ve added some notes to the Apple MacBook Pro Experience Report—Aperture is now a Universal Binary. I will likely add an experience report for Aperture in a few weeks. Internal drive mounting kits for the PowerMacG5 Want more storage inside your PowerMac G5, such as a striped RAID? See my PowerMac G5 Internal Drive Kits article. You won’t find this kind of honest product-assessment in MacWorld magazine—diglloyd has no paid advertisers to please. The first revision of my Apple MacBook Pro Experience Report is now online. Included so far are general assessments of hard disk, memory, network speed and expansion, raw computing power and an assessment of various programs, including Adobe Photoshop CS2, Nikon Capture, and Digital Photo Professional speed under Rosetta. This report will be updated in the future, so check back periodically. 11 April 2006
Nikon Capture and AMD processors with Data Execution Prevention (DEP) It turns out that the crash (blue screen of death) that I described earlier provoked by launching Nikon Capture on my dual Opteron box is a known problem. Searching Nikon’s web site, I discovered a technical note describing the problem. It’s awfully helpful of Nikon to add a technical note—but it would be even nicer to have the installer detect the processor type, and warn the hapless customer doing the install. Such blatant disregard for the customer is outrageous—I wasted 45 minutes installing, crashing, uninstalling, reinstalling, etc before I found the technical note in Nikon’s online knowledge database. And it’s insult to injury that Digital Photo Professional is free and Nikon Capture costs $99. One more reason to switch to Canon. By the way, self-modifying code (eg turning data into code or modifying machine code in-memory) has always been an Evil practice even before today’s nasty viruses appeared, but Nikon apparently is years behind in understanding sound software design. Perhaps that explains why Nikon Capture makes poor use of multiple CPUs yet allows only one worker thread and disallows more than one instance of Nikon Capture (as discussed in my 09 April entry). Running Parallels virtualization software on the MacBook Pro Though not quite ready for prime time, Parallels is an appealing way to run Windows or other operating systems on your Intel-based Macintosh as compared with Apple’s Boot Camp. Advantages include:
Aside from a few confusing minor problems, and one apparent kernel crash, the Parallels virtualization software ran flawlessly. Here are some tips if you want to try it out:
Digital Photo Professional speed on Windows 2000 using Parallels I installedDigital Photo Professional 2.1.1.4 on Windows 2000, running on my MacBook Pro 2.16 Ghz using the Parallels virtualization software. I set “” to “” and allocated 900MB of memory (out of 2GB total). I then processed 30 EOS CR2 files from a Canon 1Ds Mark II (the same files as used in the Digital Photo Professional Batch Processing Performance Tip article). I verified that there was ample free memory during processing, and I quit the DPP window while the batch was in progress. The CPU usage according to the Windows Task Manager was 100% while processing the files. The CPU usage according to Mac OS Activity Monitor was about 125% (out of a maximum of 200%)—so the two cores on the MacBook were not fully utilized. Results: 9:14 to process 30 1Ds Mark II CR2 files or 22.8 seconds per CR2 (single worker). This is not too impressive, considering that the MacBook Pro takes just 29.3 seconds per CR2 while running the PowerPC version using Rosetta (emulation) [see 07 April entry]. Compare this with the 9.8 seconds per CR2 (single worker) seen on my dual 2.4GHz Opteron box (single worker), or 7.8 seconds per CR2 (2 workers). It’s not clear if this lesser speed is due to the virtualization software, as I haven’t yet tried Apple’s Boot Camp (I don’t yet own a copy of Windows XP Service Pack 2—if anyone wants to donate a legal, unused copy, I’ll be glad to accept it, or trade for one of my articles ). 10 April 2006
“Digital Photo Professional Batch Processing Performance Tip” updated The Digital Photo Professional Batch Processing Performance Tip article, as mentioned in the 08 April entry now includes results from a dual Opteron 2.4GHz machine running Windows XP. Although I post many free articles, and intend to continue to do so, I put great time and effort into a smaller number of in-depth articles, too. Those articles are not free, but if you haven’t read one of them, I think you’ll be pleased with the value you get—detail and insights like nothing you’ll find in a magazine. These paid articles help support this site (and the equipment purchase such articles require). How much do raw converters vary in image rendering? There is a never-ending discussion on how good one raw converter is versus another. Although “good” is a subjective term, there are differences, and subject matter can influence the conclusion. Further, there are so many aspects of raw conversion, that it can be a complicated decision of which one to use. Save yourself some of the guesswork, and take a look at how fine detail is rendered with 9 different raw converters in my Raw-file Converters article. Looking for the best 28mm shift lens? If you’re looking for a 28mm shift lens for architectural work and/or for “flat” stitching (higher resolution composites), don’t buy before reading my 28mm Shift Lenses article. You won’t find a better assessment of the available lenses: the Schneider/Leica 28mm/f2.8 PC Super Angulon, the Nikon 28mm/f4 PC-Nikkor, and the Nikon 28mm/f3.5 PC-Nikkor. Although the tests were done using a Nikon D2X, Canon users can use all of these lenses, too. Nikon or EOS, full-frame of DX? If you’re in the market for the Nikon D2X, the Canon EOS 5D, or the Canon EOS 1Ds Mark II, look no further than the most comprehensive review you’ll find anywhere, my D2X vs EOS review. This is not a winner/loser type of review, but a thoughtful and detailed comparative review, intended to help you make an informed and wise choice of one of these cameras. Soft pictures? Handheld or VR? How much does the tripod matter? In the most comprehensive analysis of image sharpness that I’ve yet seen, The Sharpest Image goes into myriad details on image sharpness handheld with and without VR, burst shooting, mirror lockup vs none, tripod stability and resonance. If you enjoy critically sharp images, or if you’re in the market for a tripod, or you regularly use Vibration Reduction, you owe it to yourself to read this article. Though the article used Nikon equipment, Canon or other users will not be disappointed, as much of the article is generally applicable. Mean Time To Failure (MTTF) clarification Thanks to a reader for pointing out a good summary on Mean Time To Failure (MTTF) and Mean Time Between Failure (MTBF), which I mentioned in my 02 April entry in context of the failure of one of my nearly-new Maxtor 7V300F0 hard drives. While I do have some training in statistics, and understood perfectly well that a MTBF of 1,000,000 hours does not mean a drive will last 1,000,000 hours, I wasn’t aware that there can be an “early failure period” with a high rate of failure, as shown graphically on this page. It’s still not clear to me if this “early failure period” is included in the MTBF figure quoted by drive manufacturers, but I suspect that it isn’t given the numbers involved.
All blog entries have now been indexed. The index might not be updated more than once a week or so, so it’s best to check the current blog page first. And by the way, did you know the google trick of using "site:diglloyd.com" to search just diglloyd.com (or whatever site you choose after the "site:" entry). This was new to me within the past few months, though I’m sure many people know of it. For example, to search for anything about the Nikon D200 at diglloyd.com, you would enter: “site:diglloyd.com Nikon D200”. (Click on the link and see for yourself). No more PowerMac Quad sleep problems As I noted in my 27 February entry, I was having freezes when entering “sleep” mode with the PowerMac Quad. Well, it appears that some MacOS update must have fixed things, since it hasn’t happened for about a month now. Followup on Nikon Capture and batching To double-check that there was absolutely no way to run more than one batch in parallel with Nikon Capture as described in my 07 April entry , I tried a few things— First, I duplicated the already-installed Nikon Capture folder, and tried launching the copy after the first one was already running. No luck—the 2nd copy launches, then refuses to go on, posting a dialog stating that “another copy of Nikon Capture is already running”. Some programmer actually thought about this possibility and added this check and dialog, so a reasonable conclusion is that the underlying engine is such poorly-implemented code that no more than one progress (program) can use it at a time. Otherwise, why make such a check? Next, I ran the installer off the CD, and tried to install another copy of Nikon Capture elsewhere. Unfortunately, there is no option to install it elsewhere; there is only one place that the installer will allow another copy to install: /Applications. Still, if I had been successful, the behavior would probably have been the same as the copy I first tried. So Nikon Capture users can resign themselves to having a “speed limiter” in place on all batch conversions, compliments of Nikon. It’s a good thing Nikon doesn’t make cars. :; Adding insult to injury, today I installed Nikon Capture on my dual 2.4 GHz Opteron box, hoping to gauge its speed. Any attempt to launch it immediately crashed Windows with the blue screen of death. I tried several times. Utilizing the Quad’s 4 cores with Digital Photo Professional I’ve written up a handy tip for exploiting the full power of multiple CPU and/or multiple core computers, such as the PowerMac Quad. This tip is applicable to any program that allows multiple copies of itself to run and/or uses multiple “workers” for any sort of batch processing of files, such as mass conversion of raw files to JPEG or TIF. Need fast or large storage? Don’t miss AMUG’s review of Sonnet E4P AMUG has posted an article detailing performance of eSATA with and without “port multiplication” using the Sonnet E4P Serial ATA PCI-Express card (also considered are the PCI-X version for the previous generation PowerMac G4/G5). This article goes over many variations of eSATA configurations that would cost a considerable amount of time and money to learn on your own. A must-read if you are in the market for high capacity and/or fast storage. I currently use Sonnet’s E4i for internal storage, having used their G5 Jive kit to mount an additional 3 Maxtor 3V300F0 drives inside my PowerMac Quad. I can attest that it’s a well-engineered kit, far better than the poorly-engineered garbage I tried from another company. CPU temps do go up with additional internal drives, but the Quad cooling fans stay as quiet as ever. More on this another day, perhaps. 07 April 2006
Nikon Capture and Digital Photo Professional on MacBook Pro In general, I am *very* impressed with my new MacBook Pro 2.16 GHz, 100GB/7200rpm hard drive, 2GB RAM. If you buy a MacBook like this today, how would it compare to the fastest Mac desktop machine today, the PowerMac Quad? (I no longer own my older PowerBook G4, so I can’t make that comparison). Why compare a Quad to a MacBook? Well, that's what I have available, and the Quad offers a performance level that is the best available on Mac OS today. Could a MacBook be in striking range of the Quad once these raw converters are offered in Universal Binary form?
(1) 30 Nikon NEF files from the D2X were chosen, averaging 11.7MB (compressed). Resulting
JPEG files (“Excellent” quality) ranged in size from 4.3 to 10.5MB each (7.9MB average). Bear in mind that the fastest disk speed possible on the MacBook, even with its 7200rpm hard drive, was only 44MB/sec (tested using disktester), as compared with 240MB/sec on my Quad with its striped RAID. That accounts for 1/2 second just for I/O on the MacBook, and the figure is likely to be slower, depending on disk access habits of these programs. A native version would not change that overhead. Conclusion—The MacBook Pro 2.16 GHz is within striking range of the performance of the PowerMac Quad once these applications are offered in Universal Binary form. I’ll go out on a limb and predict a speedup of 2X-3X for a Universal Binary version, which would yield times as fast as 7 seconds/file for a D2X NEF, and 10 seconds/file for a 1DsMII CR2. Not too shabby for a laptop! Of course, results could be disappointing too, but less than a 2X improvement seems unlikely. Incompetent programming— Of course, actual speed of Nikon Capture batch processing could be considerably better, if new versions were revised to use multiple processors appropriately (the Quad is grossly underutilized by both Nikon Capture and Digital Photo Professional). As a longtime software engineer who has done much performance work, I can say that both programs are poorly programmed in terms of using the available CPU power. This is evident in the CPU graphs shown below, where large amounts of CPU power go untapped on the Quad, averaging about 130% (out of 400%). Though the MacBook appears to be better utilized, in reality a good portion of that utilization is Rosetta (the emulation of PowerPC code on the Intel processor). The last two weeks were particularly busy for me what with tax season and certain family matters that sucked up much of my time. I expect to resume regular blog entries soon. So much for a 1,000,000-hour MTTF (mean time to failure) rating. One of my one-month-old Maxtor 7V300F0 300GB hard drives failed without warning, causing complete data loss on the RAID 0 volume to which it belonged. That drive was part of a batch of five. I'm crossing my fingers that it is not a bad batch—for you statisticians out there: what is the probability of a drive failure after about 200 hours of operation when the drive is rated with a MTTF of 1,000,000 hours? I would guess it’s about as good as my chance of winning the lottery. That’s why there are lies, damn lies, and statistics. And it’s also why one of my future articles will be about best practices for not losing data. Fortunately, I practise what I preach, and got a bit lucky, too—I had made two redundant backups 12 hours before the disk failed. So I lost only a few hours work (hey, I gotta sleep sometime). I bit the bullet, sold the PowerBook G4 1.33 GHz and bought a MacBook Pro 2.16 GHz. An “impressions” report will be coming soon, including such things as how fast it runs with Nikon Capture, Digital Photo Professional, and Aperture 1.1. Watch this blog. Gigabit ethernet, PowerMac G5, and CPU usage Do you share or copy files over gigabit ethernet? Perhaps you have a central file server and store your photographs on a server shared by multiple photographers? Perhaps you perform backups, as I do, over gigabit ethernet to another machine? If so, you should be aware of a little detail that won’t be mentioned in any marketing materials extolling the dual gigabit ethernet ports on the latest PowerMac G5 models. You might guess what that secret is by taking a look at the CPU usage graphs below (taken from a quad-core PowerMac G5 Quad):
The red area represents “system time”, or time that the operating system spends on behalf of programs. In this case, 90% or more of the red area represents CPU time spent doing nothing more than transferring data over a gigabit ethernet connection to another PowerMac G5 at the rate of 90-100 megabytes/sec (roughly 800 megabits/second). [Both machines have high-speed striped RAID disk arrays]. The red area represents roughly 100% of a core, or 25% of the available processing power on the Quad. Similar usage (100% of a CPU) may be observed on the PowerMac G5 (dual 2.5GHz/single core, which was the recipient of the data being transferred). I have also confirmed similar CPU usage on a dual CPU (single core) 2.0 GHz PowerMac G5. So it seems that the entire PowerMac G5 model line shares this trait (those machines cover all three generations of the PowerMac G5). Of course, if you have a single-core machine, anything involving high-speed network transfers will run as slow as molasses, since data transfer will be competing for CPU time with productive tasks. The utility of gigabit ethernet is severely restricted on such wimpy machines—don’t edit large image files over a network if performance matters. Imagine what would happen if both of the built-in PowerMac Quad gigabit ethernet ports were fully utilized—fully half of the available CPU power would be sucked up simply to transfer data! And that’s with the 4-core Quad—the mundane dual-core (single CPU) 2.0 and 2.3 GHz PowerMac G5 models not only have just two cores, but run at a slower clock speed too. They simply don’t have enough horsepower to do anything compute-intensive while a high-speed transfer is in progress. So stick to the Quad if your work involves high-speed transfers of large amounts of data with simultaneous demands on CPU time. If you envision using a PowerMac as a server (Apple sells MacOS X Server after all), don’t even think about using it in any application requiring high bandwidth without first testing combined network and CPU performance. A PCI ethernet card capable of offloading all the ethernet-processing overhead would be ideal, thus freeing the CPU to do real work. I don’t know if such a card exists for PCI-Express for Macs yet. Apple’s Technology Overview for the PowerMac G5 makes claims about “server class” features for the dual gigabit ethernet (see page 5 and 12). I haven’t figured out how to use “jumbo frame support” to “reduce system overhead” yet; perhaps that would cut down the system CPU time usage shown above. From what I can tell, such features are only supported within Mac OS X Server, but maybe there is a way to use them in plain Mac OS X. UPDATE: a reader informed me (thank you) that jumbo frame support can be enabled by manually configuring ethernet in the Network control panel in System Preferences (click to see a larger version):  I haven’t yet tested this, nor have I checked if my dual 2.0GHz G5 supports jumbo frames; if not, then perhaps another reader will be able to make that determination. Update: the dual 2.0GHz G5 does not support jumbo frames. Whither the Nikon D200 comparative review? Some of you may be wondering about my Nikon D200 comparative review—the good news is that the D200 is functioning just fine, and that I now have quite a bit of experience with it. The bad news is that Nikon has had my D2X since Jan 31, or six weeks, so I haven’t been able to make any comparisons with the D200 since then. If that’s the kind of service you can expect on a $5000 Nikon pro body, it’s no wonder so many pros have reportedly switched to Canon [that’s my impression at least—facts might prove otherwise]. Enough time has elapsed that I’m reconsidering whether to complete the review at all due to the shelf life of such reviews, it being already four (4) months since the release of the D200. For those readers who have patiently waited for the review, I’ll summarize by saying two things: (1) the D200 is very much a mini D2X with comparable image quality (though a pinch less resolution), and (2) I prefer the D2X unless size and weight are issues. Worth noting is that the new-fangled wide-angle-view screens on the D200, Canon EOS 5D and other cameras are inferior to the white-LED-backlit D2X LCD—still the best LCD I’ve seen. Uninterruptible power supply follies Last night around 2:10 am, my two APC Smart-UPS 1000XL uninterruptible power supplies switched onto their batteries as the power failed, and began to beep loudly 4 times each minute. This might be fine for a room full of servers, but it’s hardly friendly stuff for a home. Fortunately, there is an option to disable the beeping completely—and though the fans are far from whisper-quiet, they are not annoying in pitch. I discovered a few interesting things, which I’ll share here. First, it’s no fun figuring out how to disable the beeping at 2am. It’s better to do so at a more civilized hour. Second, and perhaps taking a design award for untempered stupidity, if you turn off the Smart-UPS 1000XL, and there is no line current, you cannot turn it back on! One has to wonder at what the “Dumb-UPS 1000” did—require sending it in for service? No need to believe me—read the manual. Any attempt to turn on the unit without line current just results in a short beep. I had turned off one unit because it was for equipment I wasn’t using, and even “off” that equipment was sucking 10-20 watts (“vampire devices”). I didn’t want to drain the batteries needlessly, so off it went (the other 1000XL continuing to power my server). Then I realized that I needed some information from the APC web site on how to disable the beeping, but there was a problem: the unit I had turned off would not turn back on (no line current). Brilliant design. There’s even a troubleshooting section in the manual describing the possible reasons the unit won’t turn on—including no line current. Isn’t this a battery-powered UPS? Fortunately, the server-dedicated 1000XL was still on, so I plugged the “off” unit into the “on” unit, powered on the “off” unit, and I was up and running. I haven’t yet called APS support to delve into this. Next up was disabling the beeping. The provided CD has three programs on it—but just try to figure out which one you must install in order to “talk” to the UPS to disable the beeping. The trick is to install all three, then poke around; the beep setting is cleverly hidden, but can be found, assuming you notice the tiny “Advanced” checkbox, which then makes additional settings appear (including the beeping thing). I didn’t know it was “advanced” usage to abhor annoying beeps. Apparently there is a “personal use” version, but APC wasn’t thoughtful enough to put that on the CD; perhaps they didn’t want to use up the other 600 unused megabytes on the CD. But I digress... You can’t use an ordinary serial cable to “talk” to the UPS; you must use the APC-provided one. This is fine, except that it’s about 6 feet long, and if your UPS is 12 feet away, it means you have to power off the UPS (see earlier discussion), disconnect any additional batteries, risk a hernia moving it next to the computer (60 pounds, and sorry, no Macs, only PCs), connect the serial cable, puzzle through the software (see above), then triumphantly disable the beep, power the unit down, reconnect and recable it, then power up whatever was originally connected to it. Wait!—it won’t start without line current, so get a 12-foot extension cord, plug it into the “on” unit, and power up the “off” unit. Whew! Really fun stuff at 3am. APC also provides a USB cable, but I was not successful using it, and it too is only about 6 feet long. Now repeat the process for the 2nd unit, but discover that the software is unhappy that you unplugged the cable and plugged it into a different unit. So quit the software and start it again. Enough grumbling, these units work great, right? Yes and no. The equipment connected to the 1000XL ran flawlessly. However, within a few minutes of power failure, a load of about 30-40 watts already had drawn down one unit from 5 lights to 4 (according to the charge lights on the 1000XL), and this is with an extra battery (the Smart-UPS XL 24V Battery Pack) attached to the 1000XL (manufacture date 11/06). In theory, a Smart-UPS 1000XL and an extra battery pack should be able to power a 35 watt load for 23 hours or so (APC web site claims 16:13 for a 50 watt load). I am now very skeptical, and will have to test that theory soon. The fans are noisy enough that they sound like they might eat 20 watts all on their own. Which brings me to the topic of fans, really stupid energy-eating fans—a 35 watt load shouldn’t require activating a powerful fan—surely a temperature-sensitive fan could have been engineered (incidentally the 1000XL was not even warm to the touch). All’s well that ends well—line current was restored about two hours after failing, and I got some sleep, though the 1000XL fans (both units) continued to run loudly until fully charged. And my server never went down due to power failure (though I did have to take it down while I disabled the beeping, because I had to move the 1000XL next to the PC, due to the 6-foot serial cable). A note on SATA Port Multiplication I briefly discussed the emerging technology of SATA Port Multiplication in my 01 March entry. It turns out that there is no free lunch—as wonderful as the technology is, it exacts a price—namely a 10-20% performance penalty (see “Five Bay Port Multiplier eSATA Enclosure” review at amug.org). In other words, a 5-drive port-multiplied enclosure (single eSATA cable) won’t perform any better than a 4-drive SATA enclosure using 4 cables, and maybe even a little worse in spite of having 5 drives to 4. The convenience of one cable is not to be underestimated—a single 4-port SATA card is capable of supporting 4 enclosures, or 20 drives using just 4 cables (as compared with 5 cards and 20 cables with “direct connect” to each drive. So if your needs are about huge amounts of storage, eSATA with Port Multiplication is a truly awesome solution. Fortunately, it is apparently possible to use a Port Multiplication-capable card in non-PM mode (one cable for each drive). If you are after both high storage capacity and top-end performance, then eSATA with Port Multiplication (PM) requires careful consideration. You might instead be better served by getting a conventional 4-bay/4 port SATA enclosure and using 4 cables. For a fixed setup, this will do just fine, provided you don’t need to attach more enclosures (which would require additional cards). Alternately, you might use two (2) PM enclosures with 2 cables (4 drives total), thus avoiding some of the PM overhead. I haven’t personally tested the latter, so I can’t say for sure if it’s effective, but it’s clear that PM performance penalties increase as you move from 2 drives to 5 on a single cable. Many informative and well-witten reviews at AMUG A site perhaps overlooked by most, amug.org (Arizona Macintosh User’s Group) releases a steady stream of detailed and informative reviews covering the Macintosh and many of the available storage products available for it, particularly external storage devices, such as SATA enclosures. Unlike the fluff reviews you’ll find in most major rags, the AMUG reviews give the lowdown on speed, noise, power usage, quality, etc—all the things you should know before making a purchase, with helpful photos as well. Look for the “Recent Reviews” blurb to the right. Recent reviews include the PowerMac Quad, the MaxConnect CPU Bay Drive Assembly, the FirmTek SeriTek 1EVE4 SATA Host Adapter, and a comparison of the Seagate 300GB/16MB St3300631AS vs the Maxtor MaxLine III 7V300F0 hard drives. I highly recommend joining AMUG. The membership price will easily be repaid by the informative material and discounts on excellent products from Wiebetech and FirmTek (so far this year). BareFeats.com reports on fast hard drives If you’re in the market for fast hard disk storage, be sure to check out the latest shootout at barefeats.com. The shootout includes the 300GB Maxtor 7V300F0, which I discussed in the 16 Feb entry, and also the 500GB Maxtor DiamondMax 11, the Hitachi 7K500 and the Seagate Barracuda 7200.9, all SATA II drives. Zeiss ZF (Nikon mount) lenses. Robgalbraith.com reports on the price announcements from Carl Zeiss for their new Nikon-mount “ZF” lenses (a 50mm/f1.4 and 85mm/f1.4 so far). In rough dollar terms, the lenses will be $600 and $1200 respectively. By comparison, Nikon’s autofocus 50mm/f1.4D is $269 and Nikon’s autofocus 85mm/f1.4D is $1015. Certainly the build quality is worth a significant premium, but at more than twice the price, and no autofocus, the Zeiss 50mm needs to demonstrate very nice rendering indeed. Though the “made in China” Nikon 50mm is cheaply built, I consider it one of my best lenses optically. We can only hope that the Zeiss 50mm offers improved sharpness, contrast and bokeh. The main issue I see is focus precision (and speed of focusing). The Nikon D2X offers such high resolution, that anything but spot-on focus will make any potential sharpness differences moot. This is already a problem, even with autofocus. My experience with most Nikon lenses is that the autofocus precision is less than perfect, leading to obviously-blurred results in some cases, though it’s better with fast lenses. Using the focus assist feature in the D2X and D200 helps with manual-focus lenses, but it’s just not accurate enough at f1.4 for precision work (try focusing 10 times and observe the variability in results on a high-contrast target, and a lower contrast target). Part of the problem is the vague coverage of the D2X and D200 focus sensors, even the center one. Canon is better in this regard. Zeiss trumpets the outstanding sharpness of their new lenses in the center. Gosh, a lens that’s sharp in the center?! As anyone who has read one of my comparative reviews knows, I make comparisons between similar items whenever possible, controlling the variables as feasible, thus allowing meaningful conclusions to be drawn. The Zeiss claims on resolution are not very useful for two reasons: First, the use of only the center area to assess sharpness is of little use in understanding the sharpness across the field. There are plenty of lens designs with outstanding center sharpness and fair corner performance. As any Canon 1Ds Mark II shooter knows, finding a lens that yields excellent sharpness across the field, especially wide-open, is no easy task (see D2X vs EOS). This is also true on the Nikon D2X/D200, because the 2/3 frame sensor demands high resolving power. Second, the test was performed on special high-resolution film. While no doubt there are still film shooters, it is likely that any of them are shooting the film Zeiss used (and film flatness and MTF is always an issue). For the vast majority shooting digital, the high resolution claimed by Zeiss may or may not be significant. The only useful comparison to be done is showing how the Nikon 50mm/f1.4D compares to the Zeiss 50mm on the Nikon D2X, Nikon’s highest-resolution digital SLR (alternately the two 85mm lenses). For that matter, Zeiss could demonstrate the claimed superior bokeh of their lenses by posting direct comparisons. But perhaps Zeiss feels that reputation, not facts, are sufficient to sell their new offerings. If you’re in the market for a manual-focus 85mm lens, the $1320 Nikon 85mm/f2.8D PC-Micro-Nikkor offers outstanding optical performance, while also offering superb macro and tilt/shift capability—a far more flexible investment unless you must have f1.4 and f2. Purchasing a new MacBook Pro? Laptop hard-drive speed. If you’re purchasing a new MacBook Pro, bear in mind that the achilles heel of laptop performance is slow hard disk speeds. You’re almost certainly better off getting a 7200 RPM hard drive (versus 5400 RPM) than a slightly faster processor speed. Here are numbers taken from a DiskTester run on my PowerBook G4 1.33 GHz 17" laptop with its 80GB hard drive (5400 RPM):
Write speeds are faster because of caching. The dip in write speed from 4K to 32K is due to inadequate drive cache (this is typical, even on desktop drives). Those numbers are the best possible—the inner tracks are usually only 50% of the speed of the outer tracks on most hard drives (a fixed linear bit density in a smaller diameter). Here is DiskTester’s report for an “area test” which measures speed across the drive:
Note the dramatic drop in speed as the drive fills up! There are two ways to mitigate this: (1) Buy a higher capacity hard drive, and avoid filling up the drive so that most data resides in the faster areas of the drive. (Mac OS X uses the faster areas first when you format a new drive). You can also partition the drive, reserving the slower partition for less frequently used items. (2) Buy a hard drive that is faster to begin with. On RPM alone a 7200 RPM drive should be 33% faster than a 5400 RPM drive (and it might also have a large cache). The new MacBook Pro models use internal Serial ATA (SATA) drives. While these almost certainly will perform better than older ATA drives, the same factors hold sway: a much faster dual-core processor, but only marginally faster hard disk speed means that anything involving disk use is going to run s...l...o...w...l...y. If you’re a photographer dumping multiple gigabytes of image files onto a laptop hard drive, save yourself the migraine and spend the extra $100 or so on a 7200 RPM hard drive. Today I built a Universal Binary of DiskTester, my hard drive (and/or RAID) speed and reliability testing tool. As I don’t yet have an Intel-based Mac, I can’t yet say how well it runs, but I expect to purchase an Intel-based Mac within the next month or so. After testing, DiskTester will be re-released as a Universal Binary. Updates will be free for prior purchasers. SATA hard drives and hard drive enclosures are evolving rapidly to become ever more useful. With its outstanding bandwidth of 3.0 gigabits/sec, SATA II is spawning some interesting storage products, including cards and enclosures with port multiplication. I came across an explanatory article today which will be helpful to anyone considering high-capacity external storage: Serial ATA Port Multiplier Technology. Any external storage unit I purchase will use port multiplication—I want one enclosure and one cable, with the ability to hook up more enclosures without having to buy more PCI cards—port multiplication allows that, all with outstanding performance. With Firewire a dead technology, can motherboard support be far behind? Blog continues: Jan/Feb 2006 |
|||||||||||||||
Return to diglloyd home page |
|||||||||||||||
© Copyright 2006 digLloyd,
Inc. All Rights Reserved. Reproduction in any form requires explicit, written permission. No part of this
document may be copied, distributed, disseminated in any form, electronic or tangible, whether in whole or
in part, without the prior written authorization of digLloyd, Inc or its authorized representative. |
|||||||||||||||
