Previous Issue | Search TidBITS | TidBITS Home Page | Next Issue

TidBITS Logo


This week we look more closely at component level repair and whether or not it is reasonable to expect Apple to do it, report on a deal from Connectix that, in an unusual move, is only open to users of online services, and present much-awaited benchmarks on the just-released machines, again from Tom Thompson and BYTE Labs. Also, those of you on the Internet can check out parts of "The Internet Companion," available via anonymous FTP.


Copyright 1993 TidBITS Electronic Publishing. All rights reserved.
Information: <> Comments: <>


Lots of people have asked us where they can get the 32-bit Enabler and the Macintosh Hardware System Update. As yet, they have not appeared where the net public can find them. We have received contradictory information about whether or not these programs have even been released, so as a result, our advice is to sit tight for a bit and not worry.

Online Books

Fred Berg writes:

Parts of "The Internet Companion" by Tracey LaQuey and Jeanne C. Ryer, which was reviewed in TidBITS-164, are available via anonymous FTP from <> in the directory: </OBS/The.Internet.Companion./>.

Barry Shein adds:

Further chapters will be released in the future. See the README and COPYRIGHT files in that directory for more details. Direct comments and questions about the book to:

This pioneering effort is a step in bringing together the online electronic and print media, enabling authors to explore new avenues of publishing their works. Comments and inquiries are welcome via email to <>.

Information from:
Fred Berg --
Barry Shein --

Connectix Online Offer

Connectix, makers of Connectix PowerBook Utilities (CPU), Virtual 3.0, MAXIMA, Hand-Off II, and the new InfoLog, is offering lower than normal prices to online services users. Since you're reading TidBITS, there's a good chance you fit that category. The offer involves ordering direct from Connectix, so the prices are excellent - noticeably better than mail order (I found a few mail order prices for comparison in the table below). Add $4 for shipping in North America, $10 for international orders (for one or more), and if you're in California, add the 8.5% state sales tax. Although international orders are fine, Connectix cannot accept orders from all countries due to agreements with local resellers, to whom Connectix will refer you if necessary.

     Product                               List  Mail  Online
     Connectix PowerBook Utilities (CPU)    $99   49     39
     Virtual 3.0                             99          39
     MAXIMA 2.0                              69   45     29
     InfoLog                                149          44
     HAND-Off II                             99   55     39

If you are on AppleLink or America Online, Connectix has online order coupons you can download and redistribute. On AppleLink, check for AppleLink -> Third Parties -> Connectix -> Connectix "On-Line Coupon" and on America Online, look in a new Connectix Forum opening next week in the Macintosh Utilities Forum. It may also be worth checking on Connectix's forum on CompuServe in the Macintosh A Vendors forum (MACAVEN).

If you can't get a coupon, send your name, organization, street address, city, state, zip, and phone number; email address and service (I imagine "Internet" is fine for most of us); and credit card type (Visa, MasterCard, or American Express), number, and expiration date to Connectix at an electronic address below. Sorry to list all that out like that, but I have confidence you can figure it out.

Connectix says it will take about 10 days for delivery and the offer is void where prohibited (I like that phrase. It reminds me of a sign I saw once that said "Illegal activities prohibited." What, and they're OK everywhere else?). Finally, all of Connectix's products have a 30-day, money-back satisfaction guarantee. I approve. You can order one of each product per email address, and the prices expire 30-Jun-93.

You can order these products via email or, if you prefer (and we don't) via fax. As much as it's a bit clumsy to order this way, you can see how much money goes into the software distribution channel. If more companies conducted business online, we'd have cheaper software and less packaging waste since the companies would be interested in shipping smaller packages, not in creating hefty boxes to look good on shelves. Connectix, by the way, is good about packaging if the copy of Virtual 3.0 I just received is any indication. The disk is in an envelope inside the slim manual, and the whole thing is shrink-wrapped. No box, little waste.

And to preempt comments, yes, I know that if we rid ourselves of the software distribution channel that dealers won't be able to stay in business and provide the tech support users require (although Roy McDonald of Connectix made a point of telling me that Connectix provides toll-free support). In this instance, I fall back on running dog capitalist theory and say that the market would adapt.

Some people have legitimate concerns about sending credit card numbers through email. I expressed that concern to Roy, who said they set up the fax for this reason, and so far they have received three times as many responses via email as via fax. I'm glad that email is beating fax, but I'd like to see (and if I get bored someday I may write it) a HyperCard stack or application that has fields for the relevant data, does error checking, and then writes data to an encoded text file. The coding scheme could be as simple as a slightly modified rot13 (take the ASCII value of each letter, add 13, convert back to a letter, and repeat), but it would ensure basic privacy. There's no point in trying to protect that number too hard - if people want to steal it they will anyway and you will have to rely on your bank for help. The other value of this program would be to allow automatic data entry of the orders with a moderately intelligent mail system.

Send your email order to:
415/571-5195 (fax)
800/950-5880 (voice)

Information from:
Connectix propaganda
Roy McDonald, Connectix --

Apple Repair, Continued

Just as our modem issue garnered many comments that you'll see in a future issue, so did Fred Condo's open letter complaining about Apple's repair policies. People brought up many excellent points about why Apple doesn't officially do component level repair, which I'll try to represent here.

As Keith Bourgoin <> wrote, Apple is essentially saying, "We don't make those CD-ROM drives, so we can't get the individual parts even if we wanted them." Think of it like this: Sony builds a CD-ROM drive for Apple and stamps an Apple logo on it. That drive doesn't even necessarily appear in Apple's factories. A glaring example is the Apple Color Printer, which goes straight from Canon's factories to Apple's warehouses. Apple never gets the individual parts that make up that CD-ROM drive or color printer. Thus, when Apple provides repair parts, they provide a "unit" from Sony or Canon.

In that sense, Fred's specific complaint isn't Apple's fault, but Sony's. And, as Jeff Provost <> wrote, getting through to Sony's parts department is an exercise in frustration. If you connect to a large company's parts department, as Bill Coderre <> said in another mailfile (and to be clear, Bill says he is not the Official Apple Voice Of Truth[tm]), they may refuse to sell you a single gear. A bag of 50, sure, but since the company Bill was talking about wouldn't sell anything for less than $50 for bookkeeping reasons, the customer had to buy a bag of 50 gears. I guess that's better than buying a new widget, but not much better.

That answers the question for Apple-branded equipment that Apple doesn't make, but what about motherboards, which Apple does make? Someone wrote to us with a horror story about how a picture fell off his wall and hit the SCSI port on the back of his IIci, pulling out one of the threaded holes for the SCSI cable screw. He said the part didn't appear integral to the motherboard, but he had to replace the entire board to fix it.

The problem is threefold. First, Apple uses a just-in-time inventory method that tries to ensure when a Mac is ordered, it's built right then and shipped out. That's over-simplifying, but the idea is that Apple doesn't have a large inventory of raw parts or completed machines, ever. A large quantity of tiny repair parts in the dealer channel would cause inventory problems, tracking problems, and accounting problems, and as much as it's fun to complain, that costs money, which would raise the cost of Macs. Second, more and more of the parts on Apple's motherboards are custom designed for Apple, which means that the manufacturers only make as many as Apple will use. We're not talking the original IBM PC here, which IBM created with almost entirely off-the-shelf parts. Third and finally, if Apple made these parts available to repair centers, they would have to ensure that repair centers had people skilled enough to do component level repair. Would you trust your dealer to have someone that skilled? Some yes, some no.

Also, a company may not make parts available because that encourages untrained people to open equipment and try to fix it. That's all fine if the equipment is old, but what if it's under warranty? Does the company honor that warranty even though your soldering iron slipped and melted a hole in the controller card?

William Humphries <> passed on an interesting note. Apparently Kodak lost an antitrust suit filed against it by a group of frustrated repair centers that could not get parts. William didn't have the original suit, and I haven't found it, so I don't know if the situation is similar.

Finally, our friend Oliver Habicht <> from Cornell expressed an interesting viewpoint. As Oliver sees it, the problem is similar to the question of whether to repair or replace a broken VCR. When a mechanical system wears out, it stresses other parts of the system, and in some cases, like on a bicycle derailleurs, the parts wear together. Thus, when one breaks, it's a sign that more will break soon, and often when repairing a VCR, the technician replaces related, weakened parts. On a bike, if you replace one part of the derailleur system without replacing other parts, you may have trouble because the new part and the old ones aren't worn in the same ways and stress each other differently (and from experience, they'll make miserable whining sounds until they've worn enough so you can adjust them correctly :-)).Thus, the decision to not provide repair parts may be related to the likelihood of a successful long-term repair.

Of course, this leaves open the possibility that independent people could do component-level repair. These people do exist - a guy we knew in Ithaca would fix a broken part rather than replace the motherboard, usually cannibalizing parts from dead Macs. He was popular, and if any of you enterprising electrical engineers need a job to tide you over in this tired economic climate, think about board-level repairs. Have enough saved up to replace a motherboard should you toast it, though. We would like to see more independent repair shops that could do this, and would ask only that Apple not hinder such independent people, although for the reasons outlined above, it would be reasonable if Apple did not offer additional help.

I hope this explains the many reasons and views for why Apple doesn't provide low-level repair parts, because this policy will continue. With something like the Newton, there will be no repair - if it breaks, you'll get a new one, for free if it's in warranty, for a fee if it's not.

Postscript -- Fred Condo <CONDOF@CGSVAX.CLAREMONT.EDU> passed on the rest of the story of his CD-ROM repair quest:

Six weeks after I took my Apple CDSC CD-ROM drive in for repairs, and with the help of the Internet community, my local technician found the right Sony division to sell him the $5 gear. It cost $5. His labor cost $65, for a $70 total. That this is preferable to Apple's $500 module swap is, I should hope, self-evident. In an installation whose CD-ROM drive was an economic necessity and not something of a luxury, my decision would have been an instant replacement of the drive with something from NEC. Sony and Apple have lost my peripherals patronage until this policy changes. They have been publicly chastised here on the Internet (my email is running 100% anti-Apple on this). Surely this loss of good will must count against any fiscal savings Apple achieves by their policy.

[Just don't expect decent service from NEC - most NEC tech support horror stories involve explaining Mac basics to the technician. I don't believe Apple and Sony stand alone in this sort of policy, since I remember reading of a journalist's equivalently-convoluted quest for a new ball for his IBM PS/2 mouse. -Adam]

I will continue to buy Apple's wonderful computers whenever I need a computer. I won't buy any of their peripherals unless they make a commitment to repair the things at a reasonable price after the sale.

[Fred offered rebuttals to the explanations as to why Apple cannot officially handle component-level repair, but we cannot spare the space. Suffice it to say that Fred feels that Apple could overcome the objections if they wished. Apple, or at least the people who responded to Fred's original letter, feels differently, so for the moment we will have to agree to disagree, and we can all individually decide if we wish to consider how companies handle this issue in our purchasing plans. -Adam]

February Mac BYTE Benchmarks

by Tom Thompson --

BYTE Senior Tech Editor at Large

According to preliminary BYTE low-level benchmarks, the new Macs introduced 10-Feb-93 are performance winners. Internally, they indicate interesting directions that Apple is taking. I'll avoid some technical details, since you've probably read numerous articles on these machines already.

The Color Classic's 16 MHz 68030, 10 MB RAM ceiling, and built-in 512- by 384-pixel display sounds suspiciously like the feature set of the Mac LC II. This happens to be the case: according to developer notes, the Color Classic places most of the LC II's components into a compact Mac chassis. The low-level benchmarks confirm this: the Color Classic's performance is virtually identical to a Mac LC II. The exception is that, unlike the Mac LC II, the Color Classic can have a 68882 FPU installed.

The Mac LC III posts nearly the same performance as a Mac IIci. Note that the system we tested lacked an FPU. This is decent amount of horsepower to pack into the LC's small frame, and it should continue as one of the top sellers in Apple's product line. I've talked to several LC/LC II owners who plan to purchase the LC III upgrade when they can.

Now we get into interesting territory. Compare the test results of the Centris 610 to that of the Mac IIfx. As you can see, the 20 MHz 68040-based Centris 610 easily outguns the 40 MHz, 68030-based Mac IIfx. The floating point performance is very good, especially since the Centris 610 I tested had an FPU-less 68LC040 processor. (However, the IIfx lacked the boost of Omega SANE in its floating-point calculations.) This computer is definitely the answer to the "I can buy a killer 486 system for around two grand" argument.

The 25 MHz Centris 650 just edges ahead of the 25 MHz Quadra 700. Several design changes help. First, Apple finally cleaned up the memory sub-system, making it more efficient. They did this using an interleaved memory design in which data occupies adjacent memory banks, and on memory reads the hardware steps through these banks rapidly. The result is that burst reads eliminate two clock cycles on each initial address setup and for each successive read. Apple claims a 10 to 15 percent performance boost with interleaved memory, and on the average, that's what the BYTE CPU benchmarks saw (about 13 percent between the Centris 650 and the Quadra 700). Also, the I/O bus on the Quadra 700/900/950 has gone away: it's been folded into a custom ASIC (Application Specific Integrated Circuit) called the IOSB. The IOSB is clocked at CPU speeds, so on the Quadra 800 most I/O functions operate at 33 MHz. (An exception is Ethernet, since these transceivers are clocked at 16 MHz.) Finally, the engineers eliminated one wait state from the built-in video's VRAM frame buffer, speeding that subsystem up. The result is Quadra performance in a IIvx frame.

The Quadra 800 has all the design advantages of the Centris 650. Its performance is a tad faster than the Quadra 950's. The smaller mini-tower form factor allows you to park it under the desk or on a desktop.

The PowerBook 165c is nearly equal to the PowerBook 180 in performance. However, the display performance is much slower. The reason is that while earlier PowerBooks used dual-ported VRAM for display frame buffers, the PowerBook 165c uses DRAM. The display speed slows when the display controller and CPU contend for access to this DRAM. Note that the PowerBook 165c QuickDraw test, which uses QuickDraw heavily, ran nearly as fast as the PowerBook 180. Since most Mac applications make heavy use of QuickDraw, the PowerBook 165c's screen drawing shouldn't appear as slow as the tests indicate. For example, BYTE's Slow test algorithmically fills successive circles using a seed fill, rather than a QuickDraw region fill, so the effects of DRAM on the video subsystem are emphasized. If you think a PowerBook 165c is part of your future, check out the display quality and speed before buying.

This latest crop of Macs also indicates interesting design trends. First, many of the Macs provide (typically with a VRAM upgrade) 16-bit video. The PowerBook 165c is the exception. I call these Macs "video-ready," in that to adequately represent a frame-grabbed image or QuickTime movie made from an NTSC signal, a Mac must have a 16-bit display. In this sense Apple is incorporating multimedia support across its desktop product line. Also, with the phasing out of the 68000-based Classic, a grand unification of Mac application software can begin. Until now, a developer had to contend with supporting both old QuickDraw (the 68000-based version, which knows nothing of color) and the current 68020/68030-based QuickDraw, which knows of color, GWorlds, and pixels of various sizes. Now the Color Classic uses the same "universal code" found in all Mac ROMs since the IIci up to the Quadras. This universal code consists of 68030-based object code that implements 32-bit QuickDraw and some virtual memory mechanisms.

Certainly there are still millions of 68000-based Macs to support, but over time we can expect the developer's life to get easier because of this unified environment. Furthermore, at some point developers can compile applications exclusively to 68030 code, producing faster applications.

Last but not least, if anything moves Macs, it's the computer stores eliminating old inventory. I heard of amazing fire-sale prices on the Mac IIci ($1500 for the Mac, keyboard, and monitor) and the IIsi.

 BYTE Low-level Macintosh v2.0 benchmark results (preliminary)
    (all figures in seconds, other than the Indexes):
 Table 1
               Clr Classic  LC II  LC III   IIci    C610
     Matrix         18.7    18.6    10.5    10.2     4.3
     8-bit move     94.1    94      51.6    49.1    29.8
     16-bit move    53.9    53.8    27      24.6    16.6
     32-bit move    40.6    40.5    14.7    12.3    10.3
     Sieve           9       9       5.2     5.1     2.7
     Sort           11.4    11.4     6.2     5.8     2.5
     Math           60.2   186.9   105.6    70      37.9
     Sin(x)         17.8    95.6    54      34      19.7
     e^x            18.1   102.9    58.3    45.8    20.9
     TextEdit        3.4     3.6     1.9     1.8     1
     DrawString      1.7     1.8     1       1.1     0.5
     Slow Graphics  27.6    27.1    14.3    10       4.4
     QuickDraw       0.3     0.4     0.2     0.2     0.1
     CPU Index       1.04    1.04    2.03    2.18    3.86
     FPU Index       4.72    1.04    1.83    2.66    5.07
     Disk Index      1.24    1.01    1.48    1.06    2.12
     Video Index     1.08    1.03    1.95    2.17    4.37
     Dhrystone       2000    2083    5000    5555    16666
 Table 2
                   IIfx    C650   Q700   PB180  PB165c  Q950  Q800
     Matrix         6.1     3.5    3.5    8.2    8.9    2.6    2.6
     8-bit move    32.5    22.8   25.7   39.6   40.3   19.3   17.9
     16-bit move   16.7    12.6   15.5   21     21.9   11.7   10.2
     32-bit move    8.8     7.5   10.6   11.7   12.5    8.1    6.5
     Sieve          3.2     2.2    2.3    4.1    4.3    1.7    1.6
     Sort           3.7     2      2      4.9    5.2    1.5    1.5
     Math          45.8     6.1    6.1   25.7   27.4    4.6    4.6
     Sin(x)        21.8     2.8    2.7    8      8.4    2.1    2.1
     e^x           29.5     2.8    6      8.2    8.6    4.7    2.1
     TextEdit       1.7     0.8    1.3    2     14.3    0.8    0.7
     DrawString     1.2     0.4    0.8    1.2   13.2    0.4    0.3
     Slow Graphics  6.2     3.6    3.7   12     21.2    3.6    2.8
     QuickDraw      0.1     0.1    0.1    0.2    0.2    0.1    0.1
     CPU Index      3.36    4.98   4.4    2.6    2.47   5.85   6.36
     FPU Index      4.1    35.25  27.52  10.68  10.07  35.8   46.71
     Disk Index     1.77    3.1    1.83   1.72   1.51   3.35   3.12
     Video Index    2.68    5.29   3.93   2      0.52   5.89   7.02
     Dhrystone      10000  16666  16666  5555   5000   25000  25000
     Mac Classic II has an index = 1.
     Mac IIsi equipped with FPU and Apple video board.
     Mac IIci equipped with Apple cache board and used Thunder/24
       display board without acceleration.
     All systems ran System 7.1 with AppleTalk off and
       extensions disabled.
     Mac LC III lacked FPU.
     Centris 610 lacked FPU.
     Color Classic equipped with FPU.
     Disk tests removed for brevity.


Non-profit, non-commercial publications and Web sites may reprint or link to articles if full credit is given. Others please contact us. We do not guarantee accuracy of articles. Caveat lector. Publication, product, and company names may be registered trademarks of their companies. TidBITS ISSN 1090-7017.

Previous Issue | Search TidBITS | TidBITS Home Page | Next Issue