Email
Robert |
Daynotes
Journal
Week of 20 March
2000
Friday, 05 July 2002 08:11
A (mostly) daily
journal of the trials, tribulations, and random observations of Robert
Bruce Thompson, a writer of computer books. |
wpoison
Search [tips]
TTG Home
Robert Home
Daynotes Home
Links
Special
Reports
Current Topics
|
Jump to most recent
update
Monday,
20 March 2000
[Last
Week] [Monday] [Tuesday]
[Wednesday] [Thursday] [Friday]
[Saturday] [Sunday] [Next
Week]
I got some work done yesterday morning on Chapter 2, Working on PCs,
but I finally ran out of gas in the afternoon and ended up taking a nap
for three hours or so. I'll be back working on that chapter today and
perhaps tomorrow. After that, I get into updating the processors and
motherboards chapters, which is going to require building and testing some
new systems with the stack of processors and motherboards I have sitting
around here.
Howls of outrage this morning from Barbara. Malcolm thieved one
of her new bras and wouldn't give it back. I heard her shout for him to
stop pulling on it. He's always proud when he captures something, and
loathe to give it up. When she finally did get it away from him, more
howls of outrage. He'd chewed holes in it. I'm convinced that puppies
being cute is a survival mechanism. If they weren't, many would be
murdered very young. At least Malcolm gets his stitches out today, which
may improve his behavior. We're taking him to the vet this afternoon for
that, and also taking the other two for their annual checkups.
Back to work.
|
wpoison
Search [tips]
TTG Home
Robert Home
Daynotes Home
Links
Special
Reports
Current Topics
|
Tuesday,
21 March 2000
[Last
Week] [Monday] [Tuesday]
[Wednesday] [Thursday] [Friday]
[Saturday] [Sunday] [Next
Week]
One problem solved. I'd been having horrendous problems with my new IDE
test-bed system, particularly under Windows NT 4. I'd never seen so many
bluescreens in my life. I tried everything I could think of, and nothing
worked. The system uses first-rate components--Intel CA810e motherboard,
FC/PGA Coppermine Pentium III processor, 128 MB of Crucial memory, Maxtor
hard disk, Antec case and power supply, etc. I finally decided I must have
a defective motherboard, a bad power supply, bad memory, or something
similar. So yesterday I determined to track down the problem.
To make a long story short, it wasn't Crucial memory. It was commodity
SDRAM, PC100 so-called, that was causing the problem. What happened was my
own fault. At some point, I was installing memory in a system, and I must
have pulled that commodity SDRAM and installed two Crucial PC100 64 MB
DIMMs. I apparently ended up putting the two pulled 64 MB
"PC100" commodity DIMMs in the Crucial anti-static bag, and when
I built this system I installed those commodity DIMMs, thinking from the
label on the bag that they were Crucial. They weren't.
I replaced the memory in that system with a 64 MB Crucial PC100 DIMM
that I pulled from another system. Guess what? The problems disappeared
entirely. I had that system benchmarking all day yesterday under both
Windows 98 and Windows NT 4 with nary a hiccough. I periodically get
messages taking me to task for recommending Crucial memory so strongly.
Commodity memory is just as good, they claim. I know it isn't. This isn't
the first time, either.
I need to be writing, but I need to be doing this stuff as well. I'll
start re-write on the motherboards chapter soon, and I have to get a
handle on current Intel motherboards. Both the 810e and 820 chipsets have
come under a lot of criticism, and I need to find out whether it is
justified. The 810e and 820, for better or worse, are Intel's mainstream
desktop chipsets for the next year or more, so I need to find out how they
compare to each other and to the 440BX. I've established a 440BX, 100 MHz
FSB, SDRAM baseline, so now I have to do some serious benchmarking and
other testing to see how the 810e and 820 measure up.
I have a VC820 and a CC820 motherboard, so I can test the 820 with both
RDRAM and SDRAM. My guess is that the 820 with RDRAM will be marginally
faster than the 440BX with PC100 SDRAM, which in turn will be marginally
faster than the 810e, which in turn will be marginally faster than the 820
with SDRAM. If the differences are only a few percentage points, that's
not worth worrying about. But I've seen various reports claiming that the
820 with SDRAM is anything up to 40% slower. We'll see.
* * * * *
-----Original Message-----
From: McDonell @ The Park [mailto:mcdonell35@earthlink.net]
Sent: Monday, March 20, 2000 8:44 PM
To: thompson@ttgnet.com
Subject: HP Printer & Camera
In my notes the other day, I mentioned that
I had not heard back on the $100 rebate for buying two of that HP
"PhotoSmart" group of products. My wife informed me yesterday
that the check had come in "last week". It took 60 days but we
received a rebate!
The HP P-1000 printer began giving off
distress signals last weekend. Seems that the ink pens/ cartridges were
"dangerously" low on ink. WHAT? Those pens used to last us a
year. Maybe they had holes in them or HP Utility programs are
exaggerating the levels to stimulate sales. They did achieve the latter.
I found the black one at our new Target Store. $21.99 + tax and license.
Hmm - there was another on display that was the same series (51645
"Series") but cost $24.99 or so. It took time to sink in but
the word "Series" is significant. Something is new out there.
When I got both in my hands, one was marked "G" and the other
"A". Both listed P-1000 applications. The "G" box
had a pretty little flag waving near the top of the package, message
"...for occasional printing..."Whaaaaaaat? Oh, there it was,
"G" contains 21 ML and "A", 42 ML. At the list
price, one was $1.05 / ML, the other $0.56. I am going to buy some stock
in HP. They really know how to merchandise. So does Target. Those pens
are kept in different displays about 20 feet apart; the "G"
types get to be closest to the Printer Display.
Target did not have the color pen HP C6578
in stock. Dern. On the way to Office Depot, I stopped at Wal-Mart to
check prices. The HP 51645A was priced at $29.95 or $0.713 /ML. That is
20% higher than Target. Wal-Mart is a Dow Jones 30 company so I guess we
are supposed to be grateful.
Onward to Office Depot, where the HP 51645A
was also priced at $29.99. No "G" in stock. They did have the
color pen - Mfr List $39.99 but I fainted before they could ask me how
many I wanted! Indeed! Just like gasoline, get them while they last.
Then at the cash register, they "marked it down" to $32.99. I
revived in time to pay.
All three of those stores noted above are
within a 2 mile stretch in Carson City; which is not a metropolis.
Still, I think we need a Fry's. Although they are not noted for
discounting, their showbiz stores make it a lot of fun while you spend.
The Fry's down near Jerry Pournelle, in Burbank, is designed to look
like a gigantic mother ship.
Target has a nice selection of lower end
scanners and printers by HP, Canon & others. They also have a lot of
software for sale there. Software prices are pretty much
"equalized" as you should know. That is where I encountered
the HP 3200. I like your analysis of cost for your HP 6200. I wanted
that one badly but by the time I was in the market for it, only the HP
6250 (with paper feeder) was available. Thus, the HP 5200; which my wife
has learned to use for E-mailing copies of articles to unsuspecting
friends.
Our HP 200 digital camera has had the
equivalent of a dozen rolls of 35mm film "run through it" in
the 8 weeks we have used it. Its Compact Flash card holds 40 images at
average resolution. I think I am detecting some lens deficiencies but I
need new glasses. I noticed that HP offers a selection of detachable
lenses for it, about the size of contact lenses! Imagine juggling those
things out on a lawn. We are going to throw that camera away when it
pays for itself; that should be about the middle of July at the current
rate. The camera lacks "JetSend", an infrared image transfer
technique that the P-1000 does have. If it works, that would help get
rid of cable transfers.
Regards,
Maurice McDonell
That's exactly why I don't own an inkjet printer. The cost of
consumables is horrendous. Inkjet manufacturers follow the King Gillette
model of giving away the razor and selling the blades. There's absolutely
no reason why ink should not be sold in inexpensive quart squeeze bottles,
and print heads sold as a separate item. In fact, that's how very
expensive inkjet printers do it. And their cost-per-page is a tiny
fraction of that of the "consumer" inkjet printers. I'd just as
soon pay $400 rather than $200 an inkjet printer in return for a
consumables cost of $.02 per page instead of $0.22 per page. The
difference pays for itself in a thousand pages.
As far as image transfer, consider buying one of the USB card
readers that accepts CompactFlash cards (some also accept the SmartMedia
cards used by Olympus and others). You'll find that you can transfer your
images in less than a tenth the time that serial requires.
* * * * *
-----Original Message-----
From: Chuck Waggoner [waggoner at gis dot net]
Sent: Tuesday, March 21, 2000 6:59 AM
To: Robert Bruce Thompson (E-mail)
Subject: W2k and O2k SP1
So, what have you decided to do? Are you
going to register as apparently mandated by MS? or have you found a way
around all that?
--Chuck
Probably not much. I understand that the registration simply
requires providing your country and the init key, whereupon they return
the remaining portion of the init key required to let the product run. I
don't have a lot of problem with that, so long as they're not accumulating
personal data about me. On the other hand, I can't see what good that
process will do them.
The only Office 2000 applications I really use are FP2K and OL2K.
I'm still using Word 97 and Excel 97, and I could revert to FP98 without
much problem. I could also revert to OL98 if I could get it to install. I
have all the install files on a local disk, from which I have installed
OL98 many times, but that no longer works. The install process insists on
going out to the Internet to download files it says it needs, which can't
be found. I'm not entirely convinced that Microsoft didn't put some sort
of time out in the OL98 distribution files.
As far as Windows 2000, I'm not sure I'll deploy it in a
production environment. I need it to write books with, but since it allows
up to 50 boots without registration, I may just install it and use
DriveImage to clone it to unused space on another partition. That way, I
can simply start easily with a fresh install if and when I reach the 50
boot limit.
I'm not happy about the customer-control aspect of all of this. I
may eventually switch to Linux.
|
wpoison
Search [tips]
TTG Home
Robert Home
Daynotes Home
Links
Special
Reports
Current Topics
|
Wednesday,
22 March 2000
[Last
Week] [Monday] [Tuesday]
[Wednesday] [Thursday] [Friday]
[Saturday] [Sunday] [Next
Week]
WARNING
Do NOT install Office 2000 SR-1
until further notice. Read this tale
of woe posted by Tom Syroid, co-author of Outlook 2000 in a
Nutshell. The story starts on the Tuesday link. Scroll down to the
paragraph that starts with "Warning" in red. This is not a
pretty picture. It appears that SR-1 trashes the entire Office 2000
installation.
Now here's a sinking feeling. Yesterday afternoon I was working on
Chapter 2, Working on PCs. I needed to verify the procedure for
making a Windows 2000 Emergency Repair Disk, so I fired up the machine
that has Windows 2000 installed on it. During the process, Windows 2000
prompted me to insert a diskette. I picked a diskette up from the recycled
stack, expecting that Windows 2000 would format the diskette before
copying the ERD data to it. It didn't. It just copied the data. Okay, good
enough.
I wanted to test the diskette to make sure that, like the Windows NT 4
ERD, it wasn't bootable, so I told Windows 2000 to restart, expecting to
get an error message when it tried to boot to the diskette. Instead, the
system booted and displayed a message saying that the BIOS had been
updated. ARRRGHHH. That turned out to be the scratch diskette that I used
to update the Intel CA810e motherboard the previous day.
The message said to turn off the system and restart it to complete the
BIOS upgrade. I didn't have much hope, given that this system runs an EPoX
motherboard with a 440LX chipset and a completely different BIOS, but I
didn't see much alternative so I shut down the system. When it came back
up, it rebooted normally, although it did say CMOS was corrupt. I told it
to reset CMOS to defaults, restarted, and everything came up normally. I
thought I'd trashed a motherboard, but I was lucky.
In the who-cares department, I see that Netscape plans to begin
shipping betas of their new browser sometime in the next three weeks or
so. I can't imagine many people are likely to bother with it, except
perhaps those running an OS for which IE is not available. And most of
them are probably waiting for Opera. Let's see. The last time Netscape
shipped a new browser was what, three years ago? And that one is still
in beta, or so it appears. Netscape has the nerve to call this allegedly
forthcoming version Netscape 6, as though no one will notice that they
never bothered to ship Netscape 5 (nor, some would argue, Netscape 4).
They probably should have called this one Netscape 4.0, because from what
I've heard it won't even match the capabilities of IE5.
I'll probably download it once it reaches a so-called release version
and load it on a machine I don't care about. Not that I expect much from
it. Even if does what they promise, it's much too little and way too late
to help Netscape. If Microsoft had any sense they'd have released IE for
Linux long ago. If they get around to doing so now, it should kill any
chance Netscape has for a revival.
If you haven't read the warning
from this morning, please do so now.
10:45: Despite
Tom Syroid's problems with Office 2000 SR-1, and despite the fact that it
supposedly implements an install counter, I decided to get it. If nothing
else, I could install it on a scratch system and see what happens. When I
visited the web site, I noticed that there's apparently no option to
download a static copy of the SR-1 update that would allow me to store it
on a network drive and install it as needed. No, the only option is a
dynamic download, which means I'd have to download 26 to 40 MB each time I
wanted to update a system. No thanks.
Well, there is a static option, but it's the administrative update
version of SR-1, which can be applied to an administrative installation of
Office 2000 on a server, but doesn't work with a standard installation.
Then I noticed that Microsoft offers the SR-1 upgrades, both normal and
administrative versions, on a free CD, available here.
I went over to that URL and signed up to get the free CD, which in fact is
completely free. They don't even charge for shipping. But when I clicked
on the final button to submit my order, the server returned the following:
Server object error 'ASP 0177 : 800a004c'
Server.CreateObject Failed
/office2k_sr1/main_office2k_sr1.asp, line 617
The operation completed successfully.
So apparently, the operation both failed and completed successfully.
Hmmm. I used my Back button to return to the preceding page and resubmit
the order. The same thing happened. So it appears that six to eight weeks
from now I'll either get zero or two copies of the SR-1 CD.
I just got a call from Tim Laszly, our mechanic. Barbara leaves
on a business trip to Atlanta next month, and wanted to get the oil
changed and all the other stuff checked before she left. I followed her
out to the mechanic's place this morning. She dropped me at home and
headed for her parent's house in my Trooper. A little while ago, she
called to tell me that she'd noticed a nasty looking cut in one of the
tires, which are due for replacement anyway. I told her to talk to Tim
about whether to get four or five tires and what brand to put on. We'd
tried to get some Dunlop models the last time we bought tires, but they
weren't available in the size we needed.
A few minutes later, Tim called me to say he needed both trucks,
because he's going to trade the relatively new tires from Barbara's truck
to my truck and put the new ones on Barbara's truck. I asked him about
four versus five, and he suggested just getting four. I then asked him
what make he was planning to put on. He asked if I had a preference and I
mentioned the Dunlops. As it turns out, he runs Michelin ATXs on both of
his 4X4s, and recommended them for us. Good enough for me. I don't know
tires, and if he uses them himself that's a strong recommendation.
The only downside is that the Dunlop BabyKillers™ get pretty good
reviews for price and performance, and I'd promised myself not to install
Michelin tires. First, I hate their ads. (Those ads are why I refer to all
non-Michelin brands of tires as "BabyKillers"). Second, they're
French, and the French are not known for making anything well, except
perhaps wine. Well, Tim's advice is good enough for me, so I guess we'll
end up with Michelin tires on Barbara's truck.
|
wpoison
Search [tips]
TTG Home
Robert Home
Daynotes Home
Links
Special
Reports
Current Topics
|
Thursday,
23 March 2000
[Last
Week] [Monday] [Tuesday]
[Wednesday] [Thursday] [Friday]
[Saturday] [Sunday] [Next
Week]
More on the Office 2000 SR-1 mess over on Tom
Syroid's site. Apparently, the problem he reported yesterday isn't
widespread, although it has been reproduced by another user. Pournelle
called me yesterday to say that the bug seems to occur only on
unregistered O2K installations. If true, that's a strong inducement to
register, and is perhaps not coincidental.
I regard forced registration in the same light as copy protection. It's
simply unacceptable. I've railed in the past against Intuit, which is one
of the worst proponents of forced registration and similar
customer-control measures that I know of, and I'm sorry to see Microsoft
headed down this road. Tom Syroid mentioned on his site today that
O'Reilly has an author's template for StarOffice, which is intriguing. I
have StarOffice installed on this computer. I didn't particularly like its
consolidated desktop paradigm, but that's a relatively minor issue.
StarOffice replaces Word and Excel. That means the only Microsoft
applications I use that I need good replacements for are FrontPage and
Outlook. Magellan may well be an Outlook replacement, but that leaves
FrontPage. Unfortunately, that's problematic, because Barbara uses
FrontPage to maintain her web pages, which are on the same web site as
mine. Before I commit to StarOffice, I'll probably check out Corel Office,
which I understand is due imminently. But my goal is eventually to be able
to do my work without depending on any Microsoft applications.
Barbara is off to play golf with her father and run errands. Her
truck wasn't finished yesterday, so we'll need to pick it up this
afternoon. I'm working on the processors chapter, so I'd best get back to
it.
* * * * *
-----Original Message-----
From: Scott Kitterman [mailto:scott@kitterman.com]
Sent: Wednesday, March 22, 2000 3:06 PM
To: thompson@ttgnet.com
Subject: Anti-Cookie Privacy Tool
I'd be curious to read your thoughts on this:
Scott Kitterman
scott@kitterman.com
It looks like just another cookie blocking tool to me, albeit
with less functionality than many. From what I read, it appears to use a
mechanism similar to that used by Netscape with its Accept cookies only
from originating site option. That method is less than reliable, I've
found, on sites that use invisible frames to redirect your browser to
Imgis, Doubleclick, and so on. If I were going to use a cookie blocking
tool, I'd use one of the ones that intercepts a call to write a cookie and
examines the source of that cookie before doing so.
As it happens, I don't feel the need to use a cookie blocking
utility. I simply use the zone feature of IE to set my Internet zone for
no persistent cookies, etc. and my trusted zone to allow them. That way,
sites I visit casually cannot write cookies, but those I use regularly and
have reason to trust are added to the Trusted zone. There, they can write
cookies, so I don't lose functionality.
|
wpoison
Search [tips]
TTG Home
Robert Home
Daynotes Home
Links
Special
Reports
Current Topics
|
Friday,
24 March 2000
[Last
Week] [Monday] [Tuesday]
[Wednesday] [Thursday] [Friday]
[Saturday] [Sunday] [Next
Week]
I have been informed by a reader that Office 2000 SR-1 is now available
for download as a discrete file here.
You can download this 52 MB file, store it on a network drive, and use it
to update any number of Office 2000 machines. I downloaded it
yesterday, but I haven't gotten around to installing it yet. I won't
install it on any of my production systems, and I don't have a testbed
system with Office 2000 installed on it.
Intel ships the first Celeron III processors next week, although
I doubt they'll call them that. These new Celerons will initially ship in
566 and 600 MHz versions, and are really just modified FC-PGA Coppermine
Pentium IIIs with L2 cache halved to 128 KB and rated for use on a 66 MHz
FSB. Despite speculation by Anand and Tom, these first units will likely
not be good overclockers. In effect, they're simply down-rated Pentium
III/850 and Pentium III/900 chips, respectively. My guess is that they're
really Pentium III units that failed testing at 100 MHz FSB. So
overclocking them by boosting the FSB from 66 MHz to 100 MHz would simply
give you the same processor that Intel tested as unreliable at 100 MHz.
That'll change as Intel's yields on Coppermines improve, but only on
the low-end. The only thing that matters to a processor is how fast it
runs. It doesn't care what the FSB or clock multiplier is, just what the
product of those two is. If a processor is stable at, say, 900 MHz, it
doesn't care whether you achieve that 900 MHz by using a 100 MHz FSB and a
9.0 multiplier or a 66 MHz FSB and a 13.5 multiplier.
But indications are that 1 GHz is about the limit of the Coppermine
core. The 1 GHz Coppermine, for example, requires higher than nominal
voltage, and does not support SMP. Those two things tell me that 1 GHz is
at or beyond the design capability of Coppermine. Intel's original
intention was to parcel out faster Coppermines gradually, hitting 1 GHz in
H2/2000, just in time for Willamette to come on stream near the end of
this year. Obviously, AMD forced Intel's hand, and so they ended up
introducing all of their Coppermines in Q1/2000. That means that the rest
of this year is likely to see only minor, if any, increases in Pentium III
clock speeds. What will see is falling prices and increased availability
as Coppermine production ramps up.
But the implication for the Celeron III is that later variants will not
be overclockable in any real sense because of their locked multipliers.
For example, Intel will probably ship a Celeron/700 later this year. That
processor will use a 66 MHz FSB and a 10.5 multiplier. Attempting to
overclock that processor at 100 MHz FSB would run it at 1,050 MHz, which
is probably beyond the ability of the Coppermine core.
However, as Coppermine production ramps up, Intel will find itself in
its usual position. All of its Coppermine cores will be reliable at, say,
900 MHz. That means that, later this year, a Celeron/566 or /600 will be
an excellent overclocking candidate, since it will really be a Pentium
III/900 in disguise.
It will also be interesting to see if the Celeron III uses the PPGA
Socket 370 pinouts or the FC-PGA Socket 370 pinouts. If the former, a
Celeron III could be run in an older Socket 370 Celeron motherboard. If
the latter, it could be run only in one of the later Socket 370
motherboards like the Intel CA810e. My guess is that it will be the
latter.
* * * * *
-----Original Message-----
From: john biel [mailto:johnny51@home.com]
Sent: Thursday, March 23, 2000 7:13 PM
To: webmaster@ttgnet.com
Subject: netscape beta, or more accurately mozilla milestone release
I would dispute your opinion on this one. I
have used Mozilla releases on both Win95/98 and
Linux(Caldera2.3/Corel1.0) While they are definately beta, they are
already in my opinion far superior to opera. I generally use and like
ie5, but I would bet that by summertime I will probably start using the
"release" version of mozilla, it is faster in rendering,
smaller in size, and interestingly enough can handle some intentional
"browser-busting" sites that blow ie5 up. At present mozilla
does not have any java capability (although it can do javascript) so I
would think you might actually like to try it on that basis alone as you
have made the point that you keep these things turned off anyway. On
Linux I have already dumped the versions of netscape I had and use the
current daily build of mozilla
exclusively. On win95 I often have both up at the same time and find
mozilla much more responsive than ie5, but not yet as stable. Updates to
mozilla especially over the past month have increased it's usefulness
phenominally, assuming that continues, I'll certainly switch. It's about
5 or 6 meg, but I'll bet that get's reduced as debugging code gets
removed.
And you may be right. That's why I said I'd download it and take
a look at it. But I'm not expecting much.
* * * * *
-----Original Message-----
From: Chuck Waggoner [waggoner at gis dot net]
Sent: Thursday, March 23, 2000 9:20 PM
To: Robert Bruce Thompson (E-mail)
Subject: Cookies, etc.
After reading of your IE security set-up
early in the year, I adopted it, and after more than two months of
operating with it, I can report finding no real problems at all with
using it that way. Everything is disabled from cookies to Java,
scripting, and Active X on the standard "Internet" zone, and
all those are enabled in the "Trusted" zone.
During that period, I've been very active on
the Web, doing quite a bit of research for a project, and have found
only a very few 'bleeding edge' sites which don't work--and those
weren't likely to offer anything I needed anyway. Even Travelocity works
fine without them (although I already had their cookie before I made the
changes).
In fact, I've only needed 3 entries in the
Trusted zone: our financial institution, NYTimes, and Microsoft (hardly
anything works on the Microsoft sites without all those whiz-bangs
activated).
I've not seen it admitted on the Microsoft
site, but have read in many other places, that running with Java and
Active X enabled is just asking for trouble. Experts appear to agree
that it is quite possible for a combination of calls using Java and
Active X to penetrate a computer, and this can be happen by just
visiting a hacked site with those controls enabled. A recent CERT
advisory noted that hackers are embedding code in return fields of forms
appearing on some sites, and in some cases, that code is then passed on
to other innocents who view the site, intruding into their computer in
undesirable ways.
It's hard to imagine that there will be less
hacking attempts out there, as time flies by, so safety seems more and
more important these days.
--Chuck
I agree, which is why I haven't felt the need to install one of
those third-party cookie managers. Unfortunately, quite a few of the sites
I visit were designed by clueless webmasters who use ActiveX controls, and
IE has no way to turn off that damned ActiveX warning. Some sites go to
ridiculous extremes, for example by using JavaScript in place of simple
HTML links. Those sites I just don't visit again.
* * * * *
-----Original Message-----
From: Ray Watson [mailto:rwatson@autolinq.com]
Sent: Friday, March 24, 2000 8:16 AM
To: webmaster@ttgnet.com
Subject: Broken Links
Hello Robert,
I was just reading your daynotes page and
found some broken links on it (the current page). The links for :
Robert Home
Daynotes Home
Links
Special Reports
Current Topics
are broken. I think they are missing the /
part of the path.
Have you noticed on win2k that IE5
(5.00.2920.000) no longer has the option to launch in separate process.
The install automatically sets this ( separate process if >32megs,
otherwise same process if <32megs). I like the separate process bit
but don't appreciate the hit on resources. I don't know if you or any
other readers have a hack to change this but I would appreciate it if
someone did. I have 128 megs of ram but am almost always using 200 megs
according to task manager, so I would benefit by not having separate
processes running for 2 or 3 three IE instances.
By the way, I really enjoy your writing and
am looking forward to reading your new book.
________________
Ray Watson
AutolinQ Internet Services
Thanks for the kind words. I was about to blame FrontPage 2000
for the broken links, but they may in fact be my own fault. Or maybe not.
I'm really not sure. They should be fixed now.
What happened is this: a couple of weeks ago, I decided to change
my left column menu to use the same method I've been using on Barbara's
and other pages here, which is to use FrontPage's "Include Page"
function. That is, the left column menu is actually a separate HTML file,
and my daily journal page includes a pointer to it. That way, if I ever
want to change the left column menu, I can simply change that one included
file, and it automatically updates all the individual daily journal pages.
When I got your message, I went back and checked the include
page. Sure enough, some of the links were screwed up. Oddly enough, some
were not. One thing I will give FrontPage credit for: it is very good at
maintaining links. When you move, delete, or rename a page, it
automatically updates all linked pages with the new information. And I
think it did so in this case, which is what makes me think that the error
was probably mine.
The problem is that the include page resides in the folder /rbt,
whereas the daily journal pages reside in /daynotes/2000. When I edit the
include page, all links on it are relative to the folder /rbt. But when
FrontPage incorporates it into the daily journal page, it fixes all the
links to be relative to /daynotes/2000. Or it should. It apparently fixed
some, but left some pointing to positions relative to the actual location
of the include page rather than relative to the location of the page in
which the include page was included.
Since FrontPage has never mangled links like this, I suspect I
might have done something to cause the problem. At any rate, I went back
and fixed all the links in the include page, and they should be fixed on
the real site once I publish the changes.
* * * * *
-----Original Message-----
From: Ray Watson [mailto:rwatson@autolinq.com]
Sent: Friday, March 24, 2000 9:37 AM
To: thompson@ttgnet.com
Subject: Re: Broken Links
Aren't include pages wonderful? I could not
exist without them on the web sites that I build and maintain. I started
out using Frontpage but it mangled websites too often for me. ( At the
time I was publishing to a unix server and the case insensitive
frontpage98 kept capitalizing my links for some reason. Not good to say
the least!) I took the plunge and learned HTML and started to use
notepad. I quickly jumped to Homesite and have never looked back. I
realize that for someone not in the web page biz that hand coding can be
tedious but, you probably waste enough time waiting for FP to publish
etc that it could be interesting to compare. One thing with hand coding,
the only thing that will get changed is what you change. ( and actually
the global search and replace in Homesite is very good - you can change
links or whatever across the whole site)
Have a good day,
Ray
The sites I maintain all are at www.auto123.com
( you will find links here to the 250 - 300 sites that I maintain).
I fear that I don't have the time to maintain my sites manually,
even with such aids as global search and replace. And FrontPage 2000
really is much better about not mangling things than FP98 was. The only
problem I've had with FP2000 (which was also a problem with FP98) is that
it periodically decides to delete my bottom shared margin, thereby killing
my copyright notice. Microsoft is aware of the problem, but has not been
able to explain it. Other than that, I haven't really been bitten by any
FP2K bugs. There are still a lot of things I would change about it, but
it's the best thing I've found for casual webmasters like me.
* * * * *
-----Original Message-----
From: Ray Watson [mailto:rwatson@autolinq.com]
Sent: Friday, March 24, 2000 10:04 AM
To: thompson@ttgnet.com
Subject: Re: Broken Links
When the tool fits... it's the right one for
the job! ;-) I have not tried FP2000, so no comments from me, but from
an ease of use point of view, FP definitely makes web sites easier for
the casual or part time webmaster.
Have a good day
Ray
Which I definitely am. You should probably give
FP2000 a try. I'm no HTML expert by a long shot, but I've talked to people
who are, and they tell me that FP2000 is greatly improved from their point
of view. Apparently, it doesn't stick all the excess code in that FP98
did, and it doesn't mangle hand-coded stuff. A couple of HTML coders tell
me that they use FP2000 to rough out a site quickly and then hand-code to
tweak it.
* * * * *
-----Original Message-----
From: Barbara F. Thompson [mailto:barbara@ttgnet.com]
Sent: Friday, March 24, 2000 9:22 AM
To: Bob (E-mail)
Subject: Dog Dictionary
You probably won't think this is funny, but
I think Malcolm and Duncan could have been the lexicographer.
LEASH: A strap which attaches to your
collar, enabling you to lead your person where you want him/her to go.
DOG BED: any soft, clean surface, such as
the white bedspread in the guest room or the newly upholstered couch in
the living room.
DROOL: Is what you do when your persons have
food and you don't. To do this properly you must sit as close as you can
and look sad and let the drool fall to the floor, or better yet, on
their laps.
SNIFF: A social custom to use when you greet
other dogs. Place your nose as close as you can to the other dog s rear
end and inhale deeply, repeat several times, or until your person makes
you stop.
GARBAGE CAN: A container which your
neighbors put out once a week to test your ingenuity. You must stand on
your hind legs and try to push the lid off with your nose. If you do it
right you are rewarded with margarine wrappers to shred, beef bones to
consume and moldy crusts of bread.
BICYCLES: Two-wheeled exercise machines,
invented for dogs to control body fat. To get maximum aerobic benefit,
you must hide behind a bush and dash out, bark loudly and run alongside
for a few yards; the person then swerves and falls into the bushes, and
you prance away.
DEAFNESS: This is a malady which affects
dogs when their person want them in and they want to stay out. Symptoms
include staring blankly at the person, then running in the opposite
direction, or lying down.
THUNDER: This is a signal that the world is
coming to an end. Humans remain amazingly calm during thunderstorms, so
it is necessary to warn them of the danger by trembling uncontrollably,
panting, rolling your eyes wildly, and following at their heels.
WASTEBASKET: This is a dog toy filled with
paper, envelopes, and old candy wrappers. When you get bored, turn over
the basket and strew the papers all over the house until your person
comes home
SOFAS: Are to dogs like napkins are to
people. After eating it is polite to run up and down the front of the
sofa and wipe your whiskers clean.
BATH: This is a process by which the humans
drench the floor, walls and themselves. You can help by shaking
vigorously and frequently.
LEAN: Every good dog's response to the
command "sit !", especially if your person is dressed for an
evening out. Incredibly effective before black-tie events.
BUMP: The best way to get your human's
attention when they are drinking a fresh cup of coffee or tea.
GOOSE BUMP: A maneuver to use as a last
resort when the Regular Bump doesn't get the attention you
require.....especially effective when combined with The Sniff. See
above.
LOVE: Is a feeling of intense affection,
given freely and without restriction. The best way you can show your
love is to wag your tail. If you're lucky, a human will love you in
return.
|
wpoison
Search [tips]
TTG Home
Robert Home
Daynotes Home
Links
Special
Reports
Current Topics
|
Saturday,
25 March 2000
[Last
Week] [Monday] [Tuesday]
[Wednesday] [Thursday] [Friday]
[Saturday] [Sunday] [Next
Week]
I just processed the web stats for Pournelle's site and my own, as I
usually do on Saturday mornings. Interestingly, my own stats are beginning
to creep back up. Last week, I ran 1,100 to 1,200 page reads per day for
Monday and Tuesday, 700 to 900 pages/day for the other weekdays, and 500
to 700 pages/day on Saturday and Sunday. That's still off a bit from its
peak, but it's much better than the last few weeks have been. I have no
idea why the increase occurred, but I hope it keeps up.
Barbara is off to Raleigh with her friend Laura to deliver the
rescue Border Collie, Tess, to her new owner. They'll stop on the way back
to do a home visit or two, which they insist on doing before allowing
anyone to adopt a BC.
I'm struggling right now with re-write for the processors
chapter. Let me tell you, now is not a good time to be writing about
processors and trying to make recommendations. The problem, of course, is
Intel versus AMD. Six months or so back, I said that I thought AMD might
be forced out of the processor business, and that the Athlon was too
little, too late. And they might have been, too, except for the incredible
series of fumbles that Intel made. Those errors may result in Intel
snatching defeat from the jaws of victory.
Six months ago, AMD was moribund. Nobody other than enthusiasts and
hobbyists had much interest in the Athlon. All that Intel had to do was
execute. They fumbled the ball at every opportunity. First, they ramped
down 440BX production in the expectation that everyone would want Camino
(the i820 chipset). Well, Camino was late, and it turned out that everyone
wanted 440BX chipsets that Intel couldn't supply.
Then, when Camino finally did arrive, it was a fiasco. Literally two
days before Camino-based boards were to ship, it was discovered that there
was a fatal flaw in its support for 3 RIMM implementations, making all
those 3-RIMM motherboards scrap. Then, of course, there was the slight
problem that Camino was designed to accept only RDRAM, which cost
literally five to ten times what SDRAM cost. Even if you were willing to
pay the price, RDRAM availability was very tight.
Then Intel felt compelled to introduce the new Coppermine-core Pentium
IIIs before their manufacturing process had ramped up sufficiently to
enable them to deliver them in quantity. Intel's inability to deliver
Coppermines in quantity resulted in nine of the top ten system makers (all
but Dell) introducing AMD-based systems, which was an incredible loss of
prestige for Intel.
And even if you could get your hands on a Coppermine, there was the
problem of what motherboard to use it in. The 100 MHz FSB Coppermines
weren't a problem, but the only chipsets that supported the 133 MHz
Coppermines were the 810e, which is perceived by most as a low-end
solution, and the i820, which has obvious problems of its own.
Intel had originally planned to release the Coppermine in gradually
faster versions, topping out at 1 GHz or so in late 2000. AMD just kept
plunking along, introducing ever-faster Athlons in 50 MHz increments from
500 MHz to, earlier this month, their 1 GHz monster. That forced Intel to
ship a 1 GHz Coppermine in Q1, rather than in Q3 or Q4 as they'd planned.
The problem, of course, is that 1 GHz is about the limit for the
Coppermine core, and Intel is having problems delivering slower
Coppermines in quantity, let alone the bleeding edge 1 GHz version.
That means that Intel is stuck with the Coppermine for the rest of
2000, and they're not likely to be shipping anything much faster than 1
GHz for all that time. Obviously, as their production ramps up, they'll be
delivering faster Coppermines in larger numbers, and cutting the prices of
them, but the fact remains that Intel is pushed to match AMD speeds right
now, while AMD seems ready to continue rolling out faster Athlons
throughout 2000.
What may result is an incredible swap in positions, with AMD owning the
high-end, and Intel struggling to make profits in the lower margin low-end
and mid-range. And, just as in the past Intel used the Celeron as a club
to drive down margins in the low-end for competing AMD chips, AMD now
seems prepared to do the same to Intel. It's interesting to see that AMD
just chopped prices dramatically on the 700 MHz Athlon, which, probably
not coincidentally, is about the fastest Coppermine that Intel can deliver
in any quantity.
So right now, Intel is under the gun, and the remainder of 2000 is
beginning to look grim for them. Their new Coppermine-based Celeron/566
and /600, due to ship next week, will help, but it's not the answer.
Neither is Tinma, basically a System on A Chip (SOC) processor due to ship
later this year. The real answer is the processor code-named Willamette,
which will probably be called the Pentium IV. The problem is, that
processor won't ship until Q4, and probably won't ship in volume until at
least Q1/01. That's a long time for Intel to wait for something that can
counter Athlon.
In the mean time, AMD is getting ready to ship new generation Athlons
based on the 0.18µ K75 Thunderbird core. First out the door will be the
Spitfire, otherwise known as the Athlon Select, which will provide even
stronger competition to the Coppermine than the existing Athlon does in
the mainstream segment. Following that is Thunderbird, otherwise known as
Athlon Professional, which is a serious threat to Intel at the high end.
Beyond that is Mustang, otherwise known as the Athlon Ultra, which uses
copper interconnects. Then, of course, lurking on the horizon is the K8
Sledgehammer. All in all, Intel has a pretty frightening array of AMD
processors to compete with over the next year or two. And not much to
compete with.
The unfortunately-named 64-bit Itanium (formerly known as Merced) isn't
the answer. Many have already begun calling the Itanium the Itanic, with
the expectation that it's likely to sink from sight about as fast as its
namesake. The Itanium will initially ship in at best an 800 MHz version,
and it's likely that the first processors available will be even slower
than that. As the first IA-64 processor, Itanium can run current software
only by emulation, and it's nearly certain that Itanium performance with
existing software will be quite slow. So Itanium isn't the answer.
I have no doubt that Intel is trying desperately to get the Willamette
out the door sooner than Q4, but from what I know right now, it doesn't
seem likely that they'll be able to do so, at least in any quantity. And
even the Willamette/Pentium IV isn't a guaranteed solution. Although it
may ship with initial clock speeds in the 1.4 to 1.5 GHz range,
indications are that it will be less efficient than the Athlon and the
Pentium III. That means that a Pentium IV may have to run at a
considerably higher clock speed just to match Athlon and Pentium III
Coppermine performance.
So it appears that AMD has won, right? Not really, and that's the
problem. All of this stuff is speculative. Right now, today, the best
solution is the Celeron and the Pentium III Coppermine. Until very
recently, Athlon has been hampered by poor chipsets and motherboards. Via
has begun delivering their Athlon chipset in quantity only in the last
month, and decent Athlon motherboards are only now starting to arrive. AMD
itself is striving to replace their first-generation AMD750 chipset with
the forthcoming AMD760 chipset, which should be a barn burner.
But none of this has actually happened yet. Unlike Intel, whose series
of missteps came as a surprise to everyone, AMD is known for poor
execution. AMD has done a very good job of exploiting Intel's errors over
the last six months, but that is no guarantee that they will not revert to
their old-style inability to deliver. AMD has bitten off a lot for a
relatively small company--the Thunderbird core, copper interconnects,
developing the AMD760 chipset, and so on. I'm surprised they aren't
manufacturing motherboards. AMD has an incredible opportunity during 2000
to beat Intel at its own game. Whether they can succeed in doing so is the
question.
So what's a poor author to do? Whatever I write today will be read two
or three months from now (not to mention six months or a year from now) by
readers who unconsciously assume that I wrote the stuff about 30 seconds
before they read it. Ridiculous, but true. I can't recommend Athlon at
this moment based on what I know right now, but the situation may be very
different in three months. So I guess I'll put some form of this essay in
the book and point out that "as this is written" it is unclear
as to how the war will turn out.
I guess I better talk to AMD about getting some eval processors. The
last time I tried that, I couldn't find anyone at AMD who'd even return my
calls, but I understand they've gotten a lot better in the last six months
at taking care of the press. We'll see.
I think I'll work on the motherboards chapter instead for a while.
* * * * *
-----Original Message-----
From: J.H. Ricketson [mailto:culam@micron.net]
Sent: Friday, March 24, 2000 11:36 AM
To: thompson@ttgnet.com
Subject: Goodbye Microsoft
Dear Bob,
On Wednesday you wrote: "But my goal is
eventually to be able to do my work without depending on any Microsoft
applications."
I totally agree with your sentiments re
forced registration and applaud them. It serves no legitimate purpose
for me. It has long been my policy not to register anything. I get too
much spam as it is. It certainly can't provide "support" that
is nonexistent anyway. I will not under any circumstances move beyond
Office 97 and NT4 until Microsoft no longer demands access to my
computer to do Gods only know what damage.
Take heart: there is a very enthusiastic
review of Corel's WordPerfect Office Suite for Linux [here]:
Much as I abhor Word(im)Perfect, and the
other apps in the suite, having been forced to use them in years past, I
was intrigued. It appears to be a very viable alternative - at least for
der Linuxen.
One thing that has aroused my curiosity:
does your resolve include Internet Exploder, or will they "have to
pry that from your cold, dead hands?" <BG>
Regards,
JHR
--
[J.H. Ricketson in San Pablo]
culam@micron.net
Actually, I always much preferred WordPerfect to Word. I used WP
from the DOS 4.0 release through the DOS 5.1 release and also the first
couple of Windows versions. I think it was when I started writing books
that I had to shift to Word because all the publishers' templates were
based on it. I don't know about IE. I keep hoping that MS will see the
light and release a Linux version. I should perhaps have made it more
clear that I was talking about migrating away from MS applications in the
(very) long term, not the short term.
* * * * *
-----Original Message-----
From: Alan Donders [alan_donders
at hotmail period com]
Sent: Friday, March 24, 2000 7:10 PM
To: webmaster@ttgnet.com
Subject: O2K / FP2K
Bob, Just a note to let you know that I
tried the Microsoft site to order the O2K SR-1 CD and the process
completed without error. I got a page that says:
Thank you for your order!
Your order number is xxxxxxxxxxxx.
Please allow 4-6 weeks for delivery of your
order.
BTW, while you were trying to order the CD,
did you mind that your phone number is a required field? I don't mind
giving my e-mail address as I have one that I use just for these types
of requests but I can't imangine the need for a phine number for a
fulfillment area to mail out a CD. Actually, I think there are some
sites that offer a phone number connected to a free voice-mail service
that sends you the messages as e-mail attachments. I think I'll hunt one
down and then have a "throw-away" phone number to use too.
Also, is Jerry aware of the FP2K bottom
shared margin bug? He seems to sometimes get action when he brings these
types of issues to Microsoft's attention.
Keep up the good work on your site. Very
enjoyable and informative.
Alan Donders
alan_donders at hotmail period com
P.S. I've heard that programs that scan web
pages to harvest e-mail addresses now also look for the strings
"at" for '@' and "dot" for '.' Have you heard
anything along those lines?
Microsoft's web site blowing up on a script error is nothing new
to me. I'm glad you succeeded in getting a confirmation, but I'm not going
to worry about it. I downloaded the 52 MB SR-1 file yesterday. As far as
phone number, I always just give them my fax number or the number of the
line I use for dialing the Internet, which is nearly always off-hook and
doesn't ring anywhere anyway. Jerry is aware of the FP2K shared margin
bug, although he hadn't notice that it had deleted his copyright notice
until I pointed it out to him. I'm currently in communication with a guy
at Microsoft whose title is Product Lead for Support for Microsoft
FrontPage, so I have some hopes of getting the problem resolved or,
failing that, at least getting it recognized as a real bug that needs
fixed. As far as address grabbers, I speculated a month or so ago that
they could easily parse for the pattern "??? at ??? dot ???" and
convert it to a properly-formatted address. I have no knowledge that any
address grabber actually does this, but I'd do it that way if I were one
of the sleazy, scum-sucking, bottom-feeder programmers who write such
software.
* * * * *
-----Original Message-----
From: chuck goldie [mailto:agoldie@cin.net]
Sent: Sunday, March 26, 2000 5:39 AM
To: thompson@ttgnet.com
Subject: network help
Would you by chance know of a basic network
how-to book? I am looking to install a network system for my wife in her
real estate office. The system will only be including three PC's. This
is a very low budget operation with the main objective in the network
installation being internet service on more than one computer with only
a single ISP account.
Any help at all would be great!
Chuck Goldie
agoldie@cin.net
I don't know of any good title off the top of my head, but I
suspect one of the Dummies/Morons/Idiots/Cretins books might be just the
thing. Perhaps one of my readers will email you directly with other
suggestions.
There's not really much involved to bringing up a small network
Buy a 10/100 Ethernet card for each machine, a hub (I'd suggest an
eight-port version), and some pre-made Category 5 cables in the correct
lengths. I'd stick with Intel or another name brand for the cards and hub,
which should run you $30 to $40 per card and perhaps $125 for the hub. The
cables should be $3 to $15 each, depending on length (although very long
ones will be more). You can find pre-made Cat 5 cables in anything up to
100 foot lengths. You might also consider buying one of the small
networking kits offered by various vendors. If you have Windows 98 SE on
your "server", it already has Internet Connection Sharing
built-in, although I've never used it.
|
wpoison
Search [tips]
TTG Home
Robert Home
Daynotes Home
Links
Special
Reports
Current Topics
|
Sunday,
26 March 2000
[Last
Week] [Monday] [Tuesday]
[Wednesday] [Thursday] [Friday]
[Saturday] [Sunday] [Next
Week]
My friend David Silvis sent me this URL, outraged about yet another
invasion of privacy. This time, drivers license photos on-line. Click here
to try it for yourself.
I'd completed benchmark testing on the CA810EAL motherboard, so
I tore down that testbed yesterday and rebuilt it around an Intel CC820 Cape
Cod motherboard. That's the version that uses SDRAM via the Memory
Translator Hub (MTH). I also have a VC820 motherboard, which is identical
except that it uses Rambus RDRAM. I'm using configurations as similar as
possible to get a comparison between the i810e, i820 with SDRAM, i820 with
RDRAM, and the venerable 440BX. The 810e, of course, has embedded i752
video. For the other boards, I'm using an ATI All-In-Wonder 128 AGP card.
I have to say that my first impression of the CC820 was neutral.
Performance seems to be on a par with the 810e. For example, here are some
figures under Windows 98, the average of five runs each, with the CA810EAL
using a Coppermine/600 and the CC820 using a Coppermine/600EB. Although
the latter runs a 133 MHz FSB, remember that the 820 MTH runs memory at
100 MHz, so the figures are comparable:
ZD
WinBench 99 1.1 |
810e |
820 |
CPUmark 99 |
52.76 |
52.26 |
FPU WinMark |
3208 |
3208 |
SiSoft
Sandra 2000.3.6.4 |
810e |
820 |
Dhrystone |
1616 |
1617 |
Whetstone |
802.8 |
803 |
CPU Memory |
211.4 |
203.6 |
FPU Memory |
223.6 |
234.8 |
I'd call that pretty much a draw. The 820 is about 0.9% slower on
CPUmark 99, but there's enough variance between runs that the results
could easily have been reversed. The 820 is about 3.7% slower than the
810e on the CPU Memory test, and about 5% faster on the FPU Memory test.
Those tests show only minor variations between runs, so there may be a
real difference there. How meaningful it is is another question.
* * * * *
-----Original Message-----
From: Jeff Orvin [mailto:JeffOrvin@fni.com]
Sent: Saturday, March 25, 2000 9:18 PM
To: alan_donders@hotmail.com
Cc: Bob Thompson
Subject: Harvesting e-mail addresses
Haven't used it yet, but I ran across this
page a while back:
In part, it says, "The HTML Encoder is
a service of SiteUp Networks. The Encoder will format your E-mail
address in a way that IS NOT readable by any E-mail extractor or search
engine."
Thanks.
* * * * *
-----Original Message-----
From: J.H. Ricketson [mailto:culam@micron.net]
Sent: Saturday, March 25, 2000 11:15 PM
To: thompson@ttgnet.com
Subject: Your Sat. 3.25 Analysis of Intel vs. AMD
Dear Bob,
A very insightful and fair-minded analysis
of the situation, IMO. Thanks. Would you care to go into a similar
analysis of the SMP situation, Intel vs. AMD, and their respective
prospects? As I understand it, Intel has a lock on multiprocessor
capability. IMO, this is at least as significant as bumping heads on CPU
speed. End users are going to want SMP, and the Pros are going to need
it. I think that may be Intel's big ace in the hole, if exploited.
Regards,
JHR
--
[J.H. Ricketson in San Pablo]
culam@micron.net
Thanks. The Athlon processor itself has SMP support built-in, but
what it doesn't have for the moment is either an SMP-capable chipset or
SMP support in any operating system. The latter is the real problem,
because even if AMD released an SMP-capable chipset today, Windows
NT/2000, Linux, etc. have support for SMP on Intel only using MPS, which
is proprietary Intel technology. AMD would need to depend on Microsoft,
Torvalds, et al to build SMP support for AMD into their kernels.
* * * * *
-----Original Message-----
From: maceda@pobox.com [mailto:maceda@pobox.com]
Sent: Sunday, March 26, 2000 12:33 AM
To: webmaster@ttgnet.com
Subject: Intel's fumbles
"but the only chipsets that supported
the 133 MHz Coppermines were the 810e, which is perceived by most as a
low-end solution, and the i820, which has obvious problems of its
own."
As a matter of fact the only chipsets able
to support the 133 MHz Coppermines at the time of their launch were the
810 and VIA's chipset since the i820 was withheld by Intel. You should
also consider the efforts taken by Intel and the subsequent lawsuit to
keep VIA's chipset off the market. VIA did to Intel in the chipset
sector what AMD did in the processor front. And what about the 840?
Intel has made a lot of mistakes at a very critical point and almost all
this mistakes can be attributed to their arrogance:
1.- i820: a chipset designed from the ground
up NOT to support SDRAM. RAMBUS' choice was the choice of trying to
collect revenue of competitors. They knew that if the i820 supported
SDRAM Rambus would never take off. SDRAM at 100 MHz and 133 MHz is only
slightly slower than Rambus and with DDR SDRAM around the corner that
speed advantage was going to disappear. The price differential alone
would push motherboard makers and system integrators to the SDRAM/DDR
SDRAM camp. The only choice they had to push Rambus was to force it down
everyone’s throat.
2.- 440 BX: in a parallel road was the 440
BX. By eliminating this product they hoped to force the adoption of the
i820, which forced the adoption of Rambus. Technically speaking the i820
is not a superior product to the 440 BX, Anand, Tom and every single
hardware site on the net has proved it. What does the 440 BX lacks to be
the chipset of choice today: AGP 1/2 multiplier, official recognition to
run at 133 MHz (not necessary since every overclocker is doing it) and
AGP 4X (not very useful). Remember the HX/TX controversy?
I think I have taken enough of your time
already and you already know this. Besides I will be labeled as an Intel
basher (which I'm not) and the truth is that if the roles were reversed
AMD would do the same things or even worse. The bottom line here is that
AMD's products are good enough and I have started recommending them to
my clients. They are not going to disappear overnight and competition is
good. We live in interesting times!
Francisco Garcia Maceda
maceda@pobox.com
You're right, of course, about the VIA chipset. What I was
referring to was Intel chipsets, commenting that they'd basically shipped
a processor without any good choice of chipset to support it. As far as
RDRAM, I'm not sure I'm as charitable as you are. I don't think that SDRAM
is "only slightly slower than Rambus" at all. In fact, I suspect
it's probably slightly faster in normal applications. The great weakness
of Rambus is latency. The other great weakness, which no one seems to
mention, is that the more you use the slower it gets. That is, a
configuration with three RIMMs is slower than one with two RIMMs, which is
slower than one with one RIMM. Not good.
[Last
Week] [Monday] [Tuesday]
[Wednesday] [Thursday] [Friday]
[Saturday] [Sunday] [Next
Week] |
|