TTG Home
» Robert Home » Daynotes
Journal Home » Daynotes Journal
Daynotes
Journal
Week
of 11 December 2000
Latest
Update: Friday, 05 July 2002 08:21
|
Search Site [tips]
|
Monday,
11 December 2000
[Last
Week] [Monday] [Tuesday] [Wednesday]
[Thursday] [Friday] [Saturday]
[Sunday] [Next Week]
[TTG
Messageboard] [HardwareGuys.com
Messageboard]
|
If you haven't bought your own copy of PC Hardware in a
Nutshell, please do. If you order a copy from Fatbrain.com by clicking
this
link, we get a commission on the sale. Several readers have written to
tell us that they've ordered multiple copies to give as Christmas gifts.
We think that's an excellent idea, of course.
Intel and AMD have both cut processor prices, and the price of
AMD processors remains well below that of Intel processors running at
equivalent speeds. I am at the point now where I am considering
recommending AMD processors for everything except the very low-end (where
the Celeron has a cost advantage due to the availability of integrated
motherboards) and the very high end (where Intel remains the only choice
for dual processor systems). A glance at AMD quantity pricing (with some
Intel prices in parenthesis) reported in The
Register this morning shows why.
AMD Athlon:
1.2GHz - $254
1.1GHz - $223
1.0GHz - $179
950MHz - $161
900MHz - $143
850MHz - $125
800MHz - $110
750MHz - $95 (Intel Pentium III/733 and /750 - $173)
700MHz - $85 (Intel Pentium III/667 and /700 - $153)
AMD Duron:
800MHz - $79 (Intel Celeron/766 - $155)
750MHz - $65
700MHz - $55
I still prefer Intel chipsets to VIA chipsets, but the fact is that I'd
be equally comfortable using either an Intel or an AMD processor in one of
my main systems. For several good reasons, Intel is likely to remain the
choice of those who buy corporate PC fleets, but it appears that AMD has
become the better choice for home and SOHO systems, most of which are
equipped with better graphics and sound than typical corporate desktops,
rendering Intel's advantage in integrated motherboards much less
important.
There's been some discussion on J.
H. Ricketson's site and John
Dominik's site about their privacy concerns with the messageboards.
Just to reassure them and anyone else who's concerned, let me say this:
You need have no privacy concerns about using these
messageboards.
The main objection seems to be that the boards require registration.
The purpose of that is to prevent the boards from being spammed. The only
information required to register is a name (any name will do) and a
working email address (to which your password is mailed). Note that the
email address doesn't have to be your main email address. Any email
address that you can receive mail addressed to will do, including one you
set up on Yahoo or Hotmail, use only for registering, and then never check
again.
Even if you choose to provide your main email address, your privacy is
protected. You can specify whether or not to show your email address with
your posts. If you choose to do so, a mail-to icon appears with the post.
Anyone who clicks that icon creates a new email message to the address you
provided. If you choose not to have your email address available, it
isn't. People who read your post can still send you a message by clicking
the message icon. That invokes a javascript facility that allows the
reader to send you a private message, but the sender never sees your email
address (nor do you see his).
The messageboards use Javascript because that's the only reasonable way
to automate the processes that the board requires to function. The
messageboards set cookies, but those cookies are first-party cookies (not
ad-tracking third-party cookies), and are used only for the readers'
benefit. They do things like automate entering your name/password,
tracking the last time you accessed the board to allow you to view only
new messages, and so on.
I fail to see how any of this can reasonably raise any privacy
concerns. You have to provide a name and working email address, true, but
that name can be false and that email address can be a convenience
address. We have many users who register with an obvious alias and a
Hotmail or Yahoo email address. If you're concerned about your privacy,
you're free to do the same.
Other than that, the privacy risks are no greater than those associated
with visiting any web site. We do log IP addresses (as do all web sites),
but that information is available only to Greg Lincoln (who hosts the
boards) and me. I know I don't pay any attention to it, and I'd bet Greg
doesn't either.
Concern with privacy is all well and good. I'm a privacy advocate
myself. But crying wolf where no danger exists serves no one.
Click
here to read or post responses to this week's journal entries
[Top] |
Tuesday,
12 December 2000
[Last
Week] [Monday] [Tuesday] [Wednesday]
[Thursday] [Friday] [Saturday]
[Sunday] [Next Week]
[TTG
Messageboard] [HardwareGuys.com
Messageboard]
|
The processor for Barbara's new system showed up yesterday.
It's an Intel Pentium III/1.0G. We'll probably get started building her
system this afternoon, if she's up to it. Barbara has been hampered by a
sinus infection recently.
FrontPage screwed me again yesterday. Or I assume it was
FrontPage. Perhaps it was NT or Microsoft networking. But, at any rate,
Microsoft screwed me. I spent a fair portion of the day updating pages for
the HardwareGuys.com web site,
working directly in FrontPage on HTML documents. I'd created half a dozen
new pages and updated many others. I was about ready to publish the
changes when I decided to check one particular page one last time before
publishing. I double-clicked that page to open it, and FrontPage displayed
an hourglass. It wasn't a very large page, so after a couple of seconds I
started getting concerned. After a minute or so, I was very concerned, but
I've seen FrontPage "go away" like this in the past.
When that happened, FrontPage was always updating every document in the
web to the current date and time. Very annoying, but something I could
live with. This time, after several minutes, the hourglass went away and
the page displayed. But it was completely blank. Hoping against hope, I
switched to HTML view, but found that the page was completely empty in
HTML view as well. Okay, at that point I decided that something had
corrupted the page, so I closed FrontPage and went out to look at the
files in the directory, using first IE and then Notepad. I was horrified
to find that FrontPage (or something) had zeroed out not just that one
page, but every page in the HardwareGuys.com web.
Fortunately, I had a backup copy of the entire web elsewhere on the
network. Unfortunately, I'd skipped my usual practice of making frequent
xcopy backups as I worked, so the backup copy predated all of the changes
I'd made. I checked the recycle bin on Barbara's system, where the live
files reside, but the old versions weren't in there. So I ended up losing
what I'd worked on for much of the day. But at least I haven't lost the
web itself.
I'm sure I'll get messages asking why I continue to use a product that
periodically screws me. I guess the answer is two-fold. First, FrontPage
works and works well most of the time. It's the quickest and easiest way I
know to get a page created and published. Every six months or so it blows
up and costs me a fair amount of time and aggravation, but I continue to
accept that as the price I have to pay for being able to do quick and easy
updates the other five months and 30 days of that period. If I switched to
another product, the learning curve would occupy more time than do the
infrequent FrontPage problems. Second, I have several functioning
FrontPage webs, and no desire to try to transfer those to another product.
If there were an alternative product that I could simply point at my
FrontPage webs and tell it to "go get them" I'd consider
migrating to that product. But I know of no such product.
And I can't really blame FrontPage. It was my own fault. I know that
FrontPage sometimes corrupts its own data, and I should have been doing
backups every few minutes as is my normal practice. I should have saved
those HTML pages as I created them to a different directory on a different
server. I didn't do that for a reason that turns out to be a poor excuse.
In order to backup the files, I have to close FrontPage. I had a whole
bunch of files open in FrontPage and was toggling among them as I worked.
Closing FrontPage would have required closing all the files first and then
reopening all of them once the backup was completed. It's ironic. I
routinely do frequent xcopy backups, and never have a problem. Every time
I fail to do that, it seems I have a catastrophic problem
Ultimately, FrontPage is a dead-end product for me. I'll continue using
FrontPage 2000, but I have no plans to upgrade to Office 10. So FrontPage
2000 is the last version I'll ever use. Unfortunately, it seems that
Microsoft doesn't bother to release many bug fixes for FrontPage. Instead,
they accumulate problem reports and fix some of them in the next release.
A cynical person would claim that they do it that way to encourage people
to upgrade. Why fix bugs for the current release when you can fix them in
the next release to give people a compelling reason to upgrade? But I've
had it with that.
I'll eventually migrate to Linux, and until that day I'll just continue
to live with FrontPage 2000, warts and all.
Click
here to read or post responses to this week's journal entries
[Top] |
Wednesday,
13 December 2000
[Last
Week] [Monday] [Tuesday]
[Wednesday] [Thursday] [Friday]
[Saturday] [Sunday] [Next
Week]
[TTG
Messageboard] [HardwareGuys.com
Messageboard]
|
We didn't get around to building Barbara's new system
yesterday. I wasn't feeling very well, so we deferred working on it. I was
scrounging around for components, and found I had no spare SCSI cards. I
thought I had an Adaptec 2930U2 card on the shelf, but as it turned out
I'd already installed that in one of the test beds. It's not being used
currently, so it's easy enough to pull it, but that does involve tearing
down another system. And as long as we're going to have it open, we might
as well vacuum and clean it.
Barbara hit the roof last night when the networks broke in for
the Supreme Court decision. Her show (the one with Sela Ward which I can
never remember the name of--"Then and Now", "Now and
Again", "Back and Forth", whatever) had just started, and
the network news operation stomped all over it. As I said at the time, all
they needed to do was a 15-second break, saying "The Supreme Court
decided. Bush won on all three counts." But, no. They had to do their
useless "analysis". The Supreme Court effectively ended this
mess by reversing the Florida Supreme Court and ordering an end to
recounts. But the three network news operations, Gore supporters all,
tried to put the best face on things for Gore. Dan Rather on CBS actually
tried to make it sound as though the decision was, if not a victory for
Gore, at least not a serious loss. Give me a break.
All three broadcast news organizations mentioned that the 5-4 decision
would cause a loss of credibility for the Supreme Court. That may be true,
but if it's loss of credibility they're worried about, they should be
looking at their own operations. No one I know takes the network news
organizations seriously any more. They're partisan, arrogant, and
completely out of touch with their audience. The morning newspaper (a
liberal rag if ever there was one) reported the behavior of customers in a
bar with TV sets tuned to the network reports. Everyone listened for 15
seconds, heard that Gore had lost and then proceeded to ignore the TV for
the next hour as the network news weenies "analyzed" the
decision to death. Get a clue, folks. No one wants your opinions or your
analyses. They want the news. Period. Tell us what happened. Don't tell us
what to think about it.
Click
here to read or post responses to this week's journal entries
[Top] |
Thursday,
14 December 2000
[Last
Week] [Monday] [Tuesday]
[Wednesday] [Thursday] [Friday]
[Saturday] [Sunday] [Next
Week]
[TTG
Messageboard] [HardwareGuys.com
Messageboard]
|
Yesterday was disastrous. It all began when we started
building Barbara's new system. Before assembling stuff, we use the Olympus
digital camera to shoot photos for the book and website. A lot of those
were macro shots, and I wanted to verify that the photos were sharp and
properly framed before proceeding. So I took the digital camera back to my
office, moved the SmartMedia card into the FlashPath adapter, and stuck
that into thor, the Win98SE box that has the FlashPath driver
installed. I've done that a couple of times a week for a year or more now,
with never a problem.
But this time, when I double-clicked the Explorer icon, I got an
hourglass for a fraction of second, then nothing. After I tried a couple
other things, the system locked up solid. Okay, reboot and try again. Same
thing. Explorer won't run, nor will any other program. Using Ctrl-Alt-Del
to bring up the task list locks the computer solid. Well, I didn't have
time to mess with it any longer, so I decided to transfer the images using
kiwi, my main workstation. I used to use kiwi for transfers,
but apparently the last time I rebuilt it I neglected to install the
FlashPath driver.
Doing that should have been easy enough. I still had the original
FlashPath for NT distribution file (I seldom throw anything away), so I
ran it on kiwi. Everything appeared to proceed normally. When Setup
finished, it prompted me to reboot the system. I did that, only to get a
bluescreen error saying that a driver called Sd-something-or-other.sys was
missing. I turned off the power and tried again. Same problem. Not Good.
So I tried doing an NT repair installation. No joy. Whatever happened well
and truly hosed the system. Very Not Good.
So I decided just to punt and reinstall NT. In a moment of foolishness,
I decided that as long as I was doing a new installation from bare metal I
might as well install Windows 2000 Professional. I used Windows 2000 Setup
to blow away the contents of the primary hard drive, an 18 GB Seagate
Cheetah, created an 18 GB partition for Windows 2000 and let 'er rip.
Installation seemed to complete successfully. As it was finishing, I was
horrified when I remembered that the reason I'd not been running Windows
2000 on this system was that I could never get the Windows 2000 drivers
for the Matrox Millennium G400 to work. Oh, well, I thought, I'll just
download the latest version of those drivers. Maybe they're fixed. So I
did that on another machine while Setup finished running on my main
system.
Setup completed normally and I rebooted the system. At that point, I
was looking at Windows 2000 at 640X480 on a 19" monitor. Not a pretty
sight. So I copied the new Matrox G400 drivers over from the other system
and fired up Setup. It appeared to complete successfully, and forced a
reboot. After logging back in, I right-clicked on the desktop, chose
Properties, and started to adjust the settings. Windows 2000 recognized my
Hitachi SuperScan Elite 751 monitor as a 752, but that was the only
apparent problem. I chose the List all settings button, and picked
1280X1024 resolution at 85 Hz refresh with 32-bit color. Then the little
dialog popped up to tell me to click OK if I wanted to keep those settings
or otherwise it'd change back to the original settings in 15 seconds, 14,
13, 12, 11, 10 ....
It never got past 10. I clicked OK a couple of times in the interim,
but Windows 2000 just kept counting down. At 10 it locked solid. My mouse
cursor disappeared and the keyboard was dead. Okay, time for a hard reset.
I did that, and the system rebooted normally into 1280X1024 mode.
Temporary glitch, I thought. That turned out not to be the case. I started
moving the mouse cursor across the screen to run a program, and again the
cursor disappeared and the system locked up solid. At this point, I was
cursing out the Matrox drivers, but as it turned out it wasn't their
fault.
I finally came to my senses, realizing that I'd been foolish to try to
install Windows 2000 on a PC that was known not to like it. So I decided
to strip down things to bare metal and re-install NT4. As long as I was
doing that, I decided to install NT4 Server to replace NT4 Workstation. It
wouldn't hurt to have another system as a BDC for my network. So I changed
the boot order yet again to tell the system to boot from the ATAPI CD-ROM,
stuck in the Windows NT 4 Server CD, and restarted the system.
The BIOS boot screen told me that the processor temperature was above
the warning level, which I have set at 50C. I ignored that for the moment,
assuming that all the Windows 2000 work had caused the poor things to heat
up. The system booted normally from the NT4 Server CD and I started
installing. But halfway through the hardware detection phase, the system
hung up solid.
Obviously, I have a hardware problem, but I had neither the time nor
the inclination to address it. So I made an executive decision to stop
using kiwi as my main system and swap in something else. That
something else turned out to be thoth, a testbed system built
around an Antec full-tower case, an Intel SE440BX-2V motherboard, a
Pentium III/800 processor, 128 MB of Crucial PC100 SDRAM, a 20 GB Seagate
Barracuda ATA II hard drive, and a Plextor 12/10/32A CD writer. It already
had Windows 2000 installed on it, although little else. So I fired up the
Office 2000 CD and did a kitchen-sink install from it.
And that's how I ended up running Windows 2000 on my main system,
something I said I'd never do. But needs must, and thoth was the
most suitable system available on short notice. One thing I noticed
immediately was the reduced sound level. Kiwi has a large PC Power
& Cooling power supply, many supplementary in-case fans, and an
external fan that mounts over the power supply to push yet more air. All
told, counting drive coolers, there are something like a dozen fans in kiwi.
It also has a 10,000 RPM 18 GB Seagate Cheetah hard disk as its primary
hard drive, as well as a 50 GB Seagate Barracuda as its secondary hard
drive. With all the spinning metal in there, kiwi sounds like a jet
taking off. Thoth, on the other hand, is so quiet that the only way
I can tell it's running is to look at the LED on the PlexWriter. In fact,
it's so quiet that I can hear the Seagate Barracuda ATA II hard drive
seeking.
As far as performance, my rule of thumb holds true. Kiwi has
dual Pentium III/550 processors and 256 MB of RAM. Thoth has one
Pentium III/800 and 128 MB. Thoth "feels" a bit faster
than kiwi when I'm doing only one thing (like right now, running
FrontPage). But as soon as I get several windows open, the lack of the
second processor becomes obvious as the foreground task begins bogging
down. That's probably a combination of the single processor and only 128
MB. The next time I open the case, I'll probably add another 128 MB.
This has been a hell of a week for me. FrontPage trashes most of a
day's work. One of my secondary systems dies for no apparent reason. My
primary system dies a few minutes later. Barbara's new main system still
isn't built. And I have yet to start my holiday shopping.
I wrote the preceding material late yesterday afternoon on thoth,
my new main system. I'm writing this late last evening on my Compaq Armada
E500 notebook, using the Intel 2011 Wireless link. My initial impressions
of thoth are that it is a fast system, but not as fast as what I'm
used to. Microsoft people talk about "eating [their] own dog
food", by which they mean living with their own products. That's a
good concept, and so I decided to continue using thoth as my main
system, at least for a while.
As nice as dual processors and SCSI are, the simple truth is that the
vast majority of systems out there have neither. If I had been in any
doubt about the benefit of either, working with thoth for a short
while eliminated those doubts. Here are some simple truths:
Two slower processors beats one faster one. No contest here. Kiwi
has two Pentium III/550 processors. Thoth has one Pentium
III/800. An 800 MHz processor is about 45% faster than a 550 MHz
processor. That percentage difference is about what it takes for most
people to "feel" a significant difference between systems. A
difference of less than 25% or 30% is about the minimum that most people
will notice unless they're using two systems side-by-side. A 45%
difference is enough, for me at least, to make a system seem slightly
faster. Not hugely so, but noticeably so. And that's exactly what I
experienced when I started using the 800 MHz processor.
But, and this is a big but, that's true only when comparing apples to
apples--a dual-processor machine to another DP machine, or a single
processor machine to another SP machine. Dual processors make a huge
difference under realistic working conditions, at least under my
personal realistic working conditions. I frequently have a lot of
windows open, a lot of services running, and a lot of stuff going on in
the background. Under those conditions, having dual processors pays off
in spades. As I've noted in the past, with dual processors, the system
just doesn't bog down under load.
I think one of the biggest mistakes Intel has made is in not bringing
to market a stable, inexpensive dual processor motherboard targeted at
individual users. Rather than focus their marketing effort on trying to
convince a relatively small number of people to buy one fast,
high-margin Pentium III processor each, they should be focusing their
efforts on convincing a lot of people to buy two mid-range Pentium III
processors each.
(before anyone comments about memory, yes kiwi has 256 MB and thoth
only 128 MB. But during the tests I always had free physical memory
available, so the relative RAM sizes are not an issue, or at most a very
minor one.)
A SCSI hard disk outperforms an ATA hard disk. People
constantly tell me that I'm behind the times. "Yes, it's true that
SCSI hard drives used to outperform ATA hard drives," they tell me,
"but nowadays the new-generation IDE hard drives are so good that
they match or beat SCSI performance." Wrong.
The IDE hard drive in thoth is a 7200 RPM Seagate Barracuda
ATA II. It's still a current model, although it will soon be superceded
by the follow-on Barracuda ATA III. It'd be easy to argue all day about
which of the current-generation IDE hard drives is best. The Barracuda,
the IBM 75GXP, and the Maxtor DiamondMax Plus are all excellent drives,
and I wouldn't hesitate to use any of them. Performance-wise they're all
comparable. And they're all a step behind current SCSI drives.
Readers with sharp eyes may say that I'm being unfair to IDE. After
all, thoth is built on an Intel SE440BX2-V motherboard, which
supports only UDMA/33. The Barracuda ATA II is a UDMA/66 drive, and the
follow-on Barracuda ATA III is a UDMA/100 model. But repeated tests
we've run show very little real-world difference between UDMA/33 and the
faster variants. Current generation hard drives are just barely capable
of saturating UDMA/33, and that only when running the fastest drives
under ideal conditions.
In stepping down from a 10,000 RPM U2W SCSI drive (the Seagate
Cheetah) to a 7,200 RPM ATA drive, I noticed a big difference in
performance. My first thought was that Windows 2000 had chosen PIO mode
rather than DMA mode (Microsoft doesn't trust DMA). I checked that and
found that the system was running UDMA. Just to verify that something
wasn't horribly wrong, I ran several benchmarks on the Barracuda ATA II.
The results were normal, which is to say comparable with those I'd run
in the past and those published on other web sites. The Barracuda II was
performing exactly as expected, comparably to other 7200 RPM ATA drives.
But, boy is it slow compared to the Cheetah. The difference is
noticeable in routine operations. Large programs take noticeably longer
to load, and any time the system hits the disk the performance lags
relative to the SCSI system.
But the real killer is performance under load. I have a canned batch
file, databack.bat, which uses xcopy to copy files from theodore
(Barbara's current main system and also the main network file store)
down to the local drive. With the SCSI system, I'd fire off that batch
file and go on doing whatever I'd been doing. The batch file ran in the
background with no noticeable effect on disk operations. I recreated
that batch file on thoth yesterday and ran it to copy everything
down from theodore. While it was running, I fired up FreeCell
from the local disk. It took forever to load--two or three times as long
as normal (normal for thoth, that is). So I fired up Word. At
first, I thought the system had locked up because it took so long for
Word to load from the local drive. When you're running under load, the
advantages of SCSI become abundantly clear.
Please don't take anything I've said to mean that I think the Intel
Pentium III/800 is a bad processor or the Seagate Barracuda ATA II is a
bad hard disk. That's not what I'm saying. Both are excellent products. If
they weren't, I wouldn't be using them, and I certainly wouldn't be using
them in my main system. But there are advantages to using dual processors
and SCSI hard drives, and all the wishful thinking in the world doesn't
change that. Workstations (in the original sense of high-performance
scientific/engineering computers built for individual users) use multiple
processors and SCSI hard drives. If a single fast processor and a high-end
ATA hard drive would get the job done, you can believe that that's what
they'd use.
Both of those technologies used to be out of reach for most individual
users simply on the basis of cost. That's no longer the case. Fast Pentium
III processors are now available for less than $200 each. Good quality
dual motherboards aren't all that much more expensive than single
processor motherboards of similar quality. Good SCSI adapters still aren't
cheap, but you can get a decent one to support a hard drive and perhaps
another peripheral or two for between $100 and $150 if you shop around.
You can also pick up a Seagate Cheetah surprisingly inexpensively,
particularly if you don't need huge capacity. A 4.5 GB Cheetah, for
example, can be had for less than $150, and a 9 GB model for less than
$250.
This morning the newspaper says that George Bush is finally admitted
by everyone that matters to be the president-elect. Barbara was upset
the other night when the network news department gratuitously usurped the
10:00 p.m. hour for their useless analyses of the Supreme Court decision,
in the process tromping all over her Sela Ward program (Now and Later,
Before and After, something like that--I can never remember the name, but
Sela Ward certainly is beautiful). One of my readers posted a message on
the messageboard saying that the Canadian network had had the good sense
to break in for a 15-second announcement and then return to the program. I
made the mistake of telling Barbara about that, which only enraged her
further. Fortunately, another reader mailed her to tell her that the
program would be shown later this week on Lifetime or one of those other
cable networks, so she'll get to see it.
At any rate, Barbara was muttering darkly yesterday about how the
networks better not tromp all over Left Wing, another of her
favorite shows, which runs at 9:00 p.m. on Wednesdays. She went off
yesterday afternoon to pick up her parents, who were returning from a bus
tour, and have dinner with them. When she got back, I mentioned that Bush
was to speak on national television. Barbara immediately assumed the
worst--a 9:00 p.m. speech--but I assured her that Bush wasn't due to speak
until 10:00 p.m. She was relieved until I mentioned that Gore was also
scheduled to speak, presumably before Bush. That got her angry again.
"Enough is enough," she cried, and I don't blame her.
So we set up the VCR to tape Left Wing, just on the chance that
it'd be shown, but not really expecting anything. A few minutes before
9:00 p.m. Barbara turned on the television to NBC, which was supposed to
be showing some stupid new show--Mr. Ed or something. Instead, it
was showing a re-run of Left Wing, which was followed by an
announcement that next week would be an "all new" episode
(rather than the presumably partially-used episode previously scheduled).
We watched the first 15 seconds or so of Gore's concession speech and then
turned off the television.
I must say that I was disappointed by Gore's concession. I was hoping
for a more traditional concession. You know, the kind where Gore would be
sitting alone in a room and someone comes in and places a revolver loaded
with one round on the table beside him. After what he's put the country
through, that's the very least he should have done. Drenching himself in
gasoline and then lighting a match would have been better, but that was
perhaps too much to hope for.
Click
here to read or post responses to this week's journal entries
[Top] |
Friday,
15 December 2000
[Last
Week] [Monday] [Tuesday]
[Wednesday] [Thursday]
[Friday] [Saturday] [Sunday]
[Next Week]
[TTG
Messageboard] [HardwareGuys.com
Messageboard]
|
I'm still trying to recover from all the problems that were
caused by two machines failing. A lot of the problems that remain are
minor ones. I click on a Zip file and find that I haven't installed WinZIP
yet, or I click on an image and it comes up in a Microsoft program instead
of IrfanView. Or I attempt to download the raw web log data for
Pournelle's and my web sites, and find that FTP Voyager is pointed to a
drive on kiwi that's no longer on-line. Or I prepare to install a
program that I downloaded long ago, and then remember that it's on the 50
GB Barracuda that's still sitting in kiwi and therefore
unavailable. So I have to go find the program again, perhaps register with
the site (or enter my account information, which I don't remember and is
of course on the Barracuda.) All sorts of minor aggravations, but nothing
major at this point.
Except Outlook mail. My main personal store resides on theodore
(the server), so that's no problem. But all of the configuration data is
local, which means I now have a plain vanilla Outlook installation. All of
my rules are gone. I found something like 50 new mail messages in my inbox
this morning, so I spent half an hour or so recreating the most important
rules--those that move mailing list traffic to specific folders and so on.
Why oh why can't Microsoft store configuration information with the
personal store, or at least as a separate configuration file that could be
moved or pointed to just like the pst file? It is, as best I can
determine, impossible to migrate Outlook from one machine to another
without reconstructing everything one has done to customize the program.
And I have no tape drive on my personal system at this point. Outlook
keeps popping up a reminder to "backup Kiwi to DDS tape Week 1",
but that's not possible. I suppose I should install some kind of tape
drive in thoth (my new main workstation) but thoth is all
IDE. As long as we're tearing down thor (a Win98SE testbed) to steal its
Adaptec 2930U2 SCSI adapter for Barbara's new system, I suppose I'll also
pull the OnStream DI30 tape drive from it. That'll give me a 15/30 GB IDE
tape drive to stick in thoth.
I've started referring to Kerry as the "hall leopard".
Like a leopard, Kerry now waits for his prey to come to him. He'll lie
patiently for hours in the hall, waiting for prey he can drop on. He's not
very fast these days, but he still has a good set of fangs and a really
vicious-sounding snarl. When Malcolm is bored, he sometimes teases the
hall leopard. He'll lie just out of reach and then stretch forward to the
point where he's literally in Kerry's face. Then Kerry lets out his
vicious hall-leopard snarl and lunges at Malcolm, who withdraws in the
nick of time to avoid being chomped. Here's Malcolm in position to tease
the hall leopard.
Originally, Malcolm only did this when Kerry had something he wanted--a
toy Kerry was hogging or part of a dog treat or an ice cube that Kerry
hadn't finished eating. Now, though, I swear that Malcolm does it for fun.
One of these days, Malcolm will misjudge and we'll hear the loud snarling
and fighting noises that signify a furball in the hall.
Well, Barbara is taking Malcolm to the vet this morning for his regular
checkup and various shots, and I'd better get to work.
Click
here to read or post responses to this week's journal entries
[Top] |
Saturday,
16 December 2000
[Last
Week] [Monday] [Tuesday]
[Wednesday] [Thursday] [Friday]
[Saturday] [Sunday] [Next
Week]
[TTG
Messageboard] [HardwareGuys.com
Messageboard]
|
We finally got Barbara's new system built yesterday. We
named it sherlock, and it's one very fast system. It uses an Intel
D815BN motherboard, a Pentium III/1.0G processor, 128 MB of Kingston
PC133 SDRAM, an Adaptec 2930U2 SCSI adapter, and an 18 GB Seagate
Barracuda U2W SCSI hard drive. The motherboard has embedded video, sound,
and LAN adapters, which certainly makes it easy to build a system. For
now, sherlock just has an IDE CD-ROM drive.
We've not yet decided on a final configuration. We may boost memory to
256 MB, depending on how the system performs under Windows 2000 with only
128 MB. Barbara's personal system will now be just that, rather than
serving double duty as both client and server as her current system does.
We've not installed a tape drive, which is fine with Barbara. She prefers
to have her system backed up as a part of the overall network backup
anyway. But I'm uncomfortable with a system that has no means of local
backup, so I'll probably install an IDE Plextor PlexWriter CD burner,
which will at least allow her to backup her local registry and
configuration files to a CD-R or CD-RW disc.
Once we get the configuration finalized, I'll temporarily move sherlock
to my office for testing and benchmarking. I definitely want to find out
how the Intel 815 chipset compares to the 815E, and also how the embedded
video benchmarks with and without a GPA (Graphics Performance Accelerator)
installed in the AGP slot. The GPA is basically a $20 card that fits the
AGP slot and contains 4 MB of additional video memory. But with competent
AGP video cards available for $50 and "value" barn burners like
the GeForce2/MX available for about $100, it'll be interesting to see how
embedded video compares on a bang-for-the-buck basis.
I'm very pleased with the D815BN motherboard, and I hope that Intel
decides to make it available in retail-boxed form. I was also pleasantly
surprised by the Pentium III/1.0G processor. I expected it to run hot, but
that seems not to be the case. The processor arrived bundled with a large
heat sink and fan. With those installed, the heat sink is barely warm to
the touch even after the system has been running for some time. We wanted
to build Barbara a fast, stable system that would serve her needs for at
least the next year or 18 months, and it seems that we've achieved that
goal.
3dfx, one of the pioneers of modern 3D graphics accelerators, is
no more. Two or three years ago, the popular 3dfx Voodoo line of
3D cards was the product to beat, but a combination of poor marketing
decisions, an inability to develop new chipsets that were competitive in
performance with nVIDIA products, and delays in delivering even
those non-competitive products signed the death warrant for 3dfx. For the
last year or more, the story of 3dfx has been much too little and much too
late. This has been a simply incredible reversal of fortunes. Not all that
long ago, 3dfx owned the high-performance 3D chipset market, and nVIDIA
was teetering on the edge of bankruptcy. Now nVIDIA is
overwhelmingly dominant in the high-performance 3D chipset segment, and
has bought out its formerly robust competitor.
As weak as 3dfx has been lately, I still regret their departure from
the market. Very few real players remain standing in that segment. Matrox
has pretty much abandoned the high-performance 3D accelerator market.
Their Millennium G400 series were competitive cards when they first
shipped, but are no longer. The follow-on G450 is a good business card,
but not one that any gamer would seriously consider using. We keep hearing
rumors of the G800, but so far we have very little hard data about it. The
ATI Radeon is a competent 3D card, which can give nVIDIA products a
run for the money. So, in practical terms, we now have a two-horse race.
Still, I don't think there's much chance that nVIDIA will rest
on its laurels. Intel's embedded video initiative has really put
traditional graphics cards vendors under the gun. The simple truth is that
embedded video is Good Enough for about 95% of users about 95% of the
time. For a typical corporate/SOHO system, there's no need for anything
better than embedded video. Even for home multimedia systems, embedded
video is often sufficient. That leaves the traditional graphics cards
vendors with a rapidly shrinking market comprising only people who play
intensive 3D games and those who buy systems built on AMD processors. And
even that market will shrink as embedded video continues to improve and
integrated chipsets like the VIA KM133/266 become available for AMD-based
systems. The long-term outlook for graphics cards vendors is not good.
It's rather ironic. A few years ago, there was mainstream 2D video
which nearly everyone used and then there was the small niche market for
high-performance 3D video. The old-line mainstream video vendors like ATI
and Matrox missed the boat when the phenomenon of mainstream 3D video for
gaming began to become apparent. That allowed upstarts like 3dfx and nVIDIA
to establish a foothold in that market, and the efforts of ATI and Matrox
to dislodge them were largely unsuccessful. 3D video moved rapidly from
being a niche product to being a mainstream product, leaving ATI and
Matrox wondering what happened. Now we're on the point of another cusp,
when mainstream video (albeit competent embedded 3D mainstream video) will
again become the overwhelming choice of consumers, returning
high-performance 3D video vendors to niche status.
Once embedded video gets to the point where it provides 60 or 90 frames
per second in Quake--and that day will arrive sooner than most people
think--there won't be a lot of demand for standalone 3D accelerators. nVIDIA
will become a niche product line or, more likely, will be selling most of
their chipsets as embedded components. And that's not all bad. Building a
PC around an integrated motherboard has many advantages, including lower
cost, higher reliability, fewer compatibility problems, and so on.
And now I'd better get to work. I'm helping Barbara re-design
her diary page for the new year. I also have web stats to run for
Pournelle's and my sites, which isn't going to be easy given that the
month's and year's historical data resides on the 50 GB Barracuda in kiwi,
which isn't running at the moment. I need to rebuild kiwi as an NT4
domain controller (and I'll bet that my brief installation of Windows 2000
on it converted that 50 GB drive from NTFS-4 format to NTFS-5 format.
ARRGGHH). So perhaps I'll temporarily pull the 50 GB Barracuda from kiwi,
stick it in sherlock, and pull all the data off it. And that leaves
thor, the Win98SE test bed which started acting strangely the other
day. I guess year-end is a good time to be rationalizing the systems
around here, but I'm feeling like the proverbial one-armed paper hanger.
Click
here to read or post responses to this week's journal entries
[Top] |
Sunday,
17 December 2000
[Last
Week] [Monday] [Tuesday]
[Wednesday] [Thursday] [Friday]
[Saturday] [Sunday] [Next
Week]
[TTG
Messageboard] [HardwareGuys.com
Messageboard]
|
Brian Bilbrey
mails me to say that he wasn't familiar with the phrase "busier than
a one-armed paper hanger." His boss used it a couple of days ago, and
had to explain it to him. Then I used it yesterday, so Brian thinks it's
all a communist plot. His confusion was more understandable, though, once
he told me that they don't have wallpaper in California.
I'm feeling entirely whacked. I have a bad cold. I didn't get much done
yesterday, because I'd gotten only about four hours' sleep the night
before. Lack of sleep combined with feeling rotten from a cold just made
it hard to get much done. Then Barbara pointed out that we had a flood in
the basement. She was doing a load of laundry, and the washer leaked.
Well, it's the same age as our old dryer, so I thought perhaps we'd made a
mistake in not replacing both at the same time. But, just to make sure, I
told her to go ahead and run the second load, but this time to watch what
happened. As it turned out, the washer wasn't leaking at all. It was
working normally, but when it pumped out the used water, the standpipe
into which it pumps the water overflowed.
The standpipe is at one end of a four-foot long construct of PVC, which
enters the main drain at the other end. We've had problems with it before.
My mother's area downstairs is below grade, and so requires an ejector
pump to pump out waste water from the basement bathroom and kitchen. That
pump accumulates waste in a holding tank. When the float reaches a certain
level, the pump turns on and pumps the contents of the tank up and over to
the main drain. That pipe terminates at the main drain, very near to where
the washer standpipe enters. Apparently, the ejector pump pushes water
through the pipe so fast that the main drain can't accept it, and that
waste water was backing up through the standpipe and all over the washer,
dryer, and basement floor.
So we had our plumber come out and install a check valve to prevent
backflow. Apparently, that check valve has become clogged. I tried
yesterday to run a sink snake through the pipe, but there are too many
s-bends to get it through. So I came up with a cunning plan. I figured
that we needed some pressure. The water being pumped from the washer, you
see, is not under any pressure when it reaches the standpipe. I figured
that if we could put some water under pressure through that pipe it might
eject the clog. So Barbara connected the garden hose outdoors and we
dragged the business end of the hose into the basement. I held the hose
inside the standpipe and wrapped a rag around the top to seal it. When I
told Barbara to let 'er rip, the water flowed properly through the
standpipe and check valve. But when I eased off the pressure on the rag
even a bit, the water started fountaining out of the standpipe again.
Curses. Another cunning plan foiled.
So I called the plumber and explained the situation. He agreed that the
check valve was the likely culprit. He also mentioned that nearly all
check valves have provision for access to clean out clogs. I told him that
I was almost certain this one didn't, but that I'd check again and call
him back if it did. Otherwise, he's to show up here Monday morning. And I
have a pointed question to put to him: "Why, if nearly all check
valves make provision for cleaning, would you have installed one that
doesn't?"
This is starting to get embarrassing. Last night, meepmeep
locked up again, requiring a hard reset. And then it did it again less
than ten minutes later. So here I am, a writer of books about PC hardware,
with a bunch of PCs that are misbehaving. Barbara's current main system is
doing some weird things, but those problems are almost certainly due to
the well-known phenomenon of Windows-rot. So we start building a new main
system for her. While we're doing that, thor (the Win98SE machine
that lives under my desk) starts behaving very oddly, refusing to run
programs. Again, that sounds more like Windows-rot than a hardware
problem, but even so. Then kiwi, my own main system, crashes, and
that almost certainly is the result of a hardware problem, probably from
overheating. Then meepmeep starts crashing, and that once again is
almost certainly a hardware problem, probably a bad power supply. But then
meepmeep is built on a no-name Pacific Rim case and power supply,
so I can't say I'm surprised.
So here's what I'm going to do:
- Turn kiwi into a Windows NT 4 Server machine. Strip
down kiwi, removing the 18 GB Seagate Cheetah 10,000 RPM SCSI
drive, the Plextor 8-2-20 SCSI CD writer, the Plextor UltraPlex 40X
SCSI CD-ROM drive, and the Hitachi ATAPI DVD-ROM drive. Also remove
128 MB of the 256 MB currently installed. Leave the 50 GB Seagate
Barracuda 7,200 RPM SCSI drive and the Tecmar 3900 DDS-3 tape drive
installed. Install an ATAPI CD-ROM drive. Dual Pentium III/550
processors, the Adaptec 2940U2W SCSI host adapter, and the Matrox
Millennium G400 video card are all overkill for this system, but it's
easiest to leave them where they are, so I will. Install Windows NT
Server 4.0 and SP6a. If it turns out that my abortive attempt to
install Windows 2000 Professional on this system in fact converted the
filesystem on the 50 GB Barracuda from NTFS-4 to NTFS-5, I'll
temporarily move the Barracuda to Barbara's new Windows 2000 system
long enough to get all the data on that drive moved over to another
network drive before I reformat it for NTFS-4. Once kiwi is
configured as an NT4S BDC, we'll connect it to the network and promote
it to PDC, taking over for theodore, Barbara's current main
system.
- Finish configuring sherlock (Barbara's new machine). Install
the Plextor PlexWriter 8-2-20 CD writer and the Plextor UltraPlex Wide
40X CD-ROM drive in sherlock. The Adaptec 2930U2 SCSI adapter
in that machine can handle these two optical drives as well as the 18
GB Seagate Barracuda LVD hard drive without any problems. Install
Windows 2000 Professional.
- Refurbish and rebuild theodore as the Internet gateway
machine. Theodore already has Windows NT 4 Server
installed, and will be demoted to a BDC once we bring up kiwi
as the PDC. Then all I need to do is transfer over the cable-modem
Ethernet card from meepmeep, install WinGate, and configure
everything. Come to think of it, I'd better check and record the
current configurations of everything--hardware and software--before meepmeep
finally dies. A Pentium III/450 and 128 MB seems overkill for the
Internet gateway machine, but that's the easiest way to handle things
without an insane amount of juggling. I considered just buying one of
the inexpensive cable-modem routers like the LinkSys BEFSR11 EtherFast
1-Port Cable/DSL Router, which Onvia is selling for $90 after a $10
rebate. But I didn't want to get into the mess that's likely to result
if Time-Warner "married" the MAC address of the current
Ethernet card to the cable modem, as I suspect they did. Also, there
are a surprising number of times when it's necessary to use the
machine that is directly connected to the Internet rather than a
system that connects through the proxy. I can't ping or do a
traceroute except from the directly-connected machine, for example,
and the program I run to do DNS lookups for web statistics won't
operate through a proxy. So it makes sense to have a fairly competent
system as the Internet gateway.
Barbara is cleaning house as I write this. Ordinarily, I'd be doing
laundry, but that's not possible unless I want to flood the basement with
each load. We're going to just stay in today and try to recover. Barbara
has had a sinus infection which is still nagging her. We won't be going
out much today. Our weather is really strange. Yesterday, rain and
temperatures approaching 50F (10C). Last night, a thunderstorm. This
morning, it's supposed to snow. And later today we're expecting a rapid
drop in temperature, high winds, and a wind chill of 10F (-12C) or
thereabouts. Not a good day to be outside.
Click
here to read or post responses to this week's journal entries
[Top] |
|