Monday, 23 June 2003
9:55 - I'm still chugging away on the galley proofs. Last night, I made it up through page 509 before my eyes started to get bleary. In theory, I'm supposed to complete the galley proofs today, but I don't think I'll be able to do that. My editor did say Friday that, given the good progress I'm making, a couple extra days should be do-able.
Intel ships the 3.2 GHz Pentium 4 today, its final Pentium 4 based on the 0.13µ Northwood core. I've had a sample for some time now, but haven't had time to do exhaustive benchmarking on it. What testing I've done confirms that this is an extremely fast processor. On the other hand, you pay a very high price for that speed. On the gripping hand, sometimes that doesn't matter.
In one sense, the $700 price of the 3.2 GHz is very high, especially when you consider its performance relative to processors that cost a small fraction of that amount. But in another sense, that $700 is a non-issue for some people. The cost of the processor pales, for example, when compared to the salary of a highly-paid executive. If that processor saves even a few seconds a day of that executive's time, it pays for itself very quickly. Same thing in time-sensitive applications like commodities trading, where literally a fraction of a second can make the difference between a big profit and a big loss. For most of us, of course, fractions of a second don't matter, so we'll continue to use the less expensive Pentium 4 models. But we benefit indirectly, because each time Intel introduces a new, faster processor, their older, slower models are pushed further down the food chain, bumping the price/performance ratio of their slower models higher and higher.
AMD must be desperate to get their Lateron out the door. Their fastest Athlon processors are barely a match in most benchmarks for the 2.6 GHz Pentium 4 without Hyper-Threading Technology enabled, and their QuantiSpeed PR ratings for the Athlon became a joke around 2500+. At 3.2 GHz, the Intel Pentium 4 walks all over the fastest Athlon available. And I'm not talking about just in benchmarks. The performance difference when you're actually sitting in front of a machine is substantial. And that's without HTT enabled and running HTT-enabled benchmarks. Running HTT-enabled applications puts the fast Pentium 4s in an entirely difference class. AMD really, really needs to get the Opteron out the door if they're to have any hope of continuing to compete with Intel for the performance desktop market. With the 0.09µ Prescott-core Pentium 4s on the near horizon--which Intel will probably call the Pentium 5--even the Opteron may not do the job. It's going to be an interesting next few months.
Tuesday, 24 June 2003
9:49 - I'm still chugging away on the galley proofs. Last night, I made it up through page 738 before I keeled over. That was 229 pages proofread for the day, which doesn't sound all that much but is. I'll finish the galley proofs today, one way or another.
I got email from several readers who commented on the lower price of AMD Athlon processors relative to Pentium 4 processors with equivalent performance. That used to be true but is no longer. Or perhaps I should say it remains true only if one accepts the performance rating that AMD assigns to its processors, which it bases on a very carefully selected set of benchmarks.
In the benchmarks I use, for which I'm very careful not to allow any pro-Intel or pro-AMD bias to creep in, the fastest AMD processors (the 3000+ and 3200+) are roughly equivalent to the Intel Pentium 4 running at 2.53 or 2.6 GHz. Certainly, there are some benchmarks in which AMD processors do extremely well, sometimes even in excess of their PR numbers. But overall, AMD's PR numbers are grossly inflated. If you compare actual performance versus price, AMD processors actually provide substantially less bang for the buck than Pentium 4's, at least once you get above the low-end processors. At the mid-range and high end, AMD's pricing and performance is all smoke and mirrors. To put it bluntly, the fastest (and most expensive) AMD Athlon XP barely matches the performance of a sub-$200 Intel Pentium 4. There, I've said it.
Wednesday, 25 June 2003
9:45 - I finished the galley proofs yesterday afternoon, and they're all back to Sarah Sherman, my production editor at O'Reilly. I don't know much about O'Reilly's production process, so yesterday I asked Sarah if it would be terribly disruptive if I added a page or three to the processors chapter. I'd like to write about the Opteron/Athlon 64 and the Prescott/Pentium 5, but when I did my last Word manuscript submission in April I didn't have enough information yet to write about them. Sarah said that'd be doable if I could get the Word manuscript to her by tomorrow morning. She'll then send it to the Tools department at O'Reilly, who will convert it from Word to Frame format and send it back to her for incorporation in the chapter. She can re-flow the chapters from there with no problem, but without the Tools conversion she'd have had to manually type what I sent her into Frame. So today will be spent trying to write a coherent section on "Forthcoming Processors" or suchlike.
Also yesterday, I asked Sarah and Brian Jepson (the primary editor for the book) if it'd be okay if I posted "final" QC1 PDFs for my subscribers to look at. These PDFs will be ones that include the edits I've been making for the last week or so, and will be quite close to the actual final appearance of the printed book. Here's the email exchange:
-------- Original Message --------
Brian and Sarah,
For the last edition, I uploaded the final PDF chapter documents to my web site (with O'Reilly's permission), where they were available for my subscribers to download. Very few of them (probably half a dozen) took advantage of that, but we did catch a few errors that way. I trust my subscribers not to distribute the PDFs, so I'm not too concerned about them "getting out".
Obviously, we'd be on a very tight schedule, so I'd make it clear to them that we had only two days (or whatever) for them to download and read any chapter(s) they were particularly interested in, and comment on any errors they might find. I'd also make it clear to them that this wasn't a time for any major changes, just minor corrections, fixing typos, and so on.
I firmly believe that the more eyes that look at the final PDFs, the better. In effect, my subscribers would be acting as unpaid proofreaders.
What do you think? If you're in favor, I'll post a notice to that effect and post the PDF chapter files in a password-protected download area of my website as soon as you get them to me.
-------- Original Message --------
I think that's do-able. I can post the existing PDFs today, along with a consolidated copy of the edits I've already added, so that people won't waste time commenting on things I've already caught.
Alternatively, if Sarah can generate new PDFs with the edits I've made already incorporated I could post those PDFs (I don't know what the status of the changes Sarah is entering is, or how hard it is for her to generate new PDFs).
Tell me what to do and I'll do it (aren't I easy?)
-------- Original Message --------
That works for me. I'll let my subscribers know.
17:53 - The almost-but-not-quite
done chapters for the third edition of PC Hardware in a Nutshell are now
available for subscribers to download in PDF format. Please keep these files
confidential. I'd hate for them to get out on the Internet. I'm posting these
files for any subscriber who cares to read one or more of them and comment on
any errors he or she finds. I need any corrections back on or before 9:00 a.m.
EDT Monday morning 30 June 2003.
And I'm now chugging away on a supplementary section for the processors chapter, which I hope will cover the Opteron/Athlon 64 and the Prescott-core "Pentium 5". We'll see how it goes.
22:08 - I've posted the current draft of the supplementary material I plan to add to the processors chapter, in the form of a Word document, over on the Subscribers' page. This material covers the AMD Opteron and Athlon 64 processors, as well as the Prescott-core Pentium, which Intel may decide to call the Pentium 5. This draft is current as of 10:00 p.m. on Wednesday, June 25th. I have to get this to O'Reilly by tomorrow morning, so if you're going to comment on it, please do so quickly. After 14 almost solid hours at the keyboard, I am completely whacked.
8:40 - I can finally see the train at the end of the tunnel. I have the rest of this morning to finish up the section on Opteron/Athlon 64 and Pentium 5 that will be added to the Processors chapter. It needs to be up to Sarah at O'Reilly by lunchtime. After that, I think I'll collapse for a while.
12:15 - I've just sent the final draft of the supplementary material I plan to add to the processors chapter up to my editor at O'Reilly. It's posted in the form of a Word document on the Subscribers' page. This material covers the AMD Opteron and Athlon 64 processors, as well as the Prescott-core Pentium, which Intel may decide to call the Pentium 5. This draft is current as of noon EDT on Thursday, June 26th.
7:41 - Barbara is leaving for the weekend to visit Fletch, her Border Collie Rescue friend in South Carolina. She's taking Malcolm with her. She's been very annoyed with Malcolm lately, so I told her that I planned to count the dog in her truck when she left and compare that with the number of dog when she returned.
Tom's Hardware is talking about flat-panel displays again. Tom thinks that FPD makers should institute a zero-tolerance policy for bad pixels. That's a fine idea in theory, but there's a problem with it in the real world. The LCD panel itself makes up the bulk of the cost of an FPD, and the percentage of perfect LCD panels is not high enough to make Tom's suggestion practical. If implemented, that policy would simply mean that makers would have to eat a large number of LCD panels in order to sell only perfect ones. I don't know what percentage of LCDs have minor flaws, but it may be as high as 50%. If so, the FPD that now sells for $500 might sell for $1,000 instead.
Perhaps the solution is to have separate lines of FPDs, one with perfect panels and the others with flawed panels. Instead of the current hit-or-miss method in which you might get anything from a perfect panel to one with five or more dead pixels, you could pay your money and take your choice. Instead of all FPDs of a given model selling for $500, a perfect version of that model might sell for $1,000, a version with one or two dead pixels for $400, and one with three to five dead pixels for $275. That way, people who are bothered very much by dead pixels (not everyone is) could have guaranteed happiness, while those who don't much care about a dead pixel or two over near the edge or corner of the screen could get a perfectly acceptable FPD for much less money. Kind of like the way diamonds are graded and priced.
Yesterday was the 40th anniversary of President Kennedy's famous gaffe in Berlin. During a speech in which he wanted to convey solidarity with the people of Berlin, Kennedy intended to say, in German, "Ich bin Berliner", or "I am a Berliner". What he said instead was, "Ich bin ein Berliner", or "I am a jelly doughnut".
Don't bother to tell me that this is an urban legend, because it's not. No matter what revisionist "expert linguists" say now, the fact is that Mr. Kennedy announced to the world that he was a pastry. Not long after it happened, a woman I knew who was a native German speaker (and a native Berliner) explained to me what Mr. Kennedy had said and what he intended to say. She spoke fluent, unaccented English, and I trust her translation more than those of these unnamed "expert linguists".
Apparently, Mr. Kennedy was completely hopeless when it came to speaking foreign languages. Before the speech, his team had German speakers coach him on exactly what he was to say. Over and over again, they'd say, "Ich bin Berliner", and Mr. Kennedy would repeat, "Ich bin ein Berliner". They finally got him to the point where he was speaking the line correctly, albeit with a heavy American accent. During the speech itself, all of them were waiting with their fingers crossed. When he got to that part of the speech, sure enough, he announced, "Ich bin ein Berliner", or "I am a jelly doughnut".
Which reminds me of the time a group of Americans was touring a Polish factory soon after the Communist government fell. The factory manager was explaining, via a translator, production quotas and so on. At some point, one of the Americans asked what happened if people didn't meet their production quotas. The factory managed replied, and the translator calmly translated his reply as, "We shoot them." The Americans were taken aback, and soon departed. As they were walking toward their cars, the translator came running up behind them, shouting "Fire them! We fire them!"
That one may be an urban legend. I've never checked. I hope it's not, though. It's too perfect.
11:25 - It's official. We've updated our list of recommended motherboards.
I got a frightening phone call about 10:00 this morning. Patti, whom Barbara works with, called and asked if Barbara was here. I told her that Barbara was at work. Patti said that Barbara had never shown up at work, and they were worried because that was very unlike Barbara. I told her that Barbara had left this morning at the usual time and should have been there by 8:00. I asked Patti if she'd talked to their supervisor, and she said that he also had no idea where Barbara was. At that point, I had no idea what was going on. I tried calling Barbara's desk at work, and Patti answered. (I thought perhaps Barbara had been moved to a new office.) So I called Barbara's cell phone. No answer. I was just getting ready to call the police when Barbara called here to tell me that she'd been in the HR department all morning doing paperwork. She changes from part-time temporary to full-time permanent as of next Monday, and had a bunch of forms to fill out. All's well that ends well, but it gave me a bad moment.
13:25 - The Windows Refund Day maniacs are at it again. Are these people simply dense, or do they refuse to understand the issue as a way of forwarding their own agendum? They have arbitrarily decided that when they buy a PC that has Microsoft software bundled with it, they are somehow entitled to get a refund for just the software. That isn't how it works, as several PC vendors have made clear. The PC, including the bundled software, is a single product. Buyers of that product are not entitled to demand a refund for part of the product and keep the rest of the product. It's as if I went down to my local dealer, bought a Ford Explorer, and told them, "Oh, by the way, I want the 4X4 but not these tires that you've bundled with it." Rotsa ruck.
PC vendors and Microsoft are legally entitled to agree to bundle their products as a single product. Microsoft does not hold a gun to the head of PC vendors to force them to bundle Microsoft software. They do it voluntarily. Nor do all PC vendors bundle Microsoft software. It's quite easy to find a PC with no software at all bundled. But of course that's not good enough for these Windows Refund Day fanatics. They want to be able to choose whatever PC model they want and dictate to the maker of that PC what they are and are not willing to pay for. Jesus.
I just sent the following message to subscribers:
And I'm now sitting here trying to get my phone numbers registered. WebWasher keeps returning time-outs, but only after a lengthy delay each time. I finally got to Step Two after about a dozen retries, and I'm hopeful that I'll be able to complete the process sometime today.
14:23 - My apologies for the bad link on the Windows Refund Day post I made earlier. That link formerly pointed to an executable file, which not unreasonably scared a lot of my readers. In fact, the bad link was an artifact of copy/paste. Earlier I'd been playing around with an MS-DOS USB driver (incredible, but true), and somehow ended up with the URL for downloading that driver still in my paste buffer. The link is now fixed.
9:17 - I finally got the three confirmation emails from the Do-Not-Call Registry for the three phone numbers I'd registered. Our main phone number, my mother's phone number, and Barbara's cellphone number are now all on the DNC registry until 27 June 2008. I'd forgotten that DNC registration is effective for only five years. I'm not sure why that is. If they simply had to have an expiration date for whatever reason, I'd think they could have used a 99-year expiration rather than a 5-year one.
I also resent the large number of loopholes in the DNC registry. It exempts non-profits, which is ridiculous. I don't want to receive telemarketing calls from non-profits any more than I want telemarketing calls from for-profits. I suspect most other people feel the same. The non-profits squealed loudly and effectively, and somehow succeeded in having themselves exempted. I do know that I will never contribute anything to any non-profit that calls me to solicit a donation.
They also exempted political organizations (surprise, surprise), on the theory that restricting them would be unconstitutional. That's a true red herring. Making political organizations subject to the DNC rules is no more restricting their First Amendment rights than it is to forbid them from barging into my home and lecturing me in my den. They may have a First Amendment right to free speech, but I certainly have the right not to listen to them.
They've also exempted banks, insurance companies, pollsters, and any company with which you've done business recently. That is simply stupid. There is a clearly definable difference between ongoing business and sales efforts. If my bank needs to call me to say there's a problem with a check I deposited, or if L. L. Bean calls to say that there will be a delay in shipping something I'd ordered, fine. I'm not going to complain to the FTC, and very few other people (I am tempted to say "no one") would complain about such a legitimate business-related call. But if the bank calls me to sell me some new service or L L Bean calls me to try to sell me something additional, that's clearly beyond the pale.
At least these exempt companies are required to remove you from their lists upon request. My answering machine message currently has a message that tells callers to remove us from their lists. I'm going to modify it to say something like, "If this is a telemarketing or survey call, even if you are currently exempt from the FTC Do-Not-Call provisions, remove us from your list immediately..." They may or may not do that, but I suspect they will. The danger of not doing so is that they would be at risk of the fine of up to $11,000 that the FTC rules mandate for violating such requests.
Well, the DNC Registry isn't perfect, but it's a lot better than nothing. I'd guess we'll all see telemarketing calls reduced by 90% or more. Or at least I hope so. Presumably a lot of the telemarketing companies will be driven out of business, and that's all to the good.
Copyright © 1998, 1999, 2000, 2001, 2002, 2003, 2004 by Robert Bruce Thompson. All Rights Reserved.