|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Apple’s fastest Intel-based offering as of July 2006 is the MacBook Pro, with a 2.16GHz Intel Core Duo CPU. Around March 2006, I purchased a 15" 2.16 GHz MacBook Pro, and have been using it for the several months now. Post-purchase I did install an additional 1GB main RAM, for a total of 2GB, using the "Factory Approved" memory from satech.com, about $139 for a 1GB module. Help support free content Buying a Mac Pro or memory or anything at all (computer or not)? Help support free articles like this one by buying through amazon.com. This site earns a commission at no cost to you when you begin your shopping through the links on this page. Suggested search strings: Apple
Mac Pro desktop Processor 065-6458 2.16GHz Intel Core Duo/256MB My desktop computer is a PowerMac Quad 2.5GHz (8GB RAM, 3-drive stripe) , and many of the comparisons in this article are with the Quad —first, because that’s what I have to work with, and 2nd, because many people will want to know whether a MacBook Pro can be a viable replacement for a desktop machine. Further, the cost of a fully-loaded MacBook Pro is about the same as a PowerMac Quad (sans monitor), so comparing the two is reasonable from a cost basis. This will be an ongoing report. As I gain experience with the MBP, this page will be updated. Please watch the diglloyd blog for notes on updates, or ask to be added to the diglloyd email list. The latest revision date can be found after the copyright notice above. Please read my Overall impression for perspective before proceeding to the other parts. Contents Buyer's Guide: Configuring Your MacBook Pro Parallels virtualization software If you were to read some of the following sections without reading this one first, you might conclude that I dislike the MacBook Pro for all its shortcomings. That would be completely inaccurate—let me make it very clear: The MacBook Pro 2.16 GHz is the first Apple laptop I’ve ever used which I consider viable as a primary computer. I’ve used nearly all the fastest models from Apple as a software developer since the mid 1980's, so that’s quite an endorsement. The MacBook Pro feels fast. Even Photoshop in emulation [details] feels snappier than my old 17" PowerBook G4 1.33 GHz. While I no longer have the PowerBook G4, and so can’t make a direct comparison, I just can’t imagine going back to a G4 laptop. Buyer’s Guide to configuring your MacBook Pro The recommendations that follow are based on pricing and features as of 21 April 2006. First, the rumor mill alleges that there will be Intel price cuts on the Core Duo chip line by June 2006, so prices might drop $100-$200 on the high-end configurations and/or a 2.33 GHz model might be offered. Also rumored is that Apple might introduce a 17" model soon, though likely with very similar features to the 15" model. I have no first-hand knowledge; I’m just summarizing what I’ve heard from public web sites. If you must have a MacBook Pro, but the price is a serious issue, my recommendation is simple: get the 1.83GHz model with 1GB RAM (2 X 512) and 80GB hard disk for $2099. You will have an exceptionally fast machine that for ordinary tasks will feel just as responsive as the more expensive configurations. But review the table below to make sure you aren’t being penny-wise and pound-foolish—increasing the RAM later will mean removing one or both of the 512MB modules, and hard disk swaps are not for the faint of heart. The software you use influences which configuration is recommended. Here are examples of software for which a minimum of 1GB RAM and 7200rpm hard disk is strongly recommended:
While a few users will find that 512MB is enough, anyone using Photoshop, Aperture, raw file conversion software, virtualization software, etc will find even 1GB inadequate. Such users are strongly advised to go right to 2GB, and they’ll still wish they had 4 or 6GB! Another benefit of having 1GB or more RAM is that MacOS X takes advantage of any unused memory to cache items from the hard disk, both applications and files. For example, if you launch, then quit a program, launching it a 2nd time will be far faster. Here are some examples:
** PowerPC version, using Rosetta, a native version might show more improvement This shows how slow the hard disk speed is compared to memory, and supports the idea of opting for the 7200rpm hard disk (which my MBP has). When you purchase your MacBook Pro, I recommend starting with 1 GB RAM. It is about $100 cheaper to choose two (2) 512MB modules instead of one (1) 1GB module, and the two 512MB modules will be slightly faster. For many users, 1GB will be sufficient, so the slightly improved performance and $100 savings make sense. However, if you do choose 2X512MB, realize that to increase the RAM in the future, you’ll have to remove one or both of the 512MB modules to install 1 gigabyte module(s). If you remove just one 512MB module to add a 1GB module, this forms a mismatched pair, which works correctly, but causes slightly degraded processor performance (3-5%). Still, the $100 you saved by purchasing 2X512MB vs 1X1GB will offset the cost of upgrading. Memory prices and vendors Apple’s prices on memory are about twice as high as you can find elsewhere, such as satech.com. (Apple asks an exorbitant $500 to go from 512MB to 2GB on the 1.83 GHz model—buyer beware—another reason to start with the 2.0 GHz model). Whatever you do, don’t buy "el cheapo" memory (eg Fry’s Electronics grab-bin specials). Buy top-quality, non-refurbished "OEM quality" RAM. It will be a little more, but will save you many headaches and system crashes. There is usually a reason for the low price, such as remanufactured modules. Lower-quality memory might appear to work fine, but then fail sporadically, especially as the computer becomes hot. It’s just not worth the risk—if you’re really risk-averse, buy your memory from Apple when you buy your MacBook Pro—and keep in mind that adding more later at the Apple Store is even more expensive (and they won’t do it on the spot—I tried once). I’ve used satech.com as a source for memory for various PowerMacs and my PowerBook G4 and my MacBook Pro (Perhaps 30GB over the years). I had a bad module once, which they promptly replaced. They claim that: This is the same memory product you would buy from your system manufacturer (Apple, Compaq, Dell, HP, IBM, Gateway, Sony and others) directly when you buy the system. Verifying that statement is not something I can readily do, and of course it’s possible that Apple puts its memory through further quality control (or not). SATech.com's price for a 1GB “Factory Approved” module for the MacBook Pro is $139. Though the small sticker they use says “Samsung”, the memory chips on my module read “Elpida”. The modules are branded Excelerate, their own brand apparently. There are many vendors of memory for the MacBook Pro; it’s standard memory also used in Windows laptops. For example, OWC Technology offers 3 different 1GB modules, priced at $115, $120 and $179 which all state “Exceeds Apple/Intel Specifications”, the first two being “OWC brand” and the most expensive being Samsung (“same module Apple uses”). You might want to run an extended memory test using Apple’s Hardware Test after installing more memory, preferably after the machine has been used heavily for a while (heat stresses components). See the MacBook Pro User Guide, page 101 for details. Here are a few links for additional insight: [1], [2], [3], [4], [5], Like all laptops, the MacBook Pro is self-contained, with its own keyboard, trackpad and screen. However, the ergonomics of working with any laptop are rather poor compared to a desktop machine. I also find that the trackpad is tedious and slow as compared with a mouse. It took me about a week to adjust, but I’ll never go back to a standard Apple mouse—the Apple Mighty Mouse is terrific, with its ability to left and right click (ctrl-click) and other buttons/functions too. I always carry the Mighty Mouse with my MacBook Pro. I also greatly prefer a full-size keyboard. I use the standard Apple keyboard shipped with any Apple desktop machine. With a proper keyboard and mouse, things start to feel pretty much like a desktop machine. Although the MacBook Pro display is excellent as far as laptop displays go, ergonomically it’s Not Good. It’s only 15" diagonally, and if you set the MBP on the desk, you’re forced into a neck-damaging position (over the long term) as you stare down at it, and it also induces a hunched-over seating position. Neither is good. I used it for work one day, and had a back-ache after a few hours, not having connected it to my Apple 23" Cinema Display (read on). At 1440 X 900, the Macbook Pro display is not small, but it’s not particularly large either. It feels positively limiting in the vertical direction compared to a 1920 X 1200 display (such as the Apple 23" Cinema Display). If you have an extra display, either with a standard DVI (digital) or VGA (analog) input, simply plug it into the MacBook Pro DVI port and configure the two screens in “mirrored” mode (two screens showing the same thing) or “extended desktop” mode (a main screen with the menu bar, and the 2nd one with whatever else you like). Apple even supplies an adapter in the box so that an analog display can be attached; any display with a DVI input plugs in directly without an adapter. You don’t have to shut down, restart or even sleep the MacBook Pro; it just magically notices the screen and makes it work. Very cool. I like to attach the Apple 23" Cinema Display to the Macbook Pro as the main display, with the Macbook Pro built-in display used as additional working area for things like Mail, palettes in Photoshop, and Terminal windows. The MacBook Pro can even accept the humongous Apple 30" Cinema Display (and of course the 20", too). Any of these displays provide a viewing experience that is both easier on the eyes and the physique than the built-in display. You can attach any monitor with standard DVI input (not just Apple monitors). Once connected, arrange the screens as desired by dragging them into the desired relative positions. You can also drag the menu bar to either display:
Hard disk speed and expansion Although I opted for the speediest hard disk, the 7200rpm 100GB model, its top speed under optimal conditions (fast part of disk, large transfers—see disktester) is only about 45MB/sec, with a range of about 22-45MB/sec. Compare that to a 7200rpm desktop drive like the Maxtor 7V300F0 on the Quad [details] , which achieves speeds from 45-75MB/sec, or roughly twice as fast (and with a 16MB drive cache too). Further, the 4-drive striped RAID on my Quad offers performance of roughly up to 270MB/sec (and the 8-drive up to 500MB/sec). Thus, any task involving significant amounts of disk I/O puts the MacBook Pro at a severe disadvantage. It’s too bad that Apple crippled the 15" MacBook Pro external storage options with an aging Firewire 400 port (half the speed of Firewire 800). The MacBook Pro 17" model does have a Firewire 800 port (and is $100 cheaper with my configuration). There is no legitimate reason that the 15" model should use a Firewire 400 port instead of Firewire 800. I am not happy, because the Firewire port is what I have always used in the field for transferring multiple gigabytes of image data, backing up, etc. This is not the first time Apple has shortchanged early adopters—caveat emptor. The good news is that in July of 2006, FirmTek released their SeriTek/2SM2-E ExpressCard card for the MacBook Pro, with two exernal 3 gigabit SATA ports. See the results of that setup. DiskTester speed tests—Firewire and USB For the first test, I tried a Maxtor 500GB DiamondMax 11 in a WiebeTech ToughTech 800 case on the MacBook Pro. I tested both its Firewire 800 port, using a Firewire 800 to Firewire 400 cable, and also the USB 2.0 interface. The Maxtor DiamondMax 11 itself is capable of 60-70MB/sec top speed. The DiskTester results shown below demonstrate the miserably slow performance with either Firewire 400 or USB 2. They also offer proof that performance claims based on technical specifications are worthless; the claimed 480mbps of USB 2.0 is grossly inferior to the “slower” Firewire 400 speed. The PowerMac Quad using Firewire 800 is shown for comparison, with performance about 70% faster than the MacBook Pro Firewire 400. MacBook Pro 15" 2.16 GHz [config], Firewire 400/800 cable DiskTester speed tests—SATA via FirmTek ExpressCard adapter In late July 2006, I purchased the SeriTek/2EN2 Dual-Bay Hot-Swap Enclosure bundle with the 2SM2 adapter. The two can be purchased together as the SeriTek/2ENSM2-E External Bundle. You’ll need to supply your own SATA drives. This combination offers the best disk drive performance available on the MacBook Pro, so far as I know. FirmTek generally ships same day, and I got my unit the very next day via ground shipping. I’ve ordered from the 4 or 5 times, and they are always prompt—a great company to do business with. The SeriTek/2EN2 External Dual-Bay Hot-Swap Enclosure uses high quality aluminum construction with no internal cables, using a more reliable direct-connection to the rear of the SATA drives. An external power supply keeps heat out of the enclosure, and a 3-speed fan (user selectable) keeps fan noise to a minimum. The enclosure is petite, measuring only 9 X 4.75 X 2.75 inches, small enough to consider for portable use in the “field” (eg a vehicle). The power brick is fairly large—5 X 1.5 X 2.5 inches. Still, this is about as small as it gets for dual-drive capability, and the performance is second to none on the MacBook Pro as we shall see. The SeriTek offering is my top pick if you want top-notch disk performance on the MacBook Pro. It just brings all the right things together: size, quality, noise, performance, price. I installed two Maxtor 7V300F0 drives, the same model drives used in my PowerMac Quad RAID setup. Next, I created a striped RAID 0 pair (single volume). I then ran an extended “area test” using DiskTester 2.0, with the results shown below. The results below show a stunning performance advantage over any internal disk (at best 45MB/sec with a 7200 rpm drive) or even a Firewire 800 setup (see results above). Performance is 40% faster than even the fastest theoretical Firewire 800 performance. I have tried striped Firewire 800 2-drive RAID 0, and the best you’ll see is about 80MB/sec. Note also the write speed is pathetic with Firewire 800. These results are for sustained throughput. The 16MB cache on each drive (32MB total) also speeds up real-world use even more over the internal hard disk, which has only an 8MB cache. The results also show clearly how speed drops from the outer (fastest) tracks of the drive to the inner (slowest) tracks of the drive (click for larger graph). Note that even at 100% full (slowest part of the drive), the 2-drive stripe is still 73% faster than the absolute fastest possible with the internal hard disk! This is why for a fast scratch disk for Photoshop you should consider making a small partition (16-32 GB) on the fast side of the drive, and use the rest of the space for your main data storage volume. disktester run-area-test --chunk-size 128M --iterations 5 --delta-percent 5 --test-size 2G S2 
------------ Averages for "S2" (2GB/128MB, 5 iterations) -----------
Area (556.5GB) Write MB/sec Read MB/sec
0% 140.6 135.8
5% 140.9 136.0
10% 137.2 133.3
15% 136.8 132.7
20% 133.5 129.9
25% 130.6 127.1
30% 130.6 127.1
35% 126.9 123.6
40% 126.8 123.4
45% 123.7 120.4
50% 119.3 116.0
55% 119.2 116.1
60% 115.3 112.4
65% 110.5 108.1
70% 101.6 99.4
75% 105.8 103.6
80% 101.2 99.0
85% 96.4 94.6
90% 91.1 89.4
95% 85.0 83.5
100% 78.2 77.0
Average write speed across the volume: 117MB/sec
Average read speed across the volume: 114MB/sec
To see the entire output from DiskTester 2.0, click here. Hard disk speed aside, which will soon have various workarounds, memory expansion is the single biggest limitation of the MacBook Pro (and all previous PowerBooks for that matter). Two (2) GB, once considered an enormous amount, is barely enough to run Photoshop CS2, Digital Photo Professional and/or Nikon Capture, and perhaps Mail and a few odds and ends. Coupled with an internal hard disk that is 1/2 the speed of a desktop model (not unique to the MBP), the MacBook Pro imposes a “ceiling” on possible applications not imposed by robust desktop machines, such as the PowerMac Quad. Editing 200-300MB layered Photoshop files is completely out of the question—even on my Quad, CS2 starts hitting the disk after it has eaten the maximum RAM I can give it—3 GB for CS2 itself. On a MacBook Pro, running nothing else, you can probably manage the better part of 2GB—after that it’s paging to the slow internal hard disk. What a shame that 4GB can’t be put into the MacBook Pro; it otherwise could become a viable do-everything machine for some users. If you need more memory for your MacBook Pro, satech.com is a good source. Buy the OEM-quality memory, not the cheaper stuff. Tell them diglloyd.com sent you (diglloyd.com receives no remuneration). Another option for fast I/O is using a network server with a suitably fast disk, such as a 2 or 3 drive striped array. I observed network transfer rates (real throughput) as high as 100 MB/sec while copying files from my PowerMac Quad to the MacBook Pro. Unfortunately, to get that kind of speed, about 110% of the CPU cycles (out of 200% max) are utilized just to transfer the data (see Gigabit ethernet, PowerMac G5, and CPU Usage). Apple has yet again crippled potential performance by using an ethernet implementation that uses excessive CPU cycles, though for most users this will be of little concern. The Apple desktop machines, including the Quad, are also afflicted with this problem (as are most Windows PCs without specialized ethernet cards). On the plus side, if you take the time and effort to setup a network that supports jumbo frames (9000 bytes vs the standard 1500 bytes), the MacBook Pro does support jumbo frames, and using them can cut the CPU usage in half (roughly), according to what I’ve read (no personal experience there yet). Pun intended for all you digital camera folks. More on RAW-file conversion may be found in Digital Photo Professional Batch-Processing Tip, and see also Nikon Capture. By “raw computing power”, I mean the raw performance of the Intel Core Duo cores, excluding disk access and graphics, but including memory bandwidth. Memory bandwidth is important to programs like Photoshop, which must move large amounts of data around in memory. it is much less important to programs that primarily compute (such a mathematical or statistical modeling programs, or a program calculating π). Most applications benefit from both, with raw computing speed generally dominating the performance. With today’s fast processors, some programs run little faster, even at twice the clock rate, because the processor literally sits and waits for data to be shuttled to/from memory. That’s why with every generation of computer, you see newer and faster memory, making the old stuff useless in the new machine. I did not test memory bandwidth, but the memory used by the MBP should be quite fast, even compared to the Quad. As one measure which is quite relevant to my workflow for making backups, I measured the ability of the MacBook Pro and PowerMac Quad to perform the cryptographic SHA1 hash, used heavily in the soon-to-be-released diglloyd IntegrityChecker™ program. I tried some tests using the compression utility 'gzip', but was unsuccessful in achieving significant CPU usage because the internal disk speed is just too slow to provide data fast enough. This is a key factor that will severely limit any application that requires substantial amounts of disk access—those fast Intel Core Duo cores will stand idle much of the time, waiting for data to arrive from disk. As yet, there is no solution to this CPU-starvation problem except a server with gigabit ethernet capable of delivering data at about 110 MB/sec. But that comes at the cost of eating 110% of the CPU cycles, thus defeating the whole point. I hope to add more CPU tests, but finding a real-world application that I use which doesn’t need significant amounts of data from the hard disk has stymied me so far.
The display is bright, but shows non-uniform brightness, making it unsuitable for color or density judgments other than basic ones. The picture below was taken using a Nikon D200, and gray balanced in Nikon Capture. In “real life” the screen looks terrific—the picture shown here is not what you would perceive without a trained eye. Uniformity is just fair—with noticeable darkening in the corners—certainly nothing you’d want to do critical color work on, but about what can be expected from a moderately good laptop screen. For any use besides correction of digital photos, the screen is excellent.  I also calibrated the screen using Apple’s software in the Display control panel. This is the first LCD I’ve used that was noticeably discolored using the “native” screen settings. I had to add a fair amount of color adjustment to get a neutral gray. The number of pixels is the same as my old 17" PowerBook G4 (1440 X 900). I vastly prefer the 15" form factor, and will stay away from 17" models. At any rate, it is preferable to connect the MBP to an Apple 23" display than to use the built-in display (but beware of the idiotic cable arrangement on the Apple 23" Cinema Display, which makes it difficult to place the MBP in a convenient location). As with nearly all laptops, the keyboard and trackpad are built-in. Both are inferior to their desktop equivalents, so if you’re using the MacBook Pro in a fixed or semi-fixed location, invest in a full-sized keyboard and the indispensableMighty Mouse. It took a bit of getting used to, but I now dislike using a Mac without a Mighty Mouse. Apparently the “upgrade” from 60 watts in the G4 series to 85 watts in the MacBook Pro required a much larger power brick. The white power brick (transformer) has just about doubled in size, growing about 60% linearly (a rough guess), and is now annoyingly large and heavy compared to the power brick used by the former G4 models. Still, it’s far better than many I’ve seen, and at least it’s flat. I measured power usage using a Watt’s Up? power meter, which claims it measures true RMS power. It’s well-built and appears to be a higher quality unit than some of the other ones out there, but I have no way of verifying its accuracy.
Play a movie I inserted The Lord of The Rings—The Return of the King and got about 2 hours 11 minutes of playing time before the MacBook Pro shut itself off. This was with "Reduce the brightness of the built-in display…” unchecked, so that the screen was at full brightness. Full throttle disk whacking, 200% CPU use After a full battery drain and recharge, I used the diglloydTools 'run-stress-test' command to determine battery life under maximal battery drain. This test includes:
It is virtually impossible to generate this severe a load in any sort of normal usage, so this is the absolute worst case scenario. The MacBook Pro became quite hot by the end of the test, but not unreasonably so, considering the severe load under which it had run. At 60 minutes the “.” dialog appeared. The battery meter read 12% remaining. At 70 minutes, the MacBook Pro shut itself into deep sleep. The battery meter read 0%. See Power Usage for a note on the size of the power brick. Fan noise is more prominent than on my old PowerBook G4 1.33 GHz, with the fan running almost continuously, even under light usage. While it’s not very loud, the sound of silence is definitely missing. On my unit, the buzzing sound some users have reported seems to occur when the MBP is running on battery power, and does indeed seem to be load-dependent. I haven’t found it to be a problem, but I usually have the MBP running on wall-current. MacOS X is a stable, fast and robust Unix-based operating system with the most usable and most attractive user interface available today. Compared to the tacky and gaudy and awkward Windows world, it stands head and shoulders above it. Boot time Boot time on the MacBook Pro is very fast—mine boots in 20 seconds, measure from the “chong” sound to the time the login screen appears. Of course, most of the time you can use sleep mode, making it available in about 3 seconds. Viruses I get 400 spam emails per day, a few containing viruses (why isn’t repeatedly sending spam a felony?). I also occasionally get the particularly annoying emails from those who have my email address in their address book, and who clearly have been infected by a Microsoft Outlook and/or Windows virus that sends itself out in email to everyone in that computer’s address book. I have never run anti-virus software on Mac OS X, and indeed (to my knowledge) there are no self-propagating viruses “in the wild” that have been reported to affect MacOS X itself, though there are a few scattered trojan horses and Microsoft Office/Outlook viruses running around which do affect Mac users. The situation might degrade with the recent introduction of Intel-based Macs, but the operating system (Mac OS X vs Windows) is largely the issue for viruses, less-so the CPU chip. The problems with anti-virus software are (1) it costs money, (2) ongoing maintenance/updates are required, (3) it can cause slowdowns or even system instability, and (4) it isn’t perfect. With Windows, anti-virus is an absolutely necessity; with MacOS X it’s not. Good habits go a long way towards minimizing the chance of getting a virus. The Microsoft Windows world is a jungle swamp teeming with all sorts of nasty things, whereas by comparison the MacOS X world is a nearly-sterile stainless-steel lab bench. If you value your data or even your money (online banking, stocks, etc), accepting the risk of Windows viruses is a fool’s game, even with anti-virus software. There are simply too many dangerous viruses out there, and too-frequent introduction of new ones. If you must run Windows, run it as an alternative using software such as Parallels or Boot Camp. Viruses are of course possible on MacOS X, and it does have security vulnerabilities, but that’s a far cry from actual “in the wild” viruses and/or “exploits” than can self-propagate and cause serious data loss. There are a few “trojan horses”; if you are naive enough to download and run random software, sooner or later you might regret it. All computing platforms are subject to “phishing”, so be extra careful when receiving email—and never click on links within emails unless you are 100% sure they are legitimate (or choose a rule that you never do). A corollary is to never “verify” your password or personal information, etc. Such requests are always ruses. If a legitimate company were ever to ask you to verify your information via email, stop doing business with them—such stupidity deserves a vote with one’s “wallet”. Digital Photo Professional [software update page ] Please see the Digital Photo Professional Batch Processing Performance Tip article for details, including both MacOS X and Windows XP results for Digital Photo Professional. Hint: the MacBook Pro beats a dual 2.4GHz Opteron with a 2-drive 10,000rpm striped RAID. See also various items in the diglloyd blog: [1], [2], [3]. Nikon Capture [software update page ] Could a MacBook be in striking range of the PowerMac Quad once Nikon Capture is offered in Universal Binary form? [See the 07 April 2006 entry in the diglloyd blog for my prediction and the 09 April followup]. The WindowsXP-SP2 numbers below show tremendous potential once Nikon Capture becomes a Universal Binary. 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. The results are astonishing—the MacBook Pro is nearly a match for the PowerMac Quad with Nikon Capture! Of course, a well-written application using all the available CPU cores would push the Quad well-ahead in performance, but you get what you’re getting, and that’s all your getting. Making the results all the more believable, the relative speeds are consistent with those obtained in testing Digital Photo Professional.
(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. In the Windows XP test, the Windows partition is on the slowest part of the hard disk, capable of perhaps 25MB/sec under optimal conditions. Total I/O per file averaged 19MB for this test (reading the NEF, writing the "Excellent"-quality JPEG). Thus, at least 7/10 of a second was I/O; the same speed hard disk as the Quad might well mean the MacBook Pro could have completed the test in about 5.1 seconds, faster than the Quad. That is impressive indeed. A native version would not improve matters with I/O—the I/O would become a larger proportion of the overall time, which means that using the fastest possible hard disk is imperative for enjoying the full potential of the MacBook Pro. That means an external hard drive, preferably a 2 or 3 drive striped RAID, not the relatively sluggish internal one, not even the 7200rpm premium option, as used in My MacBook Pro. I’ve only just installed Aperture, so have little to yet report. Installing Aperture 1.0, then launching it results in the message “” []. However, you can still use Software Update to update it to Universal Binary (Intel-native) version 1.1, at which point Aperture complains no longer. Memory usage after startup is only 33MB real memory, 428 virtual, quite reasonable. UPDATE: 17 Apr 2006 Importing is poorly implemented in terms of performance, using scarcely 1/4 of the available CPU power (Quad), even on my striped RAID which is capable of 200MB/sec or more, so disk speed is not the issue—implementation “smarts” is the issue. JPEG conversion—I used the same 30 Canon EOS CR2 1DsMII raw files as in my Digital Photo Professional Batch Processing Performance Tip article and run a batch “Export” test on the MacBook Pro and PowerMac Quad. The results are substantially faster than Digital Photo Professional, but disappointing in that ample CPU cycles go unused. Why buy a Quad if the cores don’t get used? I expect to see 400% CPU usage on the Quad from a well-designed application, assuming the disk speed is up to snuff. I also tested the Quad with 2 cores disabled (CHUD tools, Processor Prefs), to make it a fair fight (2 cores each).
It’s puzzling that the CPU usage drops when going from 4 cores to 2 on the Quad. I had expected to see it stay at about 190%, same as with 4 cores. The MacBook Pro impresses. Never before has an Apple laptop come this close to the speed of a top-end desktop machine. But unlike other tests, it does not match the Quad, even with 2 of the 4 Quad’s cores disabled. Perhaps there is some processing going on with the video card that accounts for this. Adobe Photoshop CS2 performance [software update page] How does the speed of Adobe Photoshop CS2 on the MBP compare to a PowerMac Quad (2.5GHz, 8GB, 3 X 7200RPM striped RAID)? You can easily forget that the MBP runs Photoshop CS2 in emulation using Rosetta. Photoshop performance on the MacBook Pro seems quite responsive, so much so that performance is not likely to be an issue for most users—but beware inadequate memory, which could seriously degrade performance. Anything that causes disk access to the sluggish MBP hard drive is going to seriously degrade performance. Don’t even think of doing any serious photo work without the full
complement of 2GB RAM in a MacBook Pro. Apple needs to get its sh*t(1) together and
make up to 4GB RAM possible—at that point the MacBook Pro gains a comfortable margin of free
memory for many applications. Launch time and memory usage Disk speed can be a limiting factor for some operations, and CPU speed for others. Launching Photoshop was the first test. A launch/quit/launch cycle was done, timing from the launching click to the time the palettes were fully drawn. The disk speed on the MacBook Pro appears to be a major factor in the launch test, though I didn’t verify it. Here are the results: Launch on Quad (1st/2nd): 25
/ 3.5 seconds Memory usage on Quad (real/virtual): 74MB
/ 571MB Memory usage on Quad with 1 file (real/virtual): 114MB / 679MB Memory usage with CS2 on the MacBook is substantially higher. This might be due to Rosetta. Regardless of cause, higher memory usage on the MBP is troublesome, given that the MBP can accept a maximum of 2GB of physical memory. Photoshop action on 30 files Next, I created a Photoshop action and ran it on a batch of 30 large JPEG files (7.0 - 16.7MB each), created by processing Canon EOS 1Ds Mark II files with Digital Photo Professional. The script used is shown below. I did not include a “Save” as part of the script, because I felt that the inclusion of both an “Open” and a “Save” would skew the results too much towards measuring disk speed.
Parallels Workstation (Windows on MacBook Pro) Please see the diglloyd blog 11 April 2006 entry. The MacBook Pro 2.16GHz easily bests two 2.5GHz PowerPC cores for builds**— I used “Processor Prefs” from the developer CHUD tools to disable some of the Quad’s cores so that it was 1/2/4 core machine. I ran the tests several times and got consistent timings. Here are the timings building a small XCode project: PowerMac G5 Quad, 1 core: 141 sec This is just amazing—never before has there been an Apple laptop that could come anywhere close to matching the top of the line desktop machine for software development. The Quad won only when it could use all four of its cores. Otherwise, on a per-core basis, the Quad was 16%-30% slower. This is one more data point consistent with other tests showing the two 2.16 GHz Intel Core Duo cores frequently will outperform two 2.5GHz PowerPC G5 cores! ** Conclusions The MacBook Pro has been a big step up from my 17" PowerBook G4 1.33GHz. It already is a better performer, even in emulation, than my former PowerBook G4, including Photoshop performance. Things can only get better as more Universal Binary applications emerge. The MacBook Pro 2.16 GHz is the first Apple laptop I’ve ever used which I consider viable as a primary computer. However, as with most Apple offerings, form can take precedence over function. Clearly Apple invests in first-rate industrial design, but just as clearly the same effort is not put into banishing annoying limitations, such as slow hard disk speed with no viable expansion options for better performance (at this time), an aging Firewire 400 port instead of Firewire 800, and memory expansion limited to 2GB, severely handicapping its usability for graphics work that requires multiple gigabytes of RAM. More attention could also be paid to the screen, whose density varies somewhat, especially in the corners. Nevertheless, this is the best laptop Apple has yet produced. Highly recommended for those who need a laptop. Contact: email diglloyd a comment on this article.
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
, 
