Ionic Column - May 1998
by David Parry
Englishman David Parry lived in Tokyo from 1980 to 1994 and was a member of TPC from 1986. Currently based in Düsseldorf and working as a translator, he returns to Japan electronically via the Internet. A frequent contributor to this publication, he was Newsletter Publisher from late 1988 to early 1990 and began the Ionic Column in 1992. For reasons best known to readers, this column even won a prize and an honorable mention...
This column is even more hurried than usual, if that is possible, partly because of a failure of the PC system and the everything related to data transmission. As of time of writing, I have a skeleton system on both PC systems and I can produce text, but I cannot get into Compuserve. This file is being sent by an alternative route.
I hope I won't duplicate myself, since I don't have last month's file to refer to on the system I am using, but the situation at the moment is one of extreme aggravation.
The second PC was to have been upgraded with a 200 MHz "Intel Inside" Pentium chip and an IBM UltraDMA disk with 6.4 GB. The chip swap was no problem and the system is notably faster, but I do notice how slow the Quantum Fireballs are. By comparison, the IBM SCSI disk on the first PC (AMD K6 200) really moves fast. So does its EIDE cousin, but the BIOS on the second PC only recognizes 2.1 GB of its total real estate. More on that anon.
Where is Socket 7?
It is a situation that is going to become commonplace in the next year or two as a number of technical advances are finally implemented. New buses such as USB and Firewire (IEEE something, I can't find the precise number) and all-PCI motherboards will make old hardware redundant, and some of the hottest new hardware won't run on old boards such as mine.
The other factor is the chip socket. Terms such as Socket 7 are nothing to do with a far-flung outpost of the Intergalactic Empire somewhere beyond Alpha Centauri, but refer to the "old" Pentium chips. Pentium I, if you like, since we have a flood of new chips from Intel named after rivers, and the Pentium II is already out there, poised to take over the market for a few months before another Intel offering shoulders it aside. New chips will require new boards, and there is virtually no further development of the original Pentium chips to fit Socket 7. It will not be possible to upgrade again the two motherboards that I have - and the third one that I may have to buy in a day or two.
The second PC has an Award BIOS dated January 1995 and the version number is 4.50 instead of 4.51 on the first PC, but that minor revision makes it impossible for me to use the new EIDE disk properly. IDE disks have always been subject to size limitations within the BIOS, firstly up to 540 MB and then 2.1 GB. The Award homepage refers you to either the vendor of your PC, since the BIOS is frequently modified by them, or to a company that can provide BIOS updates. They ignored my first request, and I won't know if they answered my second request until I can get back on-line. Even then, it will take time for them to send me a new chip, since I don't think that motherboard has a flash EEPROM BIOS. To put it another way, a software upgrade wouldn't help, since flash EEPROM is simply a chip with a long-term rewritable memory.
Another motherboard might be useful, since I could put together a Pentium 100 system with the old motherboard and add one of the sluggard Fireball disks. The latter are only 1 GB, so the size problem would be irrelevant. But what to do with what I already have?
The AMD K6 is a nice chip, faster than the Intel 200, but its appetite for power means that I cannot put it in the second PC with its wimpy 200 watt power supply. It needs to live in the big tower case with a brawny 230 watt supply; I wonder why the 30 watts should be so significant, but they are. The big tower is an all-SCSI system and so should not be affected by the IDE disk size limit, so one option might be to heave out the CPU chips, swap the motherboards between the cases, and put the AMD into the old motherboard in the big PC. But it would mean a lot of reconnecting up of both PCs and the dire possibility of something going seriously wrong with both systems. Instead, I am going to swallow hard and buy a new board, knowing that the BIOS is up to date. Another motherboard would be useful anyway for my digital Frankenstein's monster.
The first system always had problems that I could not trace, possibly because I transferred files over instead of installing from scratch. But a recent rebuilding of the system from scratch did not work; the old bugs reappeared after a certain point. Windows 3.1 and WinWord became increasingly unstable, and then it became impossible to dial out with any communications program. Worse, when I attempted to reinstall Windows, I discovered that both my main and my backup disks had errors on disk 4. Different errors, would you believe, and not capable of being fixed by Norton Disk Doctor. I had to obtain a set of disks of what turned out to be a slightly newer version of Windows 3.1, but it does illustrate this problem of "bit rot" on floppy disks. At least CD-ROMs are immune from this. A number of other master disks and backup disks have also given trouble, just when you need them, naturally, but in almost every case I was able to get them working again.
A frequent cause of SCSI problems is missing or incorrect termination, which at first sight is not the problem on my system. One other question is whether it is necessary to use the utility provided by the SCSI controller maker to carry out a low-level format of the hard disk. The NCR controller in use now supposedly does not need such a process. I am debating whether to change over to the Adaptec controller instead, which is currently in the second PC for use with the Jaz drive and the scanner. Once I get them working. But that is likely to require a complete reformat of the SCSI hard disk, since I swapped the controllers once before and found that only drive C was visible with the Adaptec. I hurriedly quit without making any changes and reinstalled the NCR. Overall, the Adaptec might be a better choice for a SCSI system, as it is the de facto standard.
The SCSI disk currently has three separate partitions, the first one of which is the DOS - Windows 3.1 partition and has an extended partition with logical drives in it. Translated into English, it means that I have drives D, E, F and G when I am using that partition, but I cannot add extra logical drives to either of the other two partitions.
I also ran into what looks like a quirk of the NCR when I was attempting to change over to the third partition to load the new version of Windows together with the Russian fonts and drivers. The system simply would not boot from C. It booted just fine from a floppy and drive C was visible. I tried changing the boot sequence in the BIOS; usually the "SCSI, C, A" sequence is just fine. Still no joy. The answer came when looking through the System Commander manual. It stated that some SCSI controllers can only handle two drives. If this also applies to partitions, then that explains why I could not boot from the third partition. System Commander evidently has a software workaround for this problem.
A resounding "Nyet"
As always, these things come when the system is misbehaving and I have an urgent deadline. "Could I print out a WinWord file in Russian?" I first tried to install Windows in the partition I had set up for Russian. No luck. Finally, and with a deep breath, I installed the drivers on top of the Windows 3.1 system in the second PC. It is a bit like WIN/V, except that it only affects the keyboard layout and does not Russify all the menus. You can toggle between the English and Russian keyboards with the hotkey that you select or else click on a little box at the top of the screen. The fonts have names like ARIALCYR to prevent them from overwriting the existing Arial fonts, but on my system there was some kind of mismatch and Russian text was displayed as a square box for each character. The same applied for style names in Russian, the file I was looking at, or test text typed in within WinWord and with the Russian keyboard. Note: unlike with Japanese, there is no toggle key on the board. Cyrillic is variously handled on either side of the ASCII 127 demarcation zone, being either low ASCII or high ASCII characters in vulgar parlance. The net result was a big fat "nyet".
An expensive lesson
This has not been a good month. My ISDN phone system crashed about two weeks ago. This happened just after a friend and I had been working on the second PC to reconfigure it. That is the PC system with the ISDN card, the delightfully named Fritz! card, and my phone system well and truly went on the fritz. A Telekom engineer was hurriedly summoned on the Monday morning, and he had the phones working again within minutes after disconnecting the cable to the Fritz! card and pressing a little button to reset the system. He said that he was no expert in these matters, but that ISDN cards often caused problems on the telephone line if they were not set correctly. Perhaps something got changed and then conflicted with the Fritz! card? Anyway, it will remain disconnected for the time being.
High-end video cards always seem to be a source of trouble. Could the Millennium II be the cause of the system instability I have experienced in the big PC? The interaction of the various forms of hardware makes me wonder if it would be better to move over to Windows 95 and hope that Plug 'n' Play will eliminate conflicts. Plus, ISDN telephony is far better under Windows 95, and I will need the latter if I do get ADSL / cable modem.
The second PC has been running for about a month now with System Commander and the two versions of Windows in separate partitions. SC and Windows 95 seem agreeably stable, so perhaps I should move over to Windows 95 now. Recent versions such as the one that I have seem to be more stable and bug-free than the early offerings. Another point is that setup is supposed to be easier. Plus, a good deal of the newer software is for Windows 95 only. The main thing I do not like about Windows 95 is that it insists on installing or deinstalling new hardware at once, rather like the old Micro Channel Reference Disk business. That is a bore if you are swapping boards, but I suppose it eliminates the problem of references to irrelevant software or the wrong drivers.
With Windows 98 due in a few months, I do wonder if I might have to do a reinstallation in a few months time. The most radical feature of Windows 98 would be the 32-bit FAT system, which would make much of my old software useless. I might end up having to maintain a DOS partition just for the Q&A database and XyWrite, still my favorites after all these years and still unsurpassed for speed and reliability.
Losing my memory
One major issue is that is is not as necessary to tweak the DOS memory as it is in Windows 3.1, plus some of the drivers only need to be loaded in Windows and thus it is no longer necessary to worry so much about the lower 640 K. It is just not possible under DOS to load network drivers and the FOSSIL drivers for ISDN plus MSCDEX for the CD-ROM and have a decent amount of memory left. Even the cloaked utilities from Helix don't help; my Contour mouse cannot use the Helix MS mouse driver, and I have found that the Netroom memory manager (also from Helix) simply will not work if you have an ASPI driver loaded. It works fine in a SCSI system until you add the ASPI driver, which many programs require and is essential if you want to run a SCSI CD-ROM drive or external backup devices such as tape or the Jaz.
A pain in the wallet
The other issue is Windows NT. I would consider it seriously if I could be sure of being able to get all the hardware drivers I need and if there are no file compatibility issues with Windows 3.1 / 95 programs such as WinWord, but that is not so at the moment. Perhaps in a year or so, at which point I would again be shopping for a new PC. It seems to be a two-year cycle at the moment, meaning that the new hardware is a sufficient improvement after two years to justify an upgrade. Oh my poor wallet!
Comments or feedback or more information?
© Algorithmica Japonica Copyright Notice: Copyright of material rests with the individual author. Articles may be reprinted by other user groups if the author and original publication are credited. Any other reproduction or use of material herein is prohibited without prior written permission from TPC. The mention of names of products without indication of Trademark or Registered Trademark status in no way implies that these products are not so protected by law.
The Newsletter of the Tokyo PC Users Group
Submissions : Editor
Tokyo PC Users Group, Post Office Box 103, Shibuya-Ku, Tokyo 150-8691, JAPAN