The User's Column, July, 2006
Column 312, part 3
Jerry Pournelle jerryp@jerrypournelle.com
www.jerrypournelle.com
Copyright 2006 Jerry E. Pournelle, Ph.D.
(Continued from last week.)
The big news this week is that Intel has dropped the embargo on releasing the performance test results of its new Intel Core 2 Duo chips, and these chips blow the doors off the competition. Way back in February, Intel announced that when Core 2 Duo came out it would beat AMD's best chips by 20%. Whether or not it has achieved that result depends on the tests you use, but nearly everyone agrees that with the Core 2 Duo, Intel has moved back into first place in the eternal AMD vs. Intel contest. In the former Intel Dual Core chips the AMD chip architecture was more elegant than Intel's, in that AMD put the communications between the two halves of the dual processors inside the chip, while Intel had to go to the external bus for that; but with the Core 2 Duo this has changed.
Intel's new chip is a true multicore, unlike the Pentium D it replaces. Core 2 Duo shares one L2 cache and one front-side bus between its cores, so memory requests run more efficiently inside and out. The bottom line is that Intel's best is faster than AMD's best.
Bob Thompson says that with some benchmarks, Intel's $175 Core 2 Duo is faster overall than AMD's FX62. Incidentally, the AMD-2 systems have a DDR2 memory controller. Bob Thompson says "Depending on the benchmark you choose, you can make DDR2 (faster access) look faster or DDR (less latency) look faster. Overall, real world, there's essentially no difference in performance with DDR versus DDR2.
It is the general consensus that Intel now has faster systems than AMD. In practical terms this may not matter a lot. Few people will need as much performance as AMD's current best can deliver. Diana, my "AMD best" system (see July Part One), plays every game I know of, and for most games upgrading the video board will have a more noticeable effect than upping the raw CPU power. This is not to say that over time there won't be programs, beginning with games, that will take advantage of the improved CPU speeds; but for the moment there's little practical software, including games, that needs as much power as the Intel Core 2 Duo can deliver.
Now that Intel has moved ahead in this leapfrog game, and is likely to stay ahead on pure performance, we are back to competition on price, and that's good for all of us. Both AMD and Intel will have sweet spots - a speed/price combination that is clearly the right choice - and the price of systems built around that sweet spot will get progressively lower.
Apple has moved nearly all its systems to Intel chips. No Apple system yet announced (as of July 15, 2006) makes use of the Intel Core 2 Duo; all present Apple Intel systems make use of the older Intel Core Duo, which is found in the latest iMac systems.
On the other hand, Apple hasn't announced a replacement for their powerful G5 line. At the top of that line is the Quad 2.5 GHz Power PC G5. We can be fairly certain that Apple will bring out an Intel Mac G5, and I've heard speculation that the top of that line may be a Quad Intel Core 2 Duo machine. This beast would have 8 processors. No current workstation version of Windows can make use of that many CPU's. I've also heard that the Core 2 Duo can't be used in a four-processor (8-core) configuration, and a search of available tech documents doesn't tell me one way or another.
My immediate reaction was that a Quad Core 2 Duo OS X system would be in great demand from the digital imagery world, and I figured that Hollywood would be in a mad race to place orders for these machines if Apple announces them at their Developer Conference next month. However, Peter Glaskowsky reminds me that when SGI released a four-processor Windows NT workstation using Intel's Pentium III Xeon processor (the SGI 540 Visual Workstation), not many were sold. Compared to a two-processor machine, a four-processor machine will generally be more than twice as expensive and less than twice as fast.
All this is rather idle speculation for the sheer fun of it.
I've been persuaded to wait for the Apple Developer Conference before ordering a new Intel Mac to see what systems will make use of the Intel Core 2 Duo chips, but the reports I have on the Core Duo iMac with 2 GB of memory are all very positive. The Core Duo iMac can run both MAC OS X and Microsoft Windows XP at the same time using a program called Parallels (http://www.parallels.com). The OS X runs natively, of course; Windows XP doesn't go quite as fast under Parallels as it would in dual boot, but all reports have it that Windows on the Mac has acceptable performance.
You need the full 2 GB of memory to run both operating systems simultaneously. The proper way to get that memory is not to order it from Apple. You can get premium memory, Crucial or Kingston, much cheaper, and it will work just as well.
Note that the iMac doesn't come with any instructions on how to add memory, but there are plenty of illustrated step by step tutorials, such as http://www.wap.org/journal/imac/imacmemory.html, available on the Internet. It's not particularly tricky.
Of course any Mac you buy now won't have the new OS that's coming out as Beta this fall (release expected in 2007), but that's not a major reason to wait. My Mac came with OS 9, and it was no big trick to upgrade to OS X. I hadn't intended to have to do that, but the Glendale Apple Store hadn't bothered to tell me that OS X wasn't installed on the Mac PowerBook they sold me.
I've been persuaded to wait for the next Mac announcements before getting an Intel Mac, and perhaps you should too, but I'm sure hearing good reports on the systems you can buy right now.
Depending on who you listen to, AT&T and the big phone companies are about to end the Internet. John Dvorak has already written a column predicting that we live in an Internet Golden Age, and will look fondly back on it in future years.
Others say we are in the fight of our lives. It is not too late to stop the greedy monsters from eating the Internet alive, but we must all be vigilant, and lobby Congress.
We wrote about all this back in February. At bottom it comes down to this: AT&T and other big communications carriers watch as the Internet is used to deliver more and more information - movies on demand, video and audio podcasts, music and songs, and even Voice over IP - while the phone company doesn't get any more revenue. Edward Whittaker, CEO of SBC/AT&T, said back in 2005 that Google and Amazon and the Movies on Demand people "would like to use my pipes free, but I ain't going to let them do that because we have spent this capital and we have to have a return on it. So there's going to have to be some mechanism for these people who use these pipes to pay for the portion they're using. Why should they be allowed to use my pipes?"
In an attempt to get more revenue from "his pipes", Whittaker and allies have been spending lots of money lobbying in Washington to kill legislation specifically requiring "net neutrality", and seem to have been successful: net neutrality amendments to the upcoming revisions of the Communications Act have been killed. On the other hand, their lobbyists don't seem to have been able to get provisions exempting them from the class action lawsuits that are sure to be brought the instant that AT&T and other big carriers begin to discriminate against one or another kind of message traffic.
The AT&T argument is that Movies on Demand use up a lot of bandwidth, and those making those deliveries and collecting revenues ought to pay more than, say, companies that sell actual merchandise over the Internet and thus don't use up so much bandwidth. They also want to charge more for Vonage and Skype and other Voice Over IP companies. VOIP not only uses AT&T bandwidth, it also directly competes: those who sign up for VOIP, like my young friend who is spending a semester in China and daily talks with his fiancé through Skype, won't be paying AT&T rates for phone calls. To AT&T this is adding insult to injury.
The "net neutrality" argument is that discrimination against any kind of Internet service is not only unfair, but puts one on a slippery slope that leads to censorship. All this and more is presented at http://www.savetheinternet.com/, and anyone with the slightest interest in this matter ought to have a look there.
The whole situation is complicated by the total ignorance of nearly all the Senators and Members of Congress who will decide these matter. I don't mean the general membership of those distinguished bodies; one wouldn't expect the random odd Senator to understand such technical matters. I mean that those on the relevant Committees don't understand the matter, and many of them are pretty well in the pockets of the big Telcos. For example, see http://www.publicknowledge.org/node/497 for Senator Stevens of Alaska's understanding of tubes and pipes.
Stevens' addlepated remarks provided an illustration of the problem as some see it: there was an attempt to suppress dissemination of his inane speech. On the other hand, the suppression didn't work.
The argument against "net neutrality" is that those who flood the Internet with packets ought to pay more than those who don't; intense users of resources generally expect to pay more.
The counters to that argument are two: first, technically, it is very difficult to determine what's in an Internet packet without delaying it. It's a bit like the Heisenberg uncertainty principle. The attempt to sort out packets according to their content could itself bring the Internet to its knees. Secondly, the customers of AT&T, and cable modem, have already paid for use of AT&T's pipes. The companies that send movies on demand, or podcasts, pay for the bandwidth they need. Why should everyone pay twice? But once again, those who download movies are using more bandwidth than Aunt Minnie who doesn't look at pornography and mostly uses the Internet to look at her email, an occasional photo of her grandniece, and to order flowers and chocolate. Why should Aunt Minnie pay as much as the bandwidth addict?
These questions aren't going to be settled easily or quickly.
In any event, the most important thing we must do is let Congress know that there are a lot of us with interests counter to those of AT&T; and since it's unlikely that any Senator of Member will really understand the matter, we need to see that at least some of the relevant Committee Staffers - who will understand what's going on - are on our side. We also have to have some understanding of what "our side" is.
The important thing is to ensure competition. David McCord Wright points out that Marx's prediction that laissez faire competition generally ends up with a single unregulated monopoly as businesses consolidate was often correct, but in the United States the anti-trust acts prevented much of the cartelization of industry that plagued Europe. The original AT&T was a regulated public utility. The new giant is a carrier of a different color.
While some lobbyists act directly with actions that come close to outright bribes to Members, most work on the staffs, giving them free "educational" conferences (with lots of free food and drink), and providing them with technical papers slanted to favor the lobbyist's clients.
The computer user community is pretty naïve about how Washington works, which is why horrors like the Digital Millennium Communications Act get made into law. As a group we think we're pretty smart, but we don't always act that way. It's about time we wised up.
My first experience with the Microsoft Wireless Keyboard and Mouse led me to believe that the keyboard works only with machines that have a PS/2 input.
That turns out not to be the case. I did manage to get the Microsoft Wireless Keyboard working with LisaBetta (USB only) up in the monk's cell; the problem was batteries. The Wireless Keyboard uses up batteries fast, and alas, there is no way to know that. If the batteries are dead, the transceiver unit connected to the computer will hunt for a channel to the keyboard, but if it never finds one, it doesn't really tell you that. It just stops looking. There's no LED or any other indication that the keyboard is working, and no indication that the keyboard has battery power. This is a rather serious defect in what is otherwise an excellent system.
A pair, or better a trio, of lights with all three lit when the batteries are full charged, diminishing to two when it's lower and one when it's definitely time to change batteries, would fix the problem entirely.
My solution to the problem is to have four NiMH batteries and a charger that came with one of my Olympus cameras. Every night when I'm done writing in the monk's cell, I take out the batteries that I used that day, and put them in the charger. I leave the keyboard upside down with the battery case open, and next day I put in two freshly charged batteries before I start. That works very well.
Now that I have used it for several days I have become quite fond of the Microsoft Wireless Keyboard. I like the curved key rows, the useful built-in wrist rest, and the general similarity to the old IBM Selectric keyboard. Indeed, I like it enough that I am going to look for keyboards with that layout, either USB or PS/2, next time I am at Fry's. It really is a well designed keyboard, but its tendency to run out of juice without giving any indication of the problem is annoying.
This week we saw Pirates of the Caribbean, Dead Man's Chest, and it's a hoot. Roberta says "If you like mad chase scenes, you'll love this." She laughed all the way through the movie, perhaps more than I did, and I thought it was very funny. I don't share Johnny Depp's politics, but he is one fine actor; and Keira Knightly is just perfect as the not so helpless ingénue. The special effects are stunning.
The movie is two hours long. It didn't seem that long to me. However, the movie stops rather than ends: that is, it's a setup for Part Three of the Pirates saga. If you really hate cliffhanger endings, you probably ought to wait for the DVD and watch it after the release of the next sequel. Having said that, we're not at all sorry we saw it now.
If you see this movie, wait through the nearly interminable closing credit roll for what is in effect the last scene of the movie. It will take a long time, but it's worth it.
For many years my monthly columns ended with a section called "Winding Down." This contained reviews of movies, games, and books both technical and for general reading. These have been collected into one place (http://www.jerrypournelle.com/reviews/bookmonth.html), and that page is now up to date. We may reorganize and index it for Chaos Manor Reviews, but that won't happen for a while. There are more than ten years of monthly reviews on that page.
To be continued...