Email
Robert |
Week of 18 September 2000 Friday, 05 July 2002 08:03 A (mostly) daily journal of the trials, tribulations, and random observations of Robert Bruce Thompson, a writer of computer books. |
|
wpoison
Search [tips]
|
Order PC Hardware in a Nutshell from Fatbrain.com Jump to most
recent update Monday, 18 September 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week]
Our SETI@Home group now has 56 members and has passed 11,300 work units complete. That's 3 new members, and about 1,600 work units in the last week, which means we're adding about 7,000 work units per month. Congratulations to team members JCMorales, who recently passed the 250 work unit milestone, and Edmund Hack, who recently passed the 50 work unit milestone. If you haven't joined our SETI effort, please do. Here's how. Barbara is home, and that means things are back to normal. I'm happy, my mother is happy, and the dogs are really happy. Hurricane Gordon is heading our way, which means we can expect some wind and heavy rains tonight or tomorrow. And I need to get back to work on the HardwareGuys.com web site as well as all the other stuff that needs doing. But Barbara is back. -----Original Message----- An interesting spin on the MS barcode protection issue you mentioned yesterday. ATB. Phil -------------------------------------------- Thanks. -----Original Message----- Never! As I said, I am in the minority that detests ads and consequently do not, as a rule watch commercial TV and very little TV overall. I find hourly or half hourly breaks convenient enough. Apart from when I visited my mother a few months ago, I wouldn't have watched commercial TV for 5 years. And that was to share a favourite show with my son. These days he shares my disinclination to be assaulted by advertisements. I noted that when visiting my mother, who has my older sister and two of my sister's children living with her, that on one occasion they were watching shows that consisted of advertisments for their entertainment value (sexual innuendo). There were two of these running against each other at the same time, so they had taped one in order to see both. I saw my first bit of the Olympics this evening and a commentator said: "Thorpie's swimming prowess leaves me with insufficient words to say anything..." ROFLMAO I prefer a good book. Yes, we have those shows consisting of all commercials here, too. And I will admit that perhaps one in a thousand commercials is actually worth watching. We have one here, for example--I forget what they're advertising and please no one tell me--that has a Border Collie as an actor on a film set. They want the Border Collie to act sad and he's just not doing the job, so his handler goes over and tells the Border Collie to imagine the worst day of his life. The commercial cuts away to the Border Collie remembering that day. He notices an attractive female dog rolling down the street on a fire truck and sets off in pursuit across his yard. When he reaches the edge of his yard, he takes a flying leap to clear the hedge, and the last thing he sees is a panel van parked in his path. Before impact, they cut back to the film set, where the Border Collie is howling in anguish. But I don't watch TV much. With Barbara away for ten days, I had it turned on for less than 10 minutes during that time. I didn't even realize that there was a hurricane headed this way, or that the Olympics were going on again. It seems like they just had those recently. I think they should drop back to having them perhaps every ten or twenty years at the most. Every hundred years would be better. -----Original Message----- Robert, Just read your page and notice Mike Boyles comment about Fat Brain. I also cannot figure out how to cancel an order online (I wanted to for the same reason) and would find this annoying if I really had a need to cancel an order. I have to admit that although I have stopped using Amazon's because of their policies, my experiences with them have been very positive (over 50 books ordered) and find their site very easy to use. I figure you should get credit for my order (# XXX9999 [removed by RBT]) because I would have never used them except for the recommendation from you web site. I enjoyed your Windows NT TCP/IP Network Administration book and look forward to the PC Hardware in a Nutshell Book. Thanks for the kind words, and for ordering the book. I've sent your order number along to the folks at Fatbrain, and they'll credit my account. I also liked Amazon.com, and ordered probably 100 books from them over the years. But no more. As far as I'm concerned, they were operating under false pretenses the entire time. They took my personal information with the implicit understanding (explicit, even, if you consider the privacy terms they had in effect at the time of the orders) and then turn around and announce that they own that information and have the right to do whatever they want with that information. And for that, I hope all their customers abandon them. As far as I'm concerned, their business practices are completely unacceptable. With Fatbrain being bought out by Barnes & Noble, I hope that they don't change, but I'm afraid they will. Everyone's out to make a buck any way they can nowadays, it seems. But I'll continue giving Fatbrain the benefit of the doubt until and unless they start doing things to show that they don't deserve it. -----Original Message----- Robert, I didn't explain myself very well. It's the FROM field that I think should be used to knock out spammers. At least the ones who spoof addresses. The spam I get says FROM: Goofball871@whocares.ru in my email inbox list. In the body of the letter, Eudora figures out that the letter is actually FROM: lazylucy28@mail.yahoo.com. It's that difference that seems to me to be something to work with. The problem with that is that there are many valid reasons why the From: address may not match the other header information. For example, I'm set up with several accounts, and may send mail From: any of those. All of that is valid mail, and using a comparison of the From: field with the routing information in the header as you suggest would cause most of it to be filtered as spam. The X-Envelope-To: field is a much better way to filter for spam. -----Original Message----- Dear All, I notice that none of the major antivirus companies mention HOW the QAZ worm gets into a system. On Thursday I definitely did not have it. Today (Sunday) while browsing news.bbc.co.uk McAffee suddenly popped up a warning that notepad.exe was infected with the QAZ worm. I had not started notepad, in fact currently .txt files are associated with pfe32 on my pc. I moved it to another folder and had a look around. Sure enough I now had a c:\windows\note.com file. However the registry was not modified. There was no wininit.ini file. I rebooted and all was clear. As a working hypothesis I have to assume that this worm was transmitted from a web page I had visited today. McAffee would have scanned this file once it was created and before I could have run it and possibly triggered the payload. However SOMETHING renamed the original notepad.exe to note.com. Could this have been an activeX control? I hope this helps in any ongoing investigations. That is disturbing. Thanks. -----Original Message----- Hi Robert, In regard to the wireless idea the costs are coming down all the time, and the wireless equipment that was mentioned can provide T1 (1.5mbps) speeds or higher (up to 44mbps) over an area of several square miles if I recall correctly. He mentioned the Cisco WT2700 was $30,000 USD Since we want to place the unit centrally to the neighbourhood that is interested we need a way to get internet access to the unit. Cisco (and Lucent) also sell point to point wireless bridge equipment for a much lower price (aprox $4000 USD) that can support 10mbps speeds and up to 25 mile distances. I know for certain (having pursued this in the past for a corporation) that any small to medium size ISP would be happy to sell you internet access across this connection in whatever increment you wish and they would allow you to have the wireless bridge and antenna in their location, they will also provide email and newsgroup access. To give it a price, it would based on local pricing be aprox $700 USD/month for 1.5mbps. So you have a capital cost of $35k and an ongoing monthly cost of $700. You could probably support 150 users without any real problem at T1 speed and under the circumstances you'd have room to scale up fairly cheaply as 3mbps would probably not cost double, I suspect you could get 3mbps for around $1100 and you wouldn't need to add any capital equipment. (Everyone would of course have to buy their own wireless modem at a probable cost of $200-$300) For a neighbourhood that didn't have access to cable-internet or dsl and could generate enough interest, it would probably be worth pursuing in conjunction with a willing ISP. I have used the point to point line of sight wireless lan equipment for about 8 years now and it's most recent incarnations are very reliable even in inclement weather. We have had periods of 1 year at a time with no downtime even across links where we pushed the envelop of the technology. The WT2700 is designed for use where you don't have line of sight so I suspect it would be even more reliable. In fact we have wireless ISP's selling services in my city, competing directly with both cable and dsl. They also sell "cable-tv" to people that can't get cable-tv service or who are tired of the cable company. In regard to routing and RFC 1918, I can reach some 192.168.x.x addresses that are 5 hops away on my Rogers@home connection. And I get replies to pings not "destination unreachable" However remember that I am not leaving @home's networks. So I'm not going out to the big "I" Internet. I'm in Ontario and I get replies from private addresses that appear to be in Detroit and Texas based on the traceroute replies of the router names nearest them. It is also possible that these are not cable subscribers but addresses that the cable companies are using for their own equipment. Your cable modem has a 10/8 address on the "cable side" of it. To take a look at the cable network from your cable modems perspective get yourself an ethernet card that has the old style 10base2 connection and connect that to your cable. (I used a Fluke Lanmeter 686 cause I had one available however you should be able to reproduce the result with any 10base2 ethernet card) Then give your pc a 10/8 address and/or just load a packet sniffer program and you'll probably see far more than your cable company intended you to see. This worked about a year ago anyway, I suspect it still does. Well, as I've said, running an ISP is not as simple as most people seem to think, but I'd love to hear from anyone who actually tries it. As far as the private IP addresses, as you say there's no problem with those being visible within the ISPs network. Actually, there's no problem with them being visible per se outside the ISPs network, so long as the remote site isn't advertising a route for the private IP addresses. For example, if you tracert this site, you'll see at the end that you arrive at pair networks at about the third-to-the-last hop, which is the pair networks gateway. The next-to-the-last hop will be a private address, something like 192.168.1.6. That's ugly-looking, but occurs simply because they're using address translation internally, and does no harm so long as pair is not advertising that route. -----Original Message----- Hello Robert, Don't worry about your upbringing, I think mine was somekind confusing about languages. The fact is that my father is from Galicia and he talked Galego in his youth, but my mother is from León so she has always talked Spanish (well we call it Castellano, not Spanish). They met in Barcelona (Catalonia) and married here so I born in Catalonia and raised in a house where Spanish is talked. First confusing fact, almost nobody talked Spanish!! It was shocking for a six years old boy, but OK I learnt Catalonian....And finally some holidays we went to Galicia to see my father's family and of course nobody talked Spanish nor Catalonian they talked Galego....As you may have noticed I had some trouble defining my own identity. Now, I usually speak Catalonian and Spanish as an interlingua with non catalonian speakers, unfortunately I don't speak galego. About the D815EEA, I've been looking for details on this motherboard and is really impressive. I just can see a problem, will it run smoothly on Win95B? The fact is that I know that Intel UltraATA drivers do not allow to be installed on Win95, and it will slow HD performance....Well, It's worth a try, I'll recommend to buy a machine and test it intensively. By the way, do you follow some method to test new PCs or components? Regarding the MAC exchange issue on Win98, I've tested it this weekend and found it's possible and very easy to do it. My NICs are an Intel EtherExpress and a 3Com 905B, both of them with ROM recorded MACs. (Ooops, I'm sorry I've forgotten my notes on the exact registry key!) You just have to look for a key group called DhcpInfoXX (where XX may range from 00 to 99) and there you'll find the keys and values. Wow. Things must have been a bit confusing. As far as the D815EEA, I don't know if it runs Win95B although I suspect it will. I no longer run Win95 around here (something has to give, and I'm running so many OS versions already that when I add a new one like Windows Me something has to drop out the back door). As far as testing, I don't use a rigidly defined procedure. Generally, when I build a system around a new motherboard, I first burn in the system and then begin running various benchmarks, including SiSoft Sandra and the PC Magazine suite. Recently, I've added the SETI@home as an informal benchmark. Once I finish with the benchmarking (which I do more a a stress test than because I place much weight on the results), I typically begin using the system for everyday things like writing, creating web pages, browsing the web, burning CDs, and so on. After I'm satisfied that the system is stable, which frequently means running it for weeks, I generally start swapping hardware in and out of it and repeat the whole process. All of the systems get a good workout, and end up having more stuff swapped in and out than a regular user is likely to do over the life of a machine. I'll check out the MAC address issue you mention if I get a spare moment. -----Original Message----- I find it a little disturbing that Barnes & Noble bought out Fat Brain because I have had mixed luck with books I have ordered from Barnes and Noble (wrong books shipped/extended delays) and had basically stopped using them. I have also used VarsityBooks.com (5-10 books) and have been very happy with them although I have yet to have had a problem so I can not comment on their customer service. I'm not crazy about the idea, either, but like many people I hope that B&N will just leave Fatbrain alone. And given the nature of Fatbrain's customer base, I suspect there's a good chance of that.
|
|
wpoison
Search [tips]
|
Tuesday, 19 September 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Our SETI@Home group now has 57 members and has passed 11,500 work units complete. Congratulations to team member Robin Whitson, who recently passed the 100 work unit milestone, and team members Mike and Steven, both of whom recently passed the 50 work unit milestone. If you haven't joined our SETI effort, please do. Here's how. The AMD Duron/750 system is supposed to show up today, so I'll
be spending some quality time with it over the next couple of weeks. AMD
is also sending an Athlon processor, which will run in the same system, so
I'll have lots of things to Barbara went to the gym and the grocery store yesterday (we were out of everything, but all I'd remembered to write down on the grocery list was mayonnaise and laundry bleach. Bad dog! Bad, bad dog!). When Barbara returned, she went into Tasmanian Devil cleaning mode, at which pursuit she spent most of the rest of the day, muttering constantly about a filthy house. And here I thought I'd kept the place up pretty well. Women have a thing about filth and squalor. They notice it instantly in places that men can't see it even when staring directly at it. In a strong light. With a magnifying glass. At any rate, before she commenced cleaning, we had a conversation that illustrates the differences between men and women:
Apparently, our water (which we drink straight from the tap) is not clean. No, it contains microscopic sediments, which settle out on the toilet bowl if it is not cleaned at least weekly regardless of use. So, from a woman's point of view, water that we drink is not clean enough to leave sitting in the toilet. Hoping for a little moral support, I sent a pre-pub copy of the above to Jerry Pournelle, who responded: -----Original Message----- If you think I am going to write any kind of reply that might get back to either Barbara or Roberta you have lost your mind. The only secret of a successful marriage is that the man learns to grovel. -----Original Message----- Robert "I didn't explain myself very well. It's the FROM field that I think should be used to knock out spammers. At least the ones who spoof addresses. The spam I get says FROM: Goofball871@whocares.ru in my email inbox list. In the body of the letter, Eudora figures out that the letter is actually FROM: lazylucy28@mail.yahoo.com. It's that difference that seems to me to be something to work with." I have two internet accounts. The one I use at home (a cable modem) and the one I use at work (a dialup with a local ISP). I would use the cable account for all my sent mail if I could but the cable ISP will not send mail when it is received from another ISP. That's why the address below my signature is diffrent from the mailto address in the header. I can pop the mail from both with the local IPS or the cable ISP. Mike Boyle Yes, I understand. But the point is that the From: address is not a good discriminator for spam. There are many reasons why legitimate mail may have a mismatched From: address. Conversely, most spam comes from a From: address which is not easily determined to be false, short of validating the existence of that address for each received message, which is an impossible additional workload. X-Envelope-To: is the field to use, because although not all messages that include an X-Envelope-To: address are spam, (nearly) all spam messages use an X-Envelope-To: address. -----Original Message----- This is the registry key group where you can find HardwareAddress: HKEY_LOCAL_MACHINE\System\CurrentControlSet\Services\VxD\DHCP\DhcpInfo00\ Thanks. I'll check it out if I get a moment. -----Original Message----- Musicians are not the only ones with rights woes. If you haven't checked out the reference on Dr. K's page about Linda Weltner's experience with The Boston Globe (now part of The New York Times Company), it's worth the read. In fact, I'll bet that all of your credit card agreements now deny you the right to a trial, in the event of trouble, forcing an arbitration settlement, instead. Now, I might voluntarily choose arbitration if some issue arose, but DENYING me the right to trial in a criminal matter (and since we're talking money, it's most likely the underlying issue would be theft--a criminal matter) is not consistent with my reading of the Constitution--or many state constitutions, for that matter. And if you don't like that, your choice is not to have a credit (or debit) card. It's baffling to me that people hurry through their busy lives, either unconcerned or completely ignoring a creeping loss of significant rights--issues that once occupied a great deal of think time in the early history of this country. Well, I wouldn't call it unconscionable. Nearly all publishers are making electronic rights grabs in their latest contracts, and as far as I can see The Boston Globe is fully entitled to demand what they're demanding. I don't much like it, but, as they say, if I don't like it I don't have to deal with them. If the writers don't like the deal, they shouldn't sign the contract. It's as simple as that. -----Original Message----- Robert - I gave up on Amazon and Fat Brain after I found their prices undercut by BookPool (www.bookpool.com). Additionally, I needed to cancel a recent BookPool order and found they had a cancellation link for that purpose. But I really didn't think this unusual at the time - this should be normal. Your discussions may reveal BookPool's cancellation tool to be unique! Additionally, after I cancelled the order, I received an e-mail from them confirming the order was shipped. I responded that I tried to cancel the order and the kind lady responded back, instructing me to return it unopened. I did and they credited my card the full amount. And that's another reason I have a BookPool link on my page. Al Hedstrom I don't think I've ever bought a book from Bookpool. They stock very few titles (none of mine were in stock) and I find their web site difficult to use. I've bought bunches of books from Fatbrain with never a problem, so I'll stick with them unless and until they change for the worse. -----Original Message----- hello sir, I'm new in this computer field i need ur help if u please let me know how to delete ur bogus windows and how to install new one with desire partition. thanks I assume that you're talking about Windows 98. Windows NT 4 and Windows 2000 allow you to boot from startup floppies and create/delete whatever partitions you wish during setup. Before you repartition your drive, make sure you've backed up any data you wish to keep. With Windows 98, create an emergency boot floppy (Control Panel -- Add New Programs -- Startup Disk) and start the system with that floppy. When the DOS prompt is displayed, run fdisk and use it to delete all existing volumes and partitions. Then restart the system and create whatever partitions and volumes you need. Then re-install Windows 98. Alternatively, if you can get your hands on a copy of PartitionMagic, you can use it to modify existing partitions and volumes on-the-fly, without having to wipe everything out and re-install.
|
|
wpoison
Search [tips]
|
Wednesday, 20 September 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Our SETI@Home group continues to crank out work units. Congratulations to team member Peter Bruno, who recently passed the 250 work unit milestone. If you haven't joined our SETI effort, please do. Here's how. Just as I clicked "Publish" yesterday morning, FedEx showed up with the new AMD Duron/750 system, direct from AMD. I had the new computer unboxed, up and running, connected to the network, and running the SETI@home client in literally less than 10 minutes. This system "feels" very fast. Much faster than I expected, because AMD configured this system as a reference "value" system. That is to say, it uses the Duron rather than the Athlon processor, and the other components, while of superior quality, are entry-level components--5,400 RPM Seagate hard drive, PC100 rather than PC133 SDRAM, an inexpensive TNT2 Vanta video card, and so on. I'm now running SETI on eight systems, having added the new Duron/750 to my stable of SETI systems. That machine cranks out a SETI work unit in just under 11 hours, comparable to the Katmai-core Pentium III/550. The puzzle remains Barbara's Pentium III/450 system, which for some reason seems to be completing work units much faster than it should based on its processor speed. Thinking that perhaps the last time I checked it had gotten an "easy" block or something, I checked again yesterday. It was near the end of a work unit, and was on schedule to complete it in 8:45, even faster than the 9:00 that raised my suspicions in the first place. Granted, that machine has 128 MB of RAM versus 64 MB for many of the other systems, but my dual Pentium III/550 has 256 MB, and it pretty consistently turns in times of about 10:45. So why should Barbara's system be running so much faster, given that it's a slower processor, with the same amount of L2 cache, and running on the same chipset? There are some other interesting differences, summarized in the table.
With all the anomalies in this table, I decided to normalize the data and chart it. I divided the units/day figure by the MHz rating of the processor and multiplied the result by 1,000 to provide a unitized performance rating value. From left to right, the first four systems all use the Deschutes/Katmai core, which has 512 KB of L2 cache running at half processor speed. This makes it clear that the question is not why the Pentium III/450 is running so quickly, but why the Pentium III/550 is running so slowly. The fifth system, a Celeron/333, uses the Mendocino (modified Deschutes) core and has only 128 KB of L2 cache, although that cache runs at full processor speed. The sixth and seventh systems are both Coppermine-core Pentium IIIs, which have 256 KB of full-speed L2 cache. They provide nearly identical performance clock-for-clock, as they should, and in comparison to the Katmai/Deschutes, they are less efficient at running SETI, presumably because of their smaller (although faster) L2 cache. But that raises the question of why the Celeron, which has half the L2 cache of the Coppermines, is apparently more efficient. The Duron, with a smaller L2 cache than the Pentium III Coppermines, and a much smaller L2 cache than the Deschutes/Katmai Pentium II/III systems, comes off looking terrible.
So, does this mean the Pentium III/Coppermines and the Duron are slow and inefficient processors? Nope. What it means is that the SETI client is not a very good benchmark (not that anyone ever represented it to be). Unless, that is, you're buying a system solely to run the SETI client. It's pretty obvious what's going on here. The SETI data block must be small enough to fit entirely into the 512 KB L2 cache on the Deschutes/Katmai core processors, but too large for the 256 KB L2 cache of the Coppermine processors or the 192 KB L1/L2 cache of the Duron. In case you were wondering about the cache totals, Intel processors duplicate the contents of L1 in L2 cache (called an "inclusive" cache) whereas AMD processors do not duplicate the contents of L1 cache in L2 cache (called an "exclusive" cache). The only remaining questions are why the Pentium III/550 is doing so poorly and why the Celeron/333 is doing so well. Running the SiSoft Sandra benchmarks tells a very different story. Although they are synthetic rather than application-based benchmarks, the Sandra benchmarks are in our experience pretty representative of real-world performance. Once again, I normalized the data by dividing the raw Sandra benchmark numbers by the CPU speed. On the Dhrystone test, which measures integer performance, all of the Intel processors cluster remarkably closely, at about 2.70, while the AMD Duron comes in at 3.15, about 17% higher. Using this Sandra benchmark as a guide, that means that when running typical applications, a Duron will "feel" faster than its clock speed relative to Intel processors. A Duron/750, for example, might feel like an Intel Celeron processor running at about 875 MHz.
On the SiSoft Sandra Whetstone test, which measures floating-point performance, the Intel processors again cluster very tightly at 1.34/1.35. The AMD Duron, at 1.39, is about 3% faster, which may or may not be statistically significant. But it's obvious that the days when AMD was at a disadvantage in floating-point performance are over. At this point, it seems fair to say that AMD floating-point performance at least matches that of Intel processors, and may be slightly better.
I also ran the PC Magazine benchmarks, which show the Duron as having a bit less of an advantage. Under WinBench 99 1.1, the Duron and Pentium III perform at close to the same levels clock-for-clock, and the Duron outperforms the Celeron noticeably. In general, the Duron is a bit slower than the Pentium III clock-for-clock on integer calculations (where the Duron/750 is a little faster than a Pentium III/700) and a bit faster on floating-point calculations (where the Duron/750 would about match a Pentium III/766, if there were such a thing). Again, WinBench is a synthetic benchmark, but one that corresponds reasonably closely to observed real-world performance. I didn't have time to run the WinStone (application-based) benchmarks, but I'm fairly certain those would show the Duron as intermediate between the Celeron and the Pentium III clock-for-clock. Overall, after spending a day with it, my initial impressions of the AMD Duron are extremely favorable. It's fast, stable, and cheap. I can't think of anything more anyone could want from a processor. I'll be wringing the Duron system out over the next few months, so there'll be more to come. I also have an Athlon/1000 processor, which I'm looking forward to building a system around. That'll be our first 1,000 MHz system. -----Original Message----- Mr. Thompson: The September 19, 2000 issue of the Wall Street Journal contains a front page article on the topic of mandatory artbitration for consumer contracts, and the court battles that are resulting from them. I found it spooky to get my RBT daynotes fix, and then read about the very same thing in the WSJ. Not to disagree with Jerry Pournelle, I have a different method for resolving disputes in my marriage. I simply puff out my chest, look her right in the eye, and remind her, "honey, I'm the man of the house! I'm sorry, and I'll try to make sure it won't happen again." Thank you for your courtesy, Richard Micko Yeah, I suspect the Wall Street Journal keeps a close eye on my page as a source for story ideas. As far as your dispute resolution process, that sounds like it'd work. -----Original Message----- "Barbara, incidentally, has the same attitude to new computers that Jerry and I do. " Those that have 'em don't want 'em. Those that don't do. Ain't life grand :) "Apparently, our water (which we drink straight from the tap) is not clean. " Yes but do you clean your teath/drink/bathe in the water from the toliet bowl (after it's been left to fester of course)? No, of course not... so who cares if it's left for a week longer? And I'm not married so will not feel the wrath of any nearby females! ATB. Phil ____________________________________________________________________________ Oh, I knew you weren't married before you mentioned it. Husbands learn quickly that they are required to apologize (a) when they have no idea what they've done wrong, and (b) when they *know* they haven't done anything wrong. That's just the way the world works. It's been that way at least since the times of Republican Rome. I know because I've read essays about it that were written by men 2,000 years dead. If you ever plan to get married, master the following phrases, "I'm sorry, honey.", "I have no idea what I did wrong, but I promise never to do it again.", and "You're right. How could I have been so foolish?" Repeat them over and over again until you have them down pat. If you get married, you're going to need them. -----Original Message----- I've just noticed that choosing Favorites>Links from the standard drop-down menu yields results identical with the minimalist Links toolbar button, displaying sub-folders horizontally and opening any choice in the current IE window. Yep. I've used that one for a while. I add links to favorites and access them using that method until I'm sure I want to keep them, at which point I relocate the link to my local start page. -----Original Message----- Apparently, our water (which we drink straight from the tap) is not clean. No, it contains microscopic sediments, which settle out on the toilet bowl if it is not cleaned at least weekly regardless of use. So, from a woman's point of view, water that we drink is not clean enough to leave sitting in the toilet. OK, you HAVE water. Ours contains dirt. I clean dirt. I can tell YOU how to clean dirt. It's easy. <G> Keri M. Beland Eh?
|
|
wpoison
Search [tips]
|
Thursday, 21 September 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Our SETI@Home group now stands at 58 members and 12,000 work units complete. Congratulations to team member Al Carnali, who recently passed the 100 work unit milestone. If you haven't joined our SETI effort, please do. Here's how. I worked all day yesterday on the HardwareGuys.com web site, although I haven't been publishing the updates to the web server. I've been taking my time, trying to get things right, on the assumption that the book won't hit the bookstores until October 10th. Then I got email yesterday from the O'Reilly webmistress, telling me that the article we did for the O'Reilly web site will probably start running tomorrow (today). Oops. That article points to the HardwareGuys.com web site, which isn't anywhere near complete. I called her to point that out, and also to ask if it might not be a better idea to wait and run the article once the book is actually available. Not so, as it turns out. They shoot to get an article up a couple of weeks before a new book is actually available. Apparently, that creates pent-up demand for the title. It's not my place to question O'Reilly's expertise in marketing books--they do a pretty good job of it--so I told her I'd do my best to get the web site in as good a shape as possible before the article is posted. I may actually have a bit of breathing space. She said their goal was to get part 1 of the article up today, with part 2 posted next Monday, and part 3 on Wednesday. But they might not be able to make that schedule. If not, they won't post part 1 tomorrow because their traffic, like everyone's, drops precipitously on Fridays. In that case, they'd post part 1 on Monday. So I'll be working my butt off to get as much as possible posted and current on the web site between now and whenever part 1 is posted. And actually, I'll be working my butt off beyond that point, to get the site in the best possible shape even after the article posts. And then, of course, we need to start getting ready for our trip to New Hampshire and continue getting the house ready to put on the market. So I probably won't be doing much more than quick updates on this site for the next several days. Mail will just have to slide. Sorry. I'll let everyone know here when the article goes up on the O'Reilly site.
-----Original Message----- Robert Have you noticed they will rarely tell you what they want? The husband realizes they want something, and eventually starts offering things. He will undoubtably offer things that the wife didn't know she wanted. The wife takes those things, but still wants the original. Mike Boyle No, I hadn't noticed, but now that you mention it... -----Original Message----- Dear Robert, It's been a couple of months since I last wrote to you, time for an update. This letter was originally going to be a plea for help sorting out intractable network gremlins. I have four machines on my home network, and a few more at the network I administer at work. No, I've no formal experience doing this sort of thing, but I've always been a tinkerer. At work, there was no network at all until I arrived, so the worst that could happen is that the network doesn't work, and they're no worse off than when I started. Anyhow, this network problem manifested itself in the weirdest ways, and the upshot is this: if you're running the ZoneAlarm software (click here to get it and here to read a critique of it) on a network - any network - you MUST configure it properly. This configuration procedure is easy once you know how, but it is not particularly well documented. So don't do what I did and fling things around your office in frustration because your network won't resolve names...just take a look at ZoneAlarm and make sure you have done to it what I'm about to describe. Then you can feel free to slap your forehead with your palm and moan about your own idiocy. As I said, this problem manifests itself as a name resolution issue. I am using a standard Linksys network kit, with some no-name PCMCIA cards for my laptops. The protocol I use is TCP/IP - again, standard. The whole setup is cheap and reliable and relatively quick. The networked computers are all running Win2Kpro, and it is correctly installed and configured. Now comes the hard part. With ZoneAlarm installed, and with the "local" settings configured as "low," you will have intractable name-resolution difficulties. This is surprising, because the "low" setting clearly states: Minimal security setting: - Enforces application privileges and Internet lock settings only. - Internet Lock blocks only application traffic. - Allows local network access to Windows services and file/printer shares. - Leaves your computer and server applications visible to others on your local network. Essentially, that last line is a lie. Your computer MAY be visible to others on the local network, but only with the most minimal amount of consistency. Sure, you'll be able to ping other local computers, and they will usually give a response. But not always. You will usually have to ping using an IP address. You will NOT be able to ping using the computer's actual name. You'd think that disabling ZoneAlarm would be a standard troubleshooting move on my part, and you'd be right. But if you simply disable the software, you won't solve a damn thing. You must uninstall it completely in order to make things right. Thank God the uninstall procedure works well! I note that Zonelabs has a patch on their site for those of us using the Win2k service pack. Could it be that this is an undetected problem? It does not manifest itself in Win98 or NT4. If you want to run ZoneAlarm without having it disrupt your LAN, you can, as long as you configure it correctly. First, install the software. Second, re-boot. Third, move the slider for local security to "low." Finally, you have to click on the "advanced" button. This brings up an additional window that allows you to specifically tag certain items as "okay." In other words, anything selected here will be left alone by ZoneAlarm. You should see your Ethernet adapter listed. ENSURE THAT THERE IS A RED CHECK NEXT TO THIS. That's the problem, you see...you have to specifically include your OWN hardware. Then, click on the "add" button and add any IP addresses or a range of addresses that ZoneAlarm will allow. Now, you're set. This will not work if you let your computers choose their own IP addresses, so be headstrong and choose for them. Now all will be bliss. On another note, I had to write with a mini-review of a piece of hardware I just received. It's a Sony VAIO notebook that I picked up from Ubid for a song. PIII 650 w/speedster, 128 MB of RAM standard, DVD drive, bells, whistles, fog horns, you name it. Nice machine, nice as hell. I head the technical department of a company that refurbishes laptops, and we've never seen anything that comes even remotely close to this machine in terms of speed, features, and build quality. That being said, there's a problem. This VAIO (PCG-XG18) happens to come with Win98 pre-installed. That's fine if you don't care about such niggles as reliability or robustness. For me, though, I decided to go with Win2K in a dual-boot configuration. Sony has a ton of drivers on their site for running Win2k on this machine, so I figured it would be no problem to upgrade. Not only are there lots of drivers, but there are two guides that detail how Sony recommends you set up Win2k on an XG18. This is going to be a piece of cake. Yeah... But here's the difficulty. The guide recommends performing a "clean" install of Win2k in a dual-boot configuration. The machine even come with its 18-gig hard drive already partitioned into two roughly equal partitions. So, following the guide's step-by-step instructions, I installed Win2k to the second partition. This is the "D" drive, for you non-partition-savvy people out there. Sony's drivers all come in executable packages that are actually thinly disguised WinZip archives. When you double-click a driver, it unpacks everything to a temporary directory on the C drive. Then, in an amazingly demonstration of bad software engineering, the executable driver-installation routine starts to install the recently unpacked software...but it looks for this stuff in a directory on the BOOT drive. Well, the boot drive in my Win2k dual-boot systems is always the D drive; it has to be, given the installation procedure I follow (and the one recommended by Sony). I guess that usually a Win2k installation is on the C drive, but not in this case. So the upshot was that the installer potion of the driver package was unable to find the files it needed, and it errored out. The solution? Let the installer error, but don't close the error dialogue box. Instead, use Windows Explorer (or some such) to artificially create on the D drive the same directory structure that the installer created on the C drive. Then copy and paste the driver software from the C drive to the appropriate directory on the D drive. Close the error dialogue (and watch your files on the C drive vanish when the installer fails!). Run the installer again, it unpacks its stuff to the C drive, then looks for the same stuff on the D drive...bingo! It finds what it needs. And believe me, you WILL need these drivers to run on this machine. Yes, the computer itself runs Win2k wonderfully, but the drivers and the install procedure need a complete overhaul. Thankfully, most of the applications that come on CD with the XG18 are easily installable on Win2k, once you have the OS up and running, that is. BTW, you should go to the Dell web site and download their Win2k driver for the Alps touchpad. It works much better than Sony's. Well, hope you got something out of this long-winded letter. It basically describes, in a nutshell, what I've been doing with my time for the past solid week. But now that I'm a happy camper, I just had to share my trials with somebody. Feel free to leave my e-mail address with this letter, in case anybody has questions about a ZoneAlarm install. Cheers, Jeff P.S. Yes, I know the arguments against using ZoneAlarm, but what the heck...Steve Gibson's site can't crack it (yet), and best of all, it's free! Thanks. I downloaded ZoneAlarm but never installed it. I run WinGate as my proxy server, and it does a pretty good job of protecting my network, at least according to Gibson's ShieldsUp. -----Original Message----- Robert: In last week's update, you wrote "As far as your idea of stringing wire across the driveway, it does have the huge advantage of saving you each $15/month. Balanced against that, however, are a few minor disadvantages. First, your cable provider is going to terminate your service for violating the terms of service. Second, you're going to get arrested for theft of services. Third, if you have a close lightning strike, you may very well find that one or both of your homes burns down (literally). And fourth, your fire insurance isn't going to pay for it. But you will be able to save the $15." I understand points 1 and 2 but I do not have much experience with 3. Why would connecting two houses with a chunk of CAT 5 increase lightning risks? How is that different than the cable company running coax to the same two houses from a single pedestal? Thanks for helping me keep in touch with PC hardware! - Rob Campbell The problem is that each home has its own ground loop, which is defined by a grounding stake (typically an 8 foot long piece of heavy copper driven into the ground at the service entrance). If you run a piece of network cable between two homes, you're connecting two separate ground loops. If lightning strikes near one, the differing ground potentials can cause a catastrophic fire. You might think that little 24 gauge wire wouldn't be able to carry enough electricity to cause much damage, but the number of amps a given size wire can carry depends on the voltage. At the extreme voltages induced by a lightning strike (or near strike) that small wire can carry a ton of amps. Enough to literally vaporize components in any connected device and cause a catastrophic fire. NEC requires isolation devices be installed in such situations, and that is what utilities like cable and telephone companies do. -----Original Message----- Mr Thompsom, Can you give us a little more detail on Barbara's system? I'd like to know if the video settings, ram type and drive system are compareable to the other systems. Did she luck out and get SCSI? I have an NT server here with a PIII 450 that seams to hold its own against my newer servers (733s), and if you get a chance to delve into this I'd like to know. I noticed the video was in 256 color (its just a file/print server) and changed it to 65k colors. It was like throwing out an anchor . Just be carefull she doesn't see you messing with her system (she might stop cooking and pot pies get old after a while). Yancey McCalla Um, she did have an 18 GB U2W Seagate Barracuda hard drive running off an Adaptec 2930U host adapter, but I stole that drive one time when she wasn't looking and replaced it with a Maxtor UDMA drive. As far as video, her system has an Intel SR440BX motherboard, which has embedded nVIDIA TNT video, which she runs with the standard Intel drivers at 1024X768X24BPP. Video performance doesn't necessarily drop as one increases color depth. It all depends on the drivers. I've seen video cards that had better performance at 16.7M colors than at 256 colors because the card maker obviously optimized the 16.7M driver and not the 256 driver. Her system has 128 MB of Crucial PC100 SDRAM. -----Original Message----- I built a system for a friend using an asus p5a mother board and it is running windows 98SE. It is about 7 months old and his clock loses 15 minutes every 48 hours or so. I'm thinking it could be a bad cmos battery, however, the power management in the bios should warn of low voltage. Do you have any experience with this trouble on the p5a board? Thanks alot. David I don't have any experience with that exact board, but the problem you describe is not uncommon with many different types of boards. You don't say whether he leaves the system running all the time or turns it off at night. If the latter, the problem could indeed be a defective battery, and replacing it should solve the problem. If the former, the battery has nothing to do with the problem. I've only encountered that on a couple of systems over the years, and the problem was ultimately tracked down to a defective RTC in both cases. -----Original Message----- Hi Robert: First, thank you again for posting my note about my new daily journal on your site, it has helped bring over some curious readers. I've found myself truly enjoying the daily updates and feedback from those so inclined. :-) I have been having a problem with moving Outlook Express mail files (I am using version 4.72) from one PC to another. I just configured a new PC for my wife and can not yet figure out how to get it to recognize the old PC's mail files that have been moved to the new PC. I looked on the web and found various conversion utilities that will convert Outlook Express to another format that can then be imported by the new copy of Outlook Express, but they do not seem to be the optimal solution. Surely there must be a way to get the new install to simply works with all the old mail folders and files. I hope you or one of your readers may have a possible solution. Thank you for reading this and doing the wonderful job with your site. I remain and faithful reader. Sincerely, Bruce W. Edwards You're welcome. Good luck with your site. If you want to draw more traffic, it'd probably be a good idea to put your URL in your mail sig. As far as Outlook Express, I've never so much as run the program, so I have no idea. Perhaps one of my readers will know.
|
|
wpoison
Search [tips]
|
Friday, 22 September 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Our SETI@Home group continues to crank out work units. Congratulations to team members Joe Hartman, who recently passed the 250 work unit milestone, and Rick, who recently passed the 100 work unit milestone. If you haven't joined our SETI effort, please do. Here's how. Barbara wasn't feeling well yesterday, and isn't feeling much better this morning. And I wasn't and amn't feeling well either. I'm encouraging her to take it easy today--she tends to work too hard all the time, including when she's ill--and I may take a little time off as well. I do have the work on the HardwareGuys.com web site to get done, so I'll probably continue with that. The article didn't end up posted on the O'Reilly web site yesterday, so it looks like it'll be Monday for the first part. Just as well. It'll give me a chance to get things in better shape before people start hitting the site. We'll be leaving for New Hampshire a week from today (note to prospective burglars: (a) the dogs will still be here, and (b) my brother is house-sitting, and I'm leaving him with a loaded shotgun). I was considering what kind of computer, if any, to take along. I don't have a laptop, but if I want one that's really no problem. I can call up Dell, Compaq, or whoever and get an eval unit FedEx'd here overnight. (Actually, I should probably do that anyway if I'm going to write a chapter about portable systems...) But Barbara came up with a good suggestion. Since we're driving up in the Trooper, she says, why not just take a real system? We have no lack of space to carry a system unit, monitor, keyboard, etc. and we've rented a cabin as a base of operations while we're up there. It'd be nice to have a real computer for things like off-loading pictures from the digital camera, updating our diary pages (although we probably won't publish or even check email while we're gone) and maybe even getting some real work done in quiet moments. I may even take two systems, both for redundancy and so that Barbara and I can work at the same time. I'll take at least one system with a tape drive or CD burner to protect any data we generate while we're there. Short shrift time on mail, I'm afraid. -----Original Message----- > his clock loses 15 minutes every 48 hours or so < I believe this is a problem with Win9x, as I have experienced it since moving to Win95 OSR2--the time slides farther off (slower in my case) the longer the machine runs. On my Win98 system, it is NOT the system clock; rebooting to BIOS and checking the time shows the system clock to within seconds of accurate time. While Win98 is running, however, the time drifts--apparently separated from the actual system clock. My solution is to reboot at the end of the workday, which restores proper time to the OS; it's never more than 10 to 15 seconds off that way. There have been a couple Crucial Updates aimed at this problem, but so far, none have fixed it on my Win98 system. This is not a problem with our W2k machine. --Regards, Chuck Thanks. I don't run Win9X much except for doing screen shots and so on, so I wasn't aware of this. I did check on a couple of Win98SE systems that've been running for quite a while and didn't notice any problem, so it appears not be universal. -----Original Message----- Bruce Edwards asked about moving Outlook Express files from one machine to another. I think the key thing may be to get the OE files into the proper directory. In the version of OE which comes with IE 5.5 you can use the Maintenance tab to move the storage location (thanks to Dr. Keyboard for this hint). Or create the new identity and track down where that stores files and place them there. I'm pretty sure that once the files are in the right location it'll work if the versions of OE are the same. ------------------------- Thanks. -----Original Message----- Robert, You said: >[...] >For example, if you tracert this site, you'll see at the end that you >arrive at pair networks at about the third-to-the-last hop, which is >the pair networks gateway. The next-to-the-last hop will be a private >address, something like 192.168.1.6. That's ugly-looking, but occurs >simply because they're using address translation internally, and does >no harm so long as pair is not advertising that route. I ran this tracert a number of times and noted that pair is now doing something to block reporting of that penultimate hop. Perhaps they read your note and changed things? Looks better this way, IMHO. 7 141 ms 130 ms 140 ms pairnet-1.uspitb.savvis.net
[209.83.160.130] Now granted, this is only one data point and things are probably different with pair's connections to their other backbones (Digex, UUnet, Sprint, Genuity), but it is still interesting to see how one good ISP handles these things. ___________________________ It's really no big deal. Pair is using address translation on their internal networks, but isn't advertising the route. It looks ugly, but other than that there's no real downside. -----Original Message----- Robert, I think I might have some thoughts as to why Barbara's computer is running SETI faster than your main workstation. In short, Windows NT Server is different from Windows NT Workstation. Microsoft did some performance tuning to both flavors of NT to help them function more appropriately for their intended purposes. Some of the differences include: · Caching is different on NT Workstation versus NT Server. NT Server has better network throughput while NT Workstation has better local disk access time, i.e. NT Server prioritizes requests from the network. · NT Server has a server service that enables one to tune the server for its primary role: application server or file/print server. NT Workstation does not provide this feature, because it is limited to 10 inbound network sessions. · The server filesystem driver used in both versions of Windows NT (SRV.SYS) is more subject to paging under NT Workstation than under NT Server. When NT Workstation runs out of physical RAM, it pages this server code to disk, causing its network sharing performance to take a hit and conversely, giving local application performance a boost. NT Server rarely pages this server code out; to do so would impair part of its basic functionality. You may want to try some tests by varying NT's settings in these areas. Check the Performance tab in the Control Panel System applet. 'Maximum' gives the best foreground application responsiveness; the middle setting makes foreground apps slightly more responsive than background apps; and 'No Boost' makes foreground and background apps run with the same priority. On NT Server, check the Server service properties (optimization) in the Control Panel Network applet. You can gather additional data with Task Manager or Performance Monitor (i.e. how much CPU resources is SETI using on the each computer, how much Available Memory remains and how much Pagefile is being used.) Performance Monitor can be used to view the load over a certain period (say, any given SETI run). Pay special attention to the memory load, paging, processor load and the swap file. You should also check the NIC's throughput. With the above you might get a better idea what is causing the SETI performance difference between your computers. ___________________________ Thanks. Actually, as it turns out, the problem is not that Barbara's system is unusually fast, but that my system is unusually slow. And it's only one of several running NT Workstation. So perhaps the problem is that it's an SMP box. I don't know. -----Original Message----- I have seen quite a few systems which lose time. I haven't really investigated the phenomenon but it seems to relate to: 1. Power users. At work, those of us who use a number of different packages seem to suffer worst from this. People who use, say only Office apps seem to be reasonably OK. 2. Browser usage. Note thet most, but not all, our power users have Internet access through a proxy. 3. Winsock apps. I seem to lose most time when remotely administering a control system. I lose very little when working entirely locally on, say, a report using Word. This might help David Stacey Congrats on your site. It's great. Thanks, and thanks for the kind words. -----Original Message----- Bruce Edwards might like to look at Microsoft's Technet Article Q175037. This covers what he needs to do. Microsoft states "Use this information at your own risk". However it works, to the extent of being able to have your mail folder on another PC on the Network. Thanks again. -----Original Message----- Hi Robert: Thanks to your readers, it looks like I may have a solution to the Outlook Express Problem: Express Assist 2000 (http://ajsystems.com/oexhome.html) I am confused about the signature line in my e-mail, in the copy of my original message (see bottom message below) there is a signature line with URL. Thanks again for your help, Bruce :-) Glad you got it fixed. As far as the sig, dunno. Perhaps I cut it out unintentionally. A lot of stuff like that happens when I'm moving fast. -----Original Message----- Here are a couple of interesting links, if you haven't seen them: http://www.osopinion.net/Opinions/MarkSummerfield/MarkSummerfield3.html http://www.osopinion.net/Opinions/JoeriSebrechts/JoeriSebrechts5.html — Holden Thanks. Both of those articles are well worth reading, and interesting counterpoints to each other. If we're to avoid what the first article describes, it would appear that Linux is the best hope. But if Linux is to become a viable desktop alternative to Windows, the security issues brought up in the second article need to be addressed. And they need to be addressed by the Linux vendors themselves, because there's no way that 10,000,000 new Linux users are all going to get things right.
|
|
wpoison
Search [tips]
|
Saturday, 23 September 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] It's been one of those mornings when I'm too busy getting work done to get any work done. Running web stats for my own site and Pournelle's. Helping Barbara get her trip photos scanned in so that she can write her trip report. Cleaning up after Malcolm, repeatedly. Getting breakfast for mom. And so on. At least everyone seems to be over whatever bug we apparently had the last couple days. When I was running web statistic reports last week, Analog (the program I use to generate stats) blew up while running Pournelle's year-to-date report. Well, it didn't blow up. I started it and it just kept chugging away for hours without finishing the report. I finally concluded that perhaps there was a memory problem. This morning, it did pretty much the same thing on his YTD reports, which has millions of hits to be sorted and massaged. This time, though, Windows put up a notice that I was running short of virtual memory. That seemed odd. The machine has 64 MB of physical memory and another 128 MB of virtual memory. When I fired up Task Manager, though, it showed that Windows was using nearly all of the memory, physical and virtual. Analog was occupying about 50 MB itself. Not good on a system with only 64 MB of real memory. So I closed FrontPage, which I usually keep up and minimized to make it easy to publish. That helped a bit, but not enough, so I also closed the SETI client, which occupies about 15 MB. After I did that, the report ran normally. So I decided to stop running the SETI client on this box (meepmeep, my Roadrunner box). I put a copy of the stop_after_send.txt file that tells SETI to finish the current unit but not request another. I probably should have put 128 MB in this box, but 64 MB seemed enough at the time. Barbara gave me a haircut this morning, using the new Ferrari® hair clippers she got me as an anniversary present. They had a picture of Al Borland from Home Improvement on the front of the package, so they must be good, right? Barbara passed on the hair clippers with Toolman Tim on the front. Those were gasoline powered. -----Original Message----- Robert, You may be on to something re SETI on an SMP machine. I think the largest part of the improvement in the performance of my machine came from setting SETI's processor affinity to processor 2 only. Give it a try and see if that doesn't give improvement. PS that does not seem to be a persistent setting, I notice when I reboot I have to reset the affinity. Richard Sherburne Jr. I was pretty sure I'd tried that when I first installed SETI on my SMP system, but after receiving your message I tried it again. Alas, my system went from processing a work unit in about 10:45 to requiring about 12:30. The problem with Set Affinity, of course, is that it doesn't dedicate a processor to a process. All it does is forbid the process from running on any but the specified processor. During the normal course of things with no affinity set, my 2nd processor is normally pegged at 100% CPU utilization anyway, indicating that NT automatically pretty much dedicates that processor to SETI. -----Original Message----- I've come across the problem of PCs losing time an awful lot in the years I've been writing the Dr Keyboard column, and your suggestion of a duff RTC has always seemed the most likely to me. The motherboard manufacturers aren't spending a lot of money on these things - I guess they're the equivalent of a 99 cent gas station watch without the quality checks - so you're getting what you pay for. The solution is to have your machine check the time with someone's atomic clock over the Internet. I use a little applet called About Time from Paul Lutus, http://www.arachnoid.com. It'll check as often as you like and reset your system clock. It also works as a time server - I have it set up here so that the NT server checks once a day with an atomic clock in the UK and then every other PC checks with the internal server an hour later. It even adjusts for the amount of time it takes to get the correct time from the time server. So the log file for Vito my W2K workstation says: Connecting to Belushi using SNTP. Regards Chris Ward-Johnson Dr Keyboard - Computing Answers You Can Understand Thanks. I used to run a similar freeware SNTP (Simple Network Time Protocol) application on my main system, but that was when it was the machine that dialed my ISP. When I brought up meepmeep, my Roadrunner box, the app would no longer function because it wasn't running on a directly-connected system. When I rebuilt kiwi, I blew away that app unintentionally, and am not running anything similar on meepmeep. I should probably download the app you mention and install it on meepmeep. Incidentally, as you're probably aware but others may not be, one can set the time on a client machine from a network server using the command NET TIME [\\computername | /DOMAIN[:domainname]] [/SET]. For example, if I install the SNTP client on meepmeep and use it to keep meepmeep's clock accurate, I can then set the time accurately on kiwi by typing the following on kiwi at a command prompt: NET TIME \\meepmeep /set -----Original Message----- TIME etc. etc. is fine if you have a memory for running it or the batch file which runs it or whatever. The beauty if About Time is you set it and forget it. Looking at it yesterday was the first time I've opened it in months. And it works as a client and server - the same app's on the NT box and the workstations. Thanks. I may give that one a try. -----Original Message----- Bob, I work at a very small mechanical design/intellectual property company. Most of our engineering work requires use of 3D CAD software which we run on modest workstations under NT 4.0 We have observed that dual CPU Pentium Pro 200 and Pentium II systems were qualitatively slower than single CPU systems with identical processors. Not much time was spent trying to analyze this apparent difference; however, at the time we assumed that the inevitable extra overhead required to manage two CPU's was a greater burden than whatever advantage our applications were able to obtain from the additional processor. Could this be what you are observing as you track SETI times? Thanks for sharing your insights and experiences with things computerish as well as your experiences with the pooches. I frequently dog sit my daughter's Rotty and have a real soft spot for canine type people. Best regards, Frank Fitz Dunno. You may be right. But I suspect it has something to do with the number of background services and processes running, as well as the fact that I'm pounding on this system all day long. It shouldn't matter. That is, when SETI reports that it requires 10:45 of CPU time to process a unit on this system, it shouldn't matter whether it's able to get that 10:45 of CPU time in 10:45 or if it takes a week. But it seems to matter, at least some. -----Original Message----- You mentioned last week that you were pretty sure that SETI@Home is not multiprocessor aware; you're exactly right. My guess is that SETI@Home is thrashing by NT bouncing it's thread from CPU to CPU. Bring open your Task Manager (hit Ctrl-Esc), go to the Processes tab, right click on the seti.exe (whatever it's called), and choose Set Affinity (this selection is only available on multiprocessor machines), and choose just one CPU. If you want to really crank up SETI@Home, install a second copy to another directory, and set it's Affinity to your second CPU. You'll double your units/day. Unfortunately, I don't know of a utility which can set the affinity, so this is a manual step after every reboot. Thrashing alone is unlikely to account for all of your workstation's poor showing; you very likely have other CPU intensive applications that get in the way. A single one of your Pentium III 550 CPUs should certainly beat her Pentium III 450. Even a "blinking light" application will keep steal CPU cycles. Ditto for hardware that generates lots of Interrupts, especially a legacy ISA card. Doubtless others have pointed it out, but, if you're after big performance, forgo the pretty display and just run the command line version. Thanks to your WinGate and full time Internet connection, you don't need to cache Work Units, but there are utilities out there that can; SETI@Home has been very available, but folks with a full time Internet connection and a real competitive streak want to make sure that they don't have any idle cycles due to SETI@Home not being able to hand out Work Units. I stopped running SETI@Home quite a while back in favour of something I've been running even longer; I've happily signed up with the Daynotes Gang to contribute my measly 37 WU. Andrew Colbeck Thanks. Yes, the first day I started running SETI, I tried running a second instance of SETI on my dual processor system per the instructions in the FAQ. It simply didn't work. I installed a second instance to a different folder as they suggested, but when I fired up that second instance, it killed the first instance. And when I fired up the first instance, it killed the second instance. And when I uninstalled the second instance, it removed the first instance instead! Perhaps the problem was that I used the default folder for the first instance and for the second instance I simply added a "2" on the end of the folder name, giving me two similar folders in my program files directory, seti@home and seti@home2. Given that those both have more than eight characters, perhaps SETI got confused. I haven't tried re-installing a second instance with an entirely different directory name. Perhaps I should.
|
|
wpoison
Search [tips]
|
[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Thanks to Bo Leuf for pointing out the Microsoft Corporate download site. That's where you can download updates and patches that can be stored on a distribution server and applied to as many machines as necessary without downloading the patch each time. I actually had this site bookmarked at one point, but somehow lost it. Be forewarned. Bo mentions that the site requires cookies and ActiveX be enabled, but it really goes further than that. They require you have Internet Explorer configured to "have your way with me" before they're willing to let you view the page. When I first clicked on the link, I got only a blank page. So I added that URL to my Trusted Sites category and did a refresh. After clicking on literally 20 or more dialog boxes to tell IE it was okay to download or run an ActiveX control, I finally aborted the process and reconfigured Security settings for Trusted Sites. I set them to Medium. Now, as far as I'm concerned, Microsoft's settings for "High Security" are actually Low Security. Microsoft's "Medium Security" is actually what I'd call No Security at All to Speak Of. So I wasn't happy to change those settings, but I did so. Then I sat there watching as Microsoft "initialized the web site." The actual message is "Windows Update is initializing the web site. This is done without sending any information to Microsoft." What they really mean, of course, is that they're configuring my system by downloading and running a ton of ActiveX controls and other nasties, which makes me very unhappy. Why couldn't they use simple HTML for this page? There's absolutely no reason not to, except that they're Microsoft and they want absolute control of your computer. I finally got tired of watching and minimized that instance of IE to the task bar. After literally an hour, the task bar icon was still blinking between "Windows Update" and "http://corporate", indicating that stuff was still being downloaded and installed. They make me go through all of this just so that I can download patches to their buggy software in a form that's usable. Have I mentioned lately that I hate Microsoft? The "Just Say No to Microsoft" campaign that I initiated here some time ago seems to be catching on. I was just looking at preview screen shots of Office 10 the other day. As I looked at them, I found myself thinking that this was the only time I'd see them, because I'm sure never going to install Office 10 on any of my production systems. Office 2000 is the end of the line for me. My next upgrade will be to a system running Linux. Until that happens, I'm frozen at NT 4 and Office 2000 for my production systems. I'll run Windows 2000, Windows 98/Me, etc. on my test-beds for writing purposes (capturing screen shots and so on) but I've opted out of Microsoft's vicious upgrade cycle. Obviously, they want to get as many people as possible locked in to their software so they can begin renting it, but I won't be one of them. I wish I could abandon Microsoft operating systems and applications today for Linux, but Linux just isn't sufficiently developed and friendly for that to be feasible, at least for me. I have to continue to use Microsoft operating systems and applications to get my work done, but that doesn't mean I have to go along with their vicious upgrade cycle. If it takes a year, I'll still be using Windows NT 4 and Office 2000. If it takes two years, I'll still be using Windows NT 4 and Office 2000. But at some point, Linux will have matured enough as a desktop operating system to allow me to abandon Microsoft. And at that point I'll drop Microsoft and never look back. And I suspect I'll have lots of company. That's what scares Microsoft, of course. And that's why just about everything they're doing now is ultimately aimed at locking people into their operating systems and applications. That's what Windows 2000 is about. That's what Office 10/2002 is about. That's what updates to IE are about. Microsoft wants a direct line into your bank account, so they can deduct money from your account every month or every week or every time you use their software. And I'm not going to play their game. Just Say No to Microsoft. Update: I finally got tired of watching that Task Bar icon blink, so I decided to close Internet Explorer. Microsoft wouldn't let me. When I attempted to close it, the system just beeped at me. The Borg took control of my system. So I used Task Manager to kill the process. Thinking that perhaps it had just locked up in a loop, I restarted it, but with the same results. I finally killed it, searched down all changed files on my hard disk and deleted any of those downloaded from this site. I HATE Microsoft. Jerry Pournelle occasionally threatens to borrow a guided-missile frigate and sail it to Redmond. As I say to him each time, "You drive. I'll shoot." -----Original Message----- I forgot to mention one other thing, sorry I must be getting old:) When I set processor affinity to the 2nd processor, I also gave seti priority on that processor. I have ordered a set of PPro 512k prossersor to drop in place of my 256k units so we may soon have a better feel for cache size impact on seti. Now, if I could only convince my wife that a pair of 1mb cache processors was a reasonable purchase just to chase down this theory :) I'd actually considered doing that (by boosting the SETI process to High priority, or even Realtime) but the problem with doing that is it effectively turns my dual Pentium III/550 into a single Pentium III/550. As far as the processors, I wouldn't bother. According to a SETI FAQ I was reading (over on Ars Technica, I think) even a processor with 1 MB of L2 cache can't fit the entire SETI data set. Apparently, there are actually people going out and buying Pentium III Xeon processors with 2 MB of L2 cache just so they can process SETI faster. That strikes me as the height of something or other. -----Original Message----- I will admit I've had no problems at all running multiple copies on an SMP box, although I will admit I use the command line version under NT 4. With this version there is a command line option to indicate which CPU it is to run on. Hmm. Perhaps I should give that a try. I confess that I assumed that running the GUI version with the graphics display switched off would be about as efficient as running the command-line version, but perhaps that's wrong. Or perhaps I should try running the Windows 98 version. According to the SETI FAQ on Ars Technica, the Win98 client is faster on a given machine than either the NT version or the Linux version.
[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] |