THE VIEW FROM CHAOS MANOR
April 12 - 18, 1999
This is a day book. It's not all that well edited. I try to keep this up daily, but sometimes I can't. I'll keep trying. See also the monthly COMPUTING AT CHAOS MANOR column, 4,000 - 7,000 words, depending. For more on what this place is about, please go to the VIEW PAGE.
|Previous Weeks of The View:||For an index
of previous pages of view, see VIEWDEX.
See also the New Order page, which tries to make order of chaos. These will be useful.
For the rest, see What is this place? for some details on where you have got to.
If you subscribed:
If you didn't and haven't, why not?
For the BYTE story, click here.
The LINUX pages are organized as the log, my queries, and your responses and advice parts one, two, three, and four. There's four pages because I try to keep download times well under a minute. There are new updates to four.
Highlights this week:
April 12, 1999
There's a great deal of mail to be cleaned up. The first four items in Highlights refer to last week, because they were at the end of the week and were important. I have to go out now, but when I get back I have some interesting mail to post. Today the Internet World conference began but I'm not going; I will go down for the exposition and show and see what's new. Darnell and Alex and Eric are already down there so I don't think I'll miss much.
Norton Utilities 4 seems to be doing good things to my Winchip W 95 OSR2 system. And memturbo is wonderful: it really does garbage collection. I have the registered version and we will see if its optimizing eatures do anything interesting, but it is worth the price just for memory management/garbage collection.
Today I had a real problem trying to use Netscape 4.5 to download a file from an NT tools site (I'll have specifics later). I think the site may use Active X controls, but I am not sure. I do know that as soon as I looked at it with IE4 I got the downloads doing PRECISELY the same things I did with Netscape. As to why I still use Netscape (but am thinking of dumping it) I figured that Front Page would certainly create pages that looked at least as good in IE as in Netscape, and thus I examine my pages with Netscape to be sure I haven't done anything awful. But Internet Explorer 5 is sitting here waiting to be installed, and I think I will and make it the default, and Netscape can become the standby.
And I don't want Netscape 4.51 because it says it will fix you up with AOL, and I do not want random messages asking me if I want an AOL account; Netscape has embedded some stuff so deep into Roberta's machine that we STILL can't get rid of it on her 95 OSR2 machine (with 98 there's a startup manager that does the job) but I think I have a startup manager for 95 that will do it. I'll have more on that later too.
Thanks to all those who sent or sent links to startup managers. I'll check them for the means of killing the little AOL Man on Roberta's machine.
April 13, 1999
I shall do my taxes today, since tomorrow I have to go down to Internet World and see what's new there. Meanwhile, a new way to waste my time:
I have very little experience with the Bank of New York, whose letterhead proclaims it "New york's first bank -- founded 1784 by Alexander Hamilton," but what experience I have had makes me anxious to avoid more.
It began with a letter dated April 1, 1999 and arriving here on April 12 after close of business in New York. It was addressed to myself and my wife as joint tenants, the form of the address making it probable that the names and addresses had been obtained from the Prudential, which acted as my broker for some years. The letter says that "Our records indicate that we are holding funds belonging to you In accordance with abandoned property regulations, failure to contact the Bank will result in your monies being escheated to your state of residence.
"To claim these funds you may contact out Customer Service Area [gives an 800 number] or return this notice no later than April 12, 1999."
Well, since their notice didn't get to me until afternoon of April 12 and they closed before I got it, I couldn't do that. So this morning I telephoned the 800 number. After an unconscionable number of electronic messages presumably designed to get you to hang up and stop bothering them, I got a human being on the phone. I presume human, and male; I don't think artificial intelligences are up the conversation I had.
He wanted to know what this was about. I pointed out that I didn't have the foggiest notion because other than in my history books I had never heard of the Bank of New York and had no reason to understand why they had any of my money. The letter didn't say. I gave my social security number (which wasn't supposed to be a national ID number, but which seems to have become one). No joy. Way down at the bottom of the letter, off in one corner, is a number that appears to be a form number, but it was #0222xxxx. I read them that number.
Enlightenment. The bank can't find me by name or social security number, but this is an account number. Why it is in small print down in one corner of the page, and isn't labeled "account number," isn't known to me.
I had, until March 27, the grand total of $7.53 (Seven Dollars and Fifty-three cents) in their bank. On March 27 the money was turned over to the State of California. On April 1 the Bank of New York sent me a notice that they would turn the money over to the State of California if I didn't get to them by April 12; but they mailed the letter in such a way that I didn't get it until April 12. That letter was the first indication that I had any money in their silly bank.
Now, they would be willing to give me a telephone number (not toll free) that I could call to try to get California to give me my money back. Wasn't any responsibility of the Bank of New York, New York's oldest bank founded 1784 by Alexander Hamilton, no sir, not no way.
Since I had used up more than $7.53 worth of my time, I declined to go try to persuade California Treasury bureaucrats to give me back my money.
Someone should tell the Bank of New York about computers. The letter that told me of this account could as easily have given me the amount, saving us all some time.
The letter is unsigned: where there might be a signature it says "The Bank of New York, Trustee." I am not sure what Trustee means, but apparently it does not mean that the Bank of New York will inform the account holders about money paid into the Bank of New York, New York's oldest bank founded 1784 by Alexander Hamilton, as part of a class action court settlement until it is too late to collect it. If I ever need a trustee in the State of New York I will remember that. The sum was trivial and California is welcome to it. The waste of my time wasn't trivial, and I resent it.
The one clear message from WinHec was that the limiting factor in small computer speeds is no longer CPU speed, but memory speeds. This is because raising the CPU speed without raising the bus speed is getting to the point of diminishing returns. One needs to make the whole system run faster: not just internal calculations, but data transfer from Level 2 cache to chip internal level 1 cache; faster snooping to see what to put into that cache; faster transfer of calculated data to the display; and so forth. Most of this is accomplished by raising the bus speed. For some time now PC motherboards have operated at 66 MHz, but 100 is becoming increasingly popular. Bus speeds at 100 require PC-100 certified memory, which costs more (but not a LOT more) than standard memory. Now the tendency is to even faster bus speeds: 133 MHz seems to be the next plateau where we'll pause for a while.
Intel is backing a system known as RAMBUS, which I won't explain here: the important fact is that works at 133 bus speeds and is capable of operating at high transfer speed rates, but does little about the latency. Loosely defined, latency is the time it takes to get the FIRST byte of data, while transfer rates are concerned with how fast the rest of the data flows.
Traditional memory doesn't work well at 133 MHz; most PC 100 chips will begin to show soft errors at 120 or so. PC-133 chips exist but they're expensive. Rumor has it that one reason they are expensive is that some PC box makers are buying up PC-133 memory in order to hedge their bets on RAMBUS. They are doing this because RAMBUS may cost more than was thought, while so far performance improvements have been disappointing, in the 15 to 40% improvement range, not the 2x to 5x that we keep hearing about.
This may be because a great deal of memory transfer is from the program in main memory to the Level 2 cache, and the latency of memory calls can be more important than the sustained transfer rate. At this point we get into rocket science, and I don't much trust the theorists: at this point I want to see hard benchmark numbers, and I have not seen them.
All this may be rumor only, but I have heard it from more than one authoritative source. The problem is that sources tend to have attachments. Those attached to Intel don't show me test results, but they do point out that Intel isn't going to support other kinds of memory: they are committed to RAMBUS, RAMBUS is good now and will be better soon, and competition will force PC-133 RAMBUS memory down and down until it costs no more than ordinary memory does now. How long it will take for Moore's Law and competition to work this magic is not so clear.
Those attached to non-Intel chip makers point out that they can make fast PC-133 machines with traditional but upgraded memory cheaper than Intel can make RAMBUS machines, and for AMD at least their chip speeds are competitive. Non-Intel systems made great inroads into Intel's market share down at the lower price end of things. Intel doesn't like that and wants that market back. RAMBUS may be a big factor in price.
There is, in a word, a lot of FUD. (Fear, uncertainty, and doubt.) It is not, I think, deliberate rumor mongering. It's quite real uncertainty. In particular, the performance enhancements of the new systems with RAMBUS may not be enough to justify the increased prices, which can be substantial. What people will pay for 3x they won't pay for 1.3x.
And for the moment, here endeth the epistle. If anyone knows a lot more -- particularly if anyone has hard data on performance and price -- they're not talking to me.
Me, I want things to go well. I'm rooting for RAMBUS, which I wrote about in the old BYTE several years ago as a promising technology; we may even have given it a technology award. We certainly considered it. But the proof of the chipset is in benchmarks
I have opened a new discussion page on GULF WAR SYNDROME sparked by some revelations coming out of Tulane and published in Insight Magazine this week. It will be an ongoing story. There is plausible evidence that it is real, and that we know the cause. How much of that is true I don't know. The waters are very muddy on this. My own interest is as a novelist, I suppose: there's a fascinating story in there no matter how it turns out.
April 14, 1999
Was down at Internet World all morning. Short version: there's a lot more bandwidth but no one knows what to do with it all. The really ambitions applications need even more. There are several outfits that will set you up with a server, hardware and software, for a thousand dollars or so. And one that will give you the box and software if you will signe up for $70 a month for two years of their internet service.
And prices are falling at the low end. Routers are a lot less expensive. There are tons of page creation software tools. The revolution continues, at Moore's Law pace: twice the capability for half the price every 18 months.
Don Giovanni was great. I had forgotten how much Mozart hated tenors. The poor tenor has to keep thundering about how he will avenge his love's dead father. But not just yet. Not until he has a gang, and maybe not even then. What a twit.
April 15, 1999 At least
death doesn't get worse every year.
Taxes first. All the accounting is done and most of it is all enterest in TurboTax which I continued to recommend. I have not done the state part but that ought to be a bit easier and I have the program.
Confusion and doubt in Kossovo. Whose airplanes attacked a convoy of refugees? But you know, I'd be a lot more confident about those Pentagon claims that this might be Serbian aircraft if we had any prior indication that they were flying aircraft in that zone, or that we had attacked them. Surely AWACS would see them the instant they took off? Perhaps some slap happy Pentagon PR type decided to tell the first thing that came into his head? But my paper this morning has a bit "Who Done It?" headlined article that is at best annoying. I mean, surely we would know who's flying in a war zone? Because if we don't, there's worse news than friendly fire on a convoy of refugees.
Now the radio says, "well, it was our airplanes, but we only fired one missile, and we only attack military targets." The second part is certainly true by intent, but in war everything is very simple, but the simplest things are very difficult. You will always have unintended consequences. One of them will be that some well-meaning chap will say nonsense like the foggy story that maybe it was a Serb airplane, undermining the credibility of all the official stories.
Will Chechnya now ask for NATO help in forming an "autonomous republic" within the USSR with NATO troops as guarantors? If they did would we respond? But no, the Russians have nukes; a lesson Milosovec is probably capable of learning.
Incidentally, at $4 billion for the way, that is roughly $4000 per Albanian in the province. Is this a bargain or would we have done better to buy the historic parts, have Disney Coroporation develop the old battlefield and cemetary as Field of Blackbirds Land, and not destroy the Yugo factory? There could be a monorail to the Serbian border, so that any Serb who wanted to visit wouldn't have to set foot on Albanian controlled parts of Kossovo...
Republics ought not lightly make war. When they do, it ought to be Constitutional war by declaration by Congress. It is one thing to give a President authority to respond to attacks on US territory or even citizens abroad. It is quite another to transfer war making power from Congress to the executive. The Framers knew that. The King of England had and has the right to make war on whomever he pleases (of course that now means the Crown which is the Cabinet); Parliament must vote funds to pay for it, but has no part in the actual decision to go to war. So it was in 1787 and the Framers took that power from the President and lodged it in Congress in a deliberate decision. We seem now to have reversed that decision. The consequences may be more than we thought.
The Phone Company says I am 18,000 feet from their switching system and thus they can't give me DSL, but I have a commercial outfit that says they can set me up with 144K at a reasonable price. INstallation is about $750 including the router. Sounds good to me. Earthlink has been all right, really, and I'll hang on to the account as a way to get onto the net from other locations. And we'll see how it goes here. This may be the wave of the future. There's a lot of good software at the Internet show, and this will let me set more things up. All told not all that expensive. Except this is Chaos Manor, and things never go quite as smoothly as one expects...
April 16, 1999
A day devoured by locusts.
Errands, banking, hiking on the hill, the spent a good part of the day and evening trying to install Windows 98 on a new machine. It appeared to hang at the same point, about two screen into installation, after the Install Wizard began to do its thing. eventually I learned: leave it alone a LONG time. I mean a LONG time. No flashing lights, no drive activity, no screen messages, nothing. Just leave it alone a LONG time.
I don't really hope terrorists destroy Microsoft. Really. I don't.
Lot of cleanup.
April 17, 1999
Day began as a disaster and got worse. Maybe it is getting better now. We got Windows 98 installed on the new system (it's Niven's game system) and now we're upgrading with Windows 98 Scond Edition, and that seems to be going well.
The early part of the day found us taking the dog to the vet, where he has to stay for the weekend. He has some kind of abcess on his jaw. Developed very fast. Wasn't there yesterday. Anyway, it's not cheap, but they are taking care of him. It's a new vet, our former veterinarian having retired. We'd been going to him for 30 years, and his boys were in the Scout troop I led, and I suppose he's entitled to retire... The new chap seems fine. Very concerned. We like him. So does Sasha.
Cleaned up two tables in the Great Hall. Clearing away the clutter from taxes.
We're looking at ATI Rage Fury video boards; very cool, good imaging, water looks like water. The Creative Sound Blaster Live is excellent, too. Creative's DVD decoder board does bad things to text files: it dims down the display. It's easy enough to cable around it when not using DVD. Niven's machine uses a software decoder, and thus there's no image degrading, but the DVD/CDROM drive overheats if run a long time. Fixed it with a chip fan.
Windows 98 Second Edition has cured several shutdown problems, and seems just a great deal more stable. We will see, though. But it is the first time I have ever seen the "it is now safe to turn off your computer" screen on Eagle One. Everything else seems to work properly too. 98 second edition looks good. Of course we need to bash it some more...
AAARRRGGGHHH: the ATI Rage Fury does not have a DVD player in the box. Instead there is a coupon to order the DVD player. The Diamond Monster Fusion Banshee video card came with a DVD player. We removed it and alas scrubbed the drivers. This means no DVD until ATI honors their coupon. There is no way to download it, even for ready money; you can get an order form on the web, but you still have to mail that in. AAArrrgghhh.
And Windows 98 Second Edition wants to scall all my disks, but it can't find them, so it shuts itself down. Imbecilic. What I have is a second physical hard drive in IDE Primary Slave, and Windows 98 knows something is there, but doesn't know how to scan it. I am beginning to hate this. It gives instructions to exit to DOS and run scandisk /all.That does no good, but in a text file I find I can do setup /is and make it ignore the disk inspection. So I do that.
I consistently get this message:
The following addresses had permanent fatal errors -----
which I presume is in response to the automated reply. I am not sure I have a valid address for bleuf..
April 18, 1999
I'm off to the nursury to get some flowers and snail killer and stuff. I hope to catch up with a bunch of stuff tonight for tomorrow. We're still worried about the dog, of course.
We found found one bug in Windows 98 Second Edition installation. Two, actually. I'll log them here, but I need to report to Microsoft. First, if you have two physical hard drives, Windows 98 can do scandisk only on one of them; but knows there are other drives, and refuses to install because it can't run scandisk. The only remedy is to do RUN F:scandisk /is which tells it to install without doing scandisk. You'll be well advised to do a thorough disk scan before you do that, of course.
Second, if you have two CDROM drives (I have THREE on this machine: DVD as F:, CDROM as G:, and a TEAC 6x24 as R:) you must put the Windows 98 Second Edition cdrom in the lowest numbered drive. I put it in the G: (36x as opposed to 5x in the DVD drive) and when it got to the point of copying files, it died the death. Had to start over again.
Otherwise the installation went smoothly, and we will see if there is any improvement. There certainly was on Eagle One which shuts down normally for the first time in that machine's history. Second Edition includes all the Y2K and Service Pack upgrades to Windows 98, and appears pretty stable. We'll be using it on all the machines here.