01/09/2001

Mailbag:

DOS games under Windows; Today’s post; Disk I/O tweaking

Overheard: [I’d] rather spend a lot of money for a new top-notch and super-light mountain bike, dive gear or quality clothing than on another computer system. The performance of my PC is just fine for the work I do and the games I happen to play once in a blue moon. My PC is over a year old! Shocking!

–Thomas Pabst, of Tom’s Hardware Guide, in his Duron 850 review released Monday

I think that’s another argument for optimizing your PC right there. If Tom Pabst, of all people, can keep a PC for more than a year, then so should everyone else.

Let’s talk about cheap PCs. In college I lived next door to a marketing guy. In 1994, a couple of years before PCs broke the $1,000 barrier for good, I was thinking about $700 PCs. At that time, 386DX motherboards retrofitted with Cyrix 486DLC processors were dirt cheap, and I figured with 4 MB of RAM and a sub-300 MB hard disk, it would be possible to build a PC for that, and I figured it would sell on price alone.

My marketing buddy agreed. Then we did nothing with the idea. But I still think about cheap PCs.

There’s an article over at The Register today talking about a new motherboard from PC Chips that sells for about $80, based on the SiS 730S chipset. An unnamed vendor states that he can buy these PC Chips boards, slap a low-end Duron on them, put it in a case with the smallest hard drive available, and voila, you’ve got a $300 PC. He’s right. For $80, you get audio, video, modem, and LAN. All you need is a CPU ($60), memory (under $50), hard drive ($85), case ($35), and a keyboard and mouse ($15). It won’t be a high-powered system, but it’s miles ahead of anything we had three years ago and we weren’t complaining.

Now, I don’t trust PC Chips motherboards one bit, but I expect a lot of people will follow this dealer’s plan. This will lead to sales and market share for AMD, but it might come at the expense of reputation. You and I can assume people will blame the inevitable problems they’ll have with these systems on the dealer, or on the motherboard, but historically that’s never happened. People equate the CPU with the computer, so problems tend to fall on AMD or Cyrix, right or wrong. But AMD is hedging its bets, rightly focusing on the business market, where it got its big break last week: Micron will be offering Athlons in their Client Pro line of business PCs.

And speaking of Cyrix, they’ve revised the Cyrix III chip yet again, adding L2 cache. Performance is still lackluster, leading people to predict Cyrix’s demise, but I’m not so sure of that. If VIA can sell Cyrix chips at a significantly lower price than Intel or AMD are willing to charge and still make a profit, there’s room. That $300 PC can become a $250 PC if VIA prices its CPUs at $35 and dealers cut available memory.

Inexpensive PCs done right do sell. Commodore and Atari proved that. Their demise came when they reached the point where they no longer could (or would) sell a useful computer at a significantly lower price than everyone else. The Amiga was tremendously useful, but it was at a me-too price. Here was Commodore, the budget computer maker, selling a computer for twice as much as Packard Bell. Hence its demise. Atari’s price was closer to Packard Bell, but the software compatibility wasn’t there. Packard Bell’s death was inevitable once HP and Compaq started eroding their price territory. Packard Bell had a terrible reputation, but they sold until they lost their price advantage. People aren’t going to settle for a Kia if they can get a Honda for about the same price.

Mailbag:

DOS games under Windows; Today’s post; Disk I/O tweaking

Apple. you call this tech support?

This is why I don’t like Apple. Yesterday I worked on a new dual-processor G4. It was intermittent. Didn’t want to drive the monitor half the time. After re-seating the video card and monitor cable a number of times and installing the hardware the computer needed, it started giving an error message at boot:

The built-in memory test has detected a problem with cache memory. Please contact a service technician for assistance.

So I called Apple. You get 90 days’ free support, period. (You also only get a one-year warranty unless you buy the AppleCare extended warranty, which I’m loathe to do. But I we’d probably better do it for this machine since it all but screams “lemon” every time we boot it.) So, hey, we can’t get anywhere with this, so let’s start burning up the support period.

The hold time was about 15 seconds. I mention this because that’s the only part of the call that impressed me and my mother taught me to say whatever nice things I could. I read the message to the tech, who then put me on hold, then came back in about a minute.

“That message is caused by a defective memory module. Replace the third-party memory module to solve the problem,” she said.

“But the computer is saying the problem is with cache, not with the memory,” I told her. (The cache for the G4 resides on a small board along with the CPU core, sort of like the first Pentium IIs, only it plugs into a socket.) She repeated the message to me. I was very impressed that she didn’t ask whether we’d added any memory to the system (of course we had–Apple factory memory would never go bad, I’m sure).

I seem to remember at least one of my English teachers telling me to write exactly what I mean. Obviously the Mac OS 9 programmers didn’t have any of my English teachers.

I took the memory out and cleaned it with a dollar bill, then put it back in. The system was fine for the rest of the afternoon after this, but I have my doubts about this system. If the problem returns, I’ll replace the memory. When that turns out not to be the problem, I don’t know what I’ll do.

We’ve been having some problems lately with Micron tech support as well, but there’s a big difference there. With Apple, if you don’t prove they caused the problem, well, it’s your problem, and they won’t lift a finger to help you resolve it. Compare this to Micron. My boss complained to Micron about the length of time it was taking to resolve a problem with one particular system. You know what the Micron tech said? “If this replacement CPU doesn’t work, I’ll replace the system.” We’re talking a two-year-old system here.

Now I know why Micron has more business customers than Apple does. When you pay a higher price for a computer (whether that’s buying a Micron Client Pro instead of a less-expensive, consumer-oriented Micron Millenia, or an Apple G4 instead of virtually any PC), you expect quick resolution to your computer problems because, well, your business doesn’t slow down just because your computer doesn’t work right. Micron seems to get this. Apple doesn’t.

And that probably has something to do with why our business now has 25 Micron PCs for every Mac. There was a time when that situation was reversed.

The joke was obvious, but… I still laughed really hard when I read today’s User Friendly. I guess I’m showing my age here by virtue of getting this.

Then again, three or four years back, a friend walked up to me on campus. “Hey, I finally got a 64!” I gave him a funny look. “Commodore 64s aren’t hard to find,” I told him. Then he laughed. “No, a Nintendo 64.”

It’s funny how nicknames recycle themselves.

For old times’ sake. I see that Amiga, Inc. must be trying to blow out the remaining inventory of Amiga 1200s, because they’re selling this machine at unprecedented low prices. I checked out www.softhut.com just out of curiosity, and I can get a bare A1200 for $170. A model with a 260MB hard drive is $200.  On an Amiga, a drive of that size is cavernous, though I’d probably eventually rip out the 260-megger and put in a more modern drive.

The A1200 was seriously underpowered when it came out, but at that price it’s awfully tempting. It’s less than used A1200s typically fetch on eBay, when they show up. I can add an accelerator card later after the PowerPC migration plan firms up a bit more. And Amigas tend to hold their value really well. And I always wanted one.

I’m so out of the loop on the Amiga it’s not even funny, but I found it funny that as I started reading so much started coming back. The main commands are stored in a directory called c, and it gets referred to as c: (many crucial Amiga directories are referenced this way, e.g. prefs: and devs: ). Hard drives used to be DH0:, DF1:, etc., though I understand they changed that later to HD0:, HD1:, etc.

So what was the Amiga like? I get that question a lot. Commodore released one model that did run System V Unix (the Amiga 3000UX), but for the most part it ran its own OS, known originally as AmigaDOS and later shortened to AmigaOS. Since the OS being developed internally at Amiga, Inc., and later at Commodore after they bought Amiga, wasn’t going to be ready on time for a late 1984/early 1985 release, Commodore contracted with British software developer Metacomco to develop an operating system. Metacomco delivered a Tripos-derived OS, written in MC68000 assembly language and BCPL, that offered fully pre-emptive multitasking, multithreading, and dynamic memory allocation (two things even Mac OS 9 doesn’t do yet–OS 9 does have multithreading but its multitasking is cooperative and its memory allocation static).

Commodore spent the better part of the next decade refining and improving the OS, gradually replacing most of the old BCPL code with C code, stomping bugs, adding features and improving its looks. The GUI never quite reached the level of sophistication that Mac OS had, though it certainly was usable and had a much lower memory footprint. The command line resembled Unix in some ways (using the / for subdirectories rather than ) and DOS in others (you used devicename:filename to address files). Some command names resembled DOS, others resembled Unix, and others neither (presumably they were Tripos-inspired, but I know next to nothing about Tripos).

Two modern features that AmigaOS never got were virtual memory and a disk cache. As rare as hard drives were for much of the Amiga’s existance this wasn’t missed too terribly, though Commodore announced in 1989 that AmigaDOS 1.4 (never released) would contain these features. AmigaDOS 1.4 gained improved looks, became AmigaOS 2.0, and was released without the cache or virtual memory (though both were available as third-party add-ons).

As for the hardware, the Amiga used the same MC68000 series of CPUs that the pre-PowerPC Macintoshes used. The Amiga also had a custom chipset that provided graphics and sound coprocessing, years before this became a standard feature on PCs. This was an advantage for years, but became a liability in the early 1990s. While Apple and the cloners were buying off-the-shelf chipsets, Commodore continued having to develop their own for the sake of backward compatibility. They revved the chipset once in 1991, but it was too little, too late. While the first iteration stayed state of the art for about five years, it only took a year or two for the second iteration to fall behind the times, and Motorola was having trouble keeping up with Intel in the MHz wars (funny how history repeats itself), so the Amigas of 1992 and 1993 looked underpowered. Bled to death by clueless marketing and clueless management (it’s arguable who was worse), Commodore bled engineers for years and fell further and further behind before finally running out of cash in 1993.

Though the Amiga is a noncontender today, its influence remains. It was the first commercially successful personal computer to feature color displays of more than 16 colors (it could display up to 4,096 at a time), stereo sound, and pre-emptive multitasking–all features most of us take for granted today. And even though it was widely dismissed as a gaming machine in its heyday, the best-selling titles for the computer that ultimately won the battle are, you guessed it, games.

Scanner troubleshooting secrets

~Mail Follows Today’s Post~

Scanner wisdom. One of the things I did last week was set up a Umax scanner on a new iMac DV. The scanner worked perfectly on a Windows 98 PC, but when I connected it to the Mac it developed all sorts of strange diseases–not warming up properly, only scanning 1/3 of the page before timing out, making really loud noises, crashing the system…

I couldn’t resolve it, so I contacted Umax technical support. The tech I spoke with reminded me of a number of scanner tips I’d heard before but had forgotten, and besides that, I rarely if ever see them in the scanner manuals.

  • Plug scanners directly into the wall, not into a power strip. I’ve never heard a good explanation of why scanners are more sensitive to this than any other peripheral, but I’ve seen it work.
  • Plug USB scanners into a powered hub, or better yet, directly into the computer. USB scanners shouldn’t need power from the USB port, since they have their own power source, but this seems to make a difference.
  • Download the newest drivers, especially if you have a young operating system like MacOS 9, Mac OS X, Windows ME, or Windows 2000. It can take a little while for the scanner drivers to completely stabilize. Don’t install off the CD that came with the scanner, because it might be out of date. Get the newest stuff from the manufacturer’s Web site.
  • Uninstall old drivers before installing the new ones. This was the problem that bit me. The new driver didn’t totally overwrite the old one, creating a conflict that made the scanner go goofy.
  • Buy your scanner from a company that has a track record of providing updated drivers. Yes, that probably means you shouldn’t buy the $15 scanner with the $25 mail-in rebate. Yes, that means don’t buy HP. Up until a couple of years ago, getting NT drivers out of HP was like pulling teeth; now HP is charging for Windows 2000 drivers. HP also likes to abandon and then pick back up Mac support on a whim. Terrible track record.

Umax’s track record is pretty darn good. I’ve downloaded NT drivers for some really ancient Umax scanners after replacing old Macs with NT boxes. I once ran into a weird incompatibility with a seven-year-old Umax scanner–it was a B&W G3 with a wide SCSI controller (why, I don’t know) running Mac OS 8.6. Now that I think about it, I think the incompatibility was with the controller card. The scanner was discontinued years ago (before Mac OS 8 came out), so expecting them to provide a fix was way out of line.
m I’ve ever had with a Umax that they didn’t resolve, so when I spec out a scanner at work, Umax is always on my short list.

And here’s something I just found interesting. Maybe I’m the only one. But in reading the mail on Jerry Pournelle’s site, I found this. John Klos, administrator of sixgirls.org, takes Jerry to task for saying a Celeron can’t be a server. He cites his 66 MHz 68060-based Amiga 4000, which apparently acts as a mail and Web server, as proof. Though the most powerful m68k-based machine ever made, its processing power pales next to any Celeron (spare the original cacheless Celeron 266 and 300).

I think the point he was trying to make was that Unix plays by different rules. Indeed, when your server OS isn’t joined at the hip to a GUI and a Web browser and whatever else Gates tosses in on a whim, you can do a lot more work with less. His Amiga would make a lousy terminal server, but for serving up static Web pages and e-mail, there’s absolutely nothing wrong with it. Hosting a bunch of Web sites on an Amiga 4000 just because I could sounds very much like something I’d try myself if I had the hardware available or was willing to pay for the hardware necessary.

But I see Jerry Pournelle’s point as well.

It’s probably not the soundest business practice to advertise that you’re running off a several-year-old sub-100 MHz server, because that makes people nervous. Microsoft’s done a pretty admirable job of pounding everything slower than 350 MHz into obsolescence and the public knows this. And Intel and AMD have done a good job of marketing their high-end CPUs, resulting in people tending to lay blame at the CPU’s feet if it’s anything but a recent Pentium III. And, well, if you’re running off a shiny new IBM Netfinity, it’s very easy to get it fixed, or if need be, to replace it with another identical one. I know where to get true-blue Amiga parts and I even know which ones are interchangeable with PCs, but you might well be surprised to hear you can still get parts and that some are interchangeable.

But I’m sure there are far, far more sub-100 MHz machines out there in mission-critical situations functioning just fine than anyone wants to admit. I know we had many at my previous employer, and we have several at my current job, and it doesn’t make me nervous. The biggest difference is that most of them have nameplates like Sun and DEC and Compaq and IBM on them, rather than Commodore. But then again, Commodore’s reputation aside, it’s been years since I’ve seen a computer as well built as my Amiga 2000. (The last was the IBM PS/2 Model 80, which cost five times as much.) If I could get Amiga network cards for a decent price, you’d better believe I’d be running that computer as a firewall/proxy and other duties as assigned. I could probably get five years’ uninterrupted service from old Amy. Then I’d just replace her memory and get another ten.

The thing that makes me most nervous about John Klos’ situation is the business model’s dependence on him. I have faith in his A4000. I have faith in his ability to fix it if things do go wrong (anyone running NetBSD on an Amiga knows his machine better than the onsite techs who fix NetFinity servers know theirs). But there’s such thing as too much importance. I don’t let Apple certified techs come onsite to fix our Macs anymore at work, because I got tired of them breaking other things while they did warranty work and having to fix three things after they left. I know their machines better than they do. That makes me irreplaceable. A little job security is good. Too much job sercurity is bad, very bad. I’ll be doing the same thing next year and the year after that. It’s good to be able to say, “Call somebody else.” But that’s his problem, not his company’s or his customers’.

~~~~~~~~~~

From: rock4uandme
To: dfarq@swbell.net
Sent: Wednesday, October 25, 2000 1:22 PM
Subject: i`m having trouble with my canon bjc-210printer…

i`m having trouble with my canon bjc210 printer it`s printing every thing all red..Can you help???
 
 
thank you!!    john c
 
~~~~~~~~~

Printers aren’t my specialty and I don’t think I’ve ever seen a Canon BJC210, but if your printer has replacable printheads (some printers make the printhead part of the ink cartridge while others make them a separate component), try replacing them. That was the problem with the only Canon printer I’ve ever fixed.
 
You might try another color ink cartridge too; sometimes those go bad even if they still have ink in them.
 
If that fails, Canon does have a tech support page for that printer. I gave it a quick look and it’s a bit sketchy, but maybe it’ll help. If nothing else, there’s an e-mail address for questions. The page is at http://209.85.7.18/techsupport.php3?p=bjc210 (to save you from navigating the entire www.ccsi.canon.com page).
 

I hope that helps.

Dave
 
~~~~~~~~~~
 

From: Bruce Edwards
Subject: Crazy Win98 Networking Computer Problem

Dear Dave:

I am having a crazy computer problem which I am hoping you or your readers
may be able to give me a clue to.  I do have this posted on my daily
journal, but since I get very little traffic, I thought your readership or
yourself may be able to help.  Here’s the problem:

My wife’s computer suddenly and inexplicably became very slow when accessing
web sites and usually when accessing her e-mail.  We access the internet
normally through the LAN I installed at home.  This goes to a Wingate
machine which is connected to the aDSL line allowing shared access to the
internet.

My computer still sends and receives e-mail and accesses the web at full
speed.  Alice’s computer now appears to access the web text at about the
speed of a 9600 baud modem with graphics coming down even more slowly if at
all.  Also, her e-mail (Outlook Express) usually times out when going
through the LAN to the Wingate machine and then out over the internet. 
The LAN is working since she is making a connection out that way.

File transfer via the LAN between my PC and hers goes at full speed.
Something is causing her internet access to slow to a crawl while mine is
unaffected.  Also, it appears to be only part of her internet access.  I can
telnet out from her computer and connect to external servers very fast, as
fast as always.  I know telnet is just simple text, but the connection to
the server is very rapid too while connecting to a server via an http
browser is much much slower and then, once connected, the data flows so slow
it’s crazy.

Also, dial-up and connect to the internet via AOL and then use her mail
client and (external to AOL) browser works fine and is as speedy as you
would expect for a 56K modem.  What gives?

I tried reinstalling windows over the existing set-up (did not do anything)
and finally started over from “bare metal” as some like to say.  Reformat
the C drive.  Reinstall Windows 98, reinstall all the drivers, apps, tweak
the configuration, get it all working correctly.  Guess what?  Same slow
speed via the aDSL LAN connection even though my computer zips out via the
same connection.  Any suggestions?

Sincerely,

Bruce W. Edwards
e-mail:  bruce@BruceEdwards.com
Check www.BruceEdwards.com/journal  for my daily journal.

Bruce  🙂
Bruce W. Edwards
Sr. I.S. Auditor  
~~~~~~~~~~

From: Dave Farquhar [mailto:dfarq@swbell.net]Sent: Monday, October 23, 2000 6:16 PM
To: Edwards, Bruce
Cc: Diana Farquhar
Subject: Re: Crazy Win98 Networking Computer Problem

Hi Bruce,
 
The best thing I can think of is your MTU setting–have you run any of those MTU optimization programs? Those can have precisely the effect you describe at times. Try setting yor MTU back to 1500 and see what that does. While I wholeheartedly recommend them for dialup connections, MTU tweaking and any sort of LAN definitely don’t mix–to the point that I almost regret even mentioning the things in Optimizing Windows.
 
Short of that, I’d suggest ripping out all of your networking protocols and adapters from the Network control panel and add back in TCP/IP and only the other things you absolutely need. This’ll keep Windows from getting confused and trying to use the wrong transport, and eliminate the corrupted TCP/IP possibility. These are remote, but possible. Though your reinstall should have eliminated that possibility…
 
If it’s neither of those things, I’d start to suspect hardware. Make sure you don’t have an interrupt conflict (rare these days, but I just saw one a couple weeks ago so I don’t rule them out). Also try swapping in a different cable or NIC in your wife’s machine. Cables of course go bad more frequently than NICs, though I’ve had horrible luck with cheap NICs. At this point I won’t buy any ethernet NIC other than a Bay Netgear, 3Com or Intel.
 
I hope that helps. Let me know how it goes for you.

Dave 
~~~~~~~~~~
From: Bruce Edwards

Hi Dave:
 
Thank you for posting on your web site. I thought you would like an update.
 
I verified the MTU setting was still at 1500 (it was).  I have not used one of the optimizing programs on this PC.
 
I removed all the adapters from the PC via the control panel.  Rebooted and only added back TCP/IP on the Ethernet card. 
 
I double checked the interrupts in the control panel, there do not appear to be any conflicts and all devices report proper function.
 
I still need to 100% verify the wiring/hubs.  I think they are O.K. since that PC, using the same adapter, is able to file share with other PCs on the network.  That also implies that the adapter is O.K.
 
I will plug my PC into the same hub and port as my wife’s using the same cable to verify that the network infrastructure is O.K.
 
Then, I’ll removed the adapter and try a different one.
 
Hopefully one of these things will work.
 
Cheers,
 
Bruce
~~~~~~~~~~

This is a longshot, but… I’m wondering if maybe your DNS settings are off, or if your browser might be set to use a proxy server that doesn’t exist. That’s the only other thing I can think of that can cause sporadic slow access, unless the problem is your Web browser itself. Whichever browser you’re using, have you by any chance tried installing and testing the other one to see if it has the same problems?
 
In my experience, IE 5.5 isn’t exactly the greatest of performers, or when it does perform well, it seems to be by monopolizing CPU time. I’ve gotten much better results with IE 5.0. As for Netscape, I do wish they’d get it right again someday…
 
Thanks for the update. Hopefully we can find an answer.

Dave 
~~~~~~~~~~ 

Abandoned intellectual property

Abandoned Intellectual Property. I read a piece on this subject at OSOpinion over the weekend, and I’ve been thinking about it ever since. There are, of course, a lot of people calling for abolition of copyright or radical changes. This is, believe it or not, one of the tamer proposals I’ve read.

I’m definitely of two minds on this one. Take my first ever publication for money, in 1991. Compute Magazine, before Bob Guccione had managed to totally ram it into the ground, opted to buy my spring break project I collaborated on with a friend. We were writing a video game for the Commodore 64 and 128 and we were getting tired of trying to draw the title screen manually with graphics commands (bad enough on the 128 which had Basic commands to do such things, but on the 64 you were talking peeks and pokes all over the place–someone really should have written this thing back in 1982!) so we wrote a program to do the work for us. You loaded the sprites, moved ’em around, hit a key, and it gave you the Basic code to re-create the screen, suitable for inclusion in your program. We never finished the game, but we got a cool $350 and international recognition (OK, so it was a dwindling audience, but how many high school kids can say they’re published authors at age 16?).

Now, the problem. General Media whittled Compute down until it was basically just another PC mag, abandoning the multiplatform support that made it so great (I read about my beloved Commie 8-bits but still got the opportunity to learn about Macs, Amigas and PCs–what could be better?), market share continued to dwindle, and eventually Guccione and GM sold out to Ziff-Davis, who fulfilled your subscription with a choice of mags (I remember I opted for PC/Computing). So the copyright went to Ziff-Davis, who never did anything with the old Compute stuff. A few years later, Ziff-Davis fell on hard times and eventually hacked itself up into multiple pieces. Who owns the old Compute stuff now? I have no idea. The copyrights are still valid and enforcable. I seriously doubt if anyone cares anymore whether you have the Nov. 1991 issue of Compute if you’re running MOB Mover on your 64/128 or emulator, but where do you go for permission?

The same goes for a lot of old software. Sure, it’s obsolete but it’s useful to someone. A 68020-based Mac would be useful to someone if they could get software for it. But unless the original owner still has his/her copies of WriteNow, Aldus SuperPaint and Aldus Persuasion (just to name a few desirable but no-longer-marketable abandoned titles) to give you, you’re out of luck. Maybe you can get lucky and find some 1995 era software to run on it, but it’ll still be a dog of a computer.

But do we have an unalienable right to abandoned intellectual property, free of charge? Sure, I want the recordings Ric Ocasek made with his bands before The Cars. A lot of people want to get their hands on that stuff, but Ocasek’s not comfortable with that work. Having published some things that I regret, I can sympathize with the guy. I like how copyright law condemns that stuff to obscurity for a time. (Hopefully it’d be obscure in the public domain too because it’s not very good, but limiting the number of copies that can exist clinches it.)

Obscurity doesn’t mean no one is exploited by stealing it. I can’t put it any better than Jerry Pournelle did.

I don’t like my inability to walk into record stores and buy Seven Red Seven’s Shelter or Pale Divine‘s Straight to Goodbye or The Caulfields’ Whirligig, but I couldn’t easily buy them in 1991 when they were still in print either. But things like that aren’t impossible to obtain: That’s what eBay and Half.com are for.

For the majority of the United States’ existence, copyright law was 26 years, renewable for another 26. This seems to me a reasonable compromise. Those who produce content can still make a living, and if it’s no longer commercially viable 26 years later, it’s freely available. If it’s still viable, the author gets another 26-year-ride. And Congress could sweeten the deal by offering tax write-offs for the premature release of copyrighted material into the public domain, which would offer a neat solution to the “But by 2019, nobody would want WriteNow anymore!” problem. Reverting to this older, simpler law also solves the “work for hire” problem that exploits musicians and some authors.

All around, this scenario is certainly more desirable for a greater number of people than the present one.

From: Bruce Edwards

Dear Dave:

I am having a crazy computer problem which I am hoping you or your readers may be able to give me a clue to.  I do have this posted on my daily journal, but since I get very little traffic, I thought your readership or
yourself may be able to help.  Here’s the problem:

My wife’s computer suddenly and inexplicably became very slow when accessing web sites and usually when accessing her e-mail.  We access the internet normally through the LAN I installed at home.  This goes to a Wingate machine which is connected to the aDSL line allowing shared access to the internet.

My computer still sends and receives e-mail and accesses the web at full speed.  Alice’s computer now appears to access the web text at about the speed of a 9600 baud modem with graphics coming down even more slowly if at
all.  Also, her e-mail (Outlook Express) usually times out when going through the LAN to the Wingate machine and then out over the internet.  The LAN is working since she is making a connection out that way.

File transfer via the LAN between my PC and hers goes at full speed. Something is causing her internet access to slow to a crawl while mine is unaffected.  Also, it appears to be only part of her internet access.  I can
telnet out from her computer and connect to external servers very fast, as fast as always.  I know telnet is just simple text, but the connection to the server is very rapid too while connecting to a server via an http
browser is much much slower and then, once connected, the data flows so slow it’s crazy.

Also, dial-up and connect to the internet via AOL and then use her mail client and (external to AOL) browser works fine and is as speedy as you would expect for a 56K modem.  What gives?

I tried reinstalling windows over the existing set-up (did not do anything) and finally started over from “bare metal” as some like to say.  Reformat the C drive.  Reinstall Windows 98, reinstall all the drivers, apps, tweak the configuration, get it all working correctly.  Guess what?  Same slow speed via the aDSL LAN connection even though my computer zips out via the
same connection.  Any suggestions?

Sincerely,

Bruce W. Edwards

~~~~~~~~~~

Hi Bruce,

The best thing I can think of is your MTU setting–have you run any of those MTU optimization programs? Those can have precisely the effect you describe at times. Try setting yor MTU back to 1500 and see what that does. While I wholeheartedly recommend them for dialup connections, MTU tweaking and any sort of LAN definitely don’t mix–to the point that I almost regret even mentioning the things in Optimizing Windows.

Short of that, I’d suggest ripping out all of your networking protocols and adapters from the Network control panel and add back in TCP/IP and only the other things you absolutely need. This’ll keep Windows from getting confused and trying to use the wrong transport, and eliminate the corrupted TCP/IP possibility. These are remote, but possible. Though your reinstall should have eliminated that possibility…

If it’s neither of those things, I’d start to suspect hardware. Make sure you don’t have an interrupt conflict (rare these days, but I just saw one a couple weeks ago so I don’t rule them out). Also try swapping in a different cable or NIC in your wife’s machine. Cables of course go bad more frequently than NICs, though I’ve had horrible luck with cheap NICs. At this point I won’t buy any ethernet NIC other than a Bay Netgear, 3Com or Intel.

I hope that helps. Let me know how it goes for you.

Amiga influence on Linux

Amiga lives! (Well, sort of). When it comes to GUIs, I’m a minimalist. Call me spoiled; the first GUI I used was on a 7.16-MHz machine with a meg of RAM, and it was fast. Sure, it wasn’t long before software bloat set in and I had to add another meg, and then another, but at a time when Windows 3.1 was running like crap on 4 megs and only decently on 8, I had 6 megs on my Amiga and didn’t really know what to do with all of it. So I left 3 megs available to the system, ran a 3-meg ramdisk, and all was well with the world. Until Commodore’s raw dead fish marketing caught up with it and pulled it and the company under.
Under Linux, KDE and GNOME look good, but they run slower than Windows on my PCs. And I like the idea of my P120 being a usable box. I can do that under Linux, but not with KDE as my Window manager. There’s IceWM, which is nice and lean, and there’s xFCE, which resembles HP’s implementation of CDE (and also resembles OS/2, bringing back fond memories for me–why is it everything I like is marketed as raw dead fish?), and now, two years after its release, I’ve discovered AmiWM.

AmiWM (http://www.lysator.liu.se/~marcus/amiwm.html) is a clone of the Amiga Workbench, the Amiga’s minimalist GUI. It’s small and fast and reminds me of the good old days when computers were computers, and didn’t try to be CD players, dishwashers, toaster ovens, televisions, and the like. For an aging PC (or for a new one that you want to run as quickly as possible–hey, you must be mildly interested in that, seeing as you’re reading my site and that’s my specialty), this one’s hard to beat.

eMachine upgrade advice

I got some mail some time back about eMachine upgrades that I never got around to posting. I’ll just summarize because that’s easier (it keeps me off the mouse).
First off, definitely look into a new hard drive. You can pick up a 7200-rpm drive of decent size (10-15 gig) for under $100 these days. I’ve had trouble getting Western Digital drives to work with older disk controllers, but no problems with Maxtors, and I get better performance and reliability from Maxtors anyway.

Next, eMachines tend to have problems with their power supplies. Get a replacement from PC Power & Cooling. It’s $45. Cheap insurance. And chances are the hard drive will perform better, since the PCP&C box will actually be supplying the wattage it claims to supply (which may or may not be true of the factory box). And remember: low-cost PCs have always had skimpy power supplies. Commodore and Atari made great low-cost computers 20 years ago, but they had horrendous power supplies. Given a properly made third-party power supply, a Commodore or Atari could run for 10-15 years or more (and often did).

Finally, get 128 megs of RAM in the system somehow. If you’ve got 32, just go buy a 128-meg stick. If you’ve got 64, get a 64-meg stick or, if you can afford it, get a 128.

Since eMachines have pretty wimpy integrated video, you might also look into a PCI video card with a Matrox, nVidia or 3Dfx chipset. Matrox gives slightly better 2D display quality, while nVidia and 3Dfx give better speed with 3D games. If you’re into gaming, definitely look into a new card.

That’s the strategy I follow with any upgrade. Get a modern disk in there, then get more memory, and replace anything else that seems underpowered. Do the disk first, then deal with memory, then possibly the video. Then, and only then, do I start looking at CPU upgrades. I’ve turned 200-MHz junkers into very useful machines again just by adding memory and a fast disk. The CPU isn’t the bottleneck in most systems.

Optimizing Windows startup

Mail. More Windows optimization questions.

From: ChiefZeke
Subject: Re: Items to consider
To: Dave Farquhar

Dave,

Would things be a bit faster if the user opted to start programs via the ‘RUN’ function of the Registry rather than via the Startup folder? I have seen this option mentioned in a couple of magazine articles.

Jerry

I imagine they would be slightly faster, since the file and path names, etc. would be stored in registry keys all in one place as opposed to individual icons, one per program, scattered all over the place.

You might also use run= strings in win.ini instead–I suspect that technique would be faster still, being a flat text file rather than a convoluted database.

Now, whether doing this would make any noticeable difference on a modern PC is another question. We may be talking shaving fractions of a second off your boot time. I imagine the difference would be more noticeable on marginal machines (though I’m not very eager to re-commission my 486SX/20 to try it). I just saw a 486DX4/75 laptop today that takes 1.5 minutes to boot Windows even without any items in any of the startup places and a fully optimized msdos.sys–a decked-out modern system similarly configured could boot in about 15 seconds. I can’t imagine your system needing much more than 20 seconds to go from POST to desktop (I’m not familiar with modern Western Digital drives like you have but I imagine their performance must be comparable to the Quantums and Maxtors I use).

This trick dates back to the Win3.1 days, and it was a really good idea way back then–the startup group actually consumed system resources, plus valuable entries in the Windows directory, so eliminating startup and placing items in win.ini could seriously improve a system’s performance back then. Today, Win9x has much better resource management, hard drives and CPUs are much faster, so you don’t hear about it as much anymore.

Very little else for today. I found my copy of the Lost Treasures of Infocom (both volumes) this week, including Bureaucracy, a text adventure I was never able to beat. I found a walk-through that got me past the part that had me stuck.

It’s a whole lot faster on my PC than it was on my Commodore 128 (which was the machine I originally bought it for, what, 11 years ago?). Amazing how much fun a 12K executable paired up with a 240K data file can provide… (And I’m running this on a dual-processor machine with 96 MB RAM and an 8.4-gig hard drive, both due to be upgraded? Something’s wrong here…)