Dr. Jerry Pournelle

Email Me

Why not subscribe now?

Chaos Manor Subscribe Now

Useful Link(s)...


Hosting by

Powered by Apache

Computing At Chaos Manor:
The Mailbag

Mailbag for March 26, 2007
Jerry Pournelle jerryp@jerrypournelle.com
Copyright 2007 Jerry E. Pournelle, Ph.D.

March 26, 2007

The mail continues to flow, but I am in Phoenix for the SAS conference, and so my replies are going to be brief.

Rich Heimlich is a long time participant in the computer revolution.

Jerry, just got back from the Game Developer's Conference and all everyone was talking about was massive multicore. It's happening. AMD is doing their FUSION product and one of their guys told me it'll be exactly what we were talking about. They anticipate 20+ cores in less than 2 years.

The industry seems to be slowly coming around on this and now starting to recognize it as a serious potential reality.

BTW, you might want to get into the whole Verizon/Vonage thing. Some serious concerns there for millions of people and lots to discuss. Are Verizon's patents too broad? Can VoIP legally exist outside of providers like Verizon? Will Vonage have to turn off 3 million customers overnight?


We've seen this era of computing plenty coming for a couple of years. It's going to change everything.

As to Verizon/Vonage, they'll work something out or Congress will do it for them. They'd be a lot better off doing it themselves.

Peter Glaskowsky cautions:

As I've said before, massive multicore processing is not this easy. It isn't like Moore's Law, where more bits and more gigahertz alone could gave us higher effective performance. Multiprocessing works great for certain applications such as video and audio processing, web serving, and some kinds of scientific computing, but it doesn't help at all on other applications.

Most of the things you do with computers, for example-- word processing, Web browsing, strategy games, etc.-- will work better with two to four cores than they do with just one. But they won't work much better with 20 than they do with four. So for most users, the benefits of multiprocessing will level off quickly.

There are some important lessons here. In the future, PCs for different applications will be designed differently, so users will have to learn how to select the right PC for their intended uses. Most people will probably buy 2-core or 4-core PCs long after it becomes practical to build 16-core PCs.

The other key point: since adding more cores doesn't automatically make new PCs faster than old PCs, microprocessor designers will have to keep making the individual cores faster. We need more gigahertz, more instructions per clock, more cache, and so on. Continuing improvements in these areas are needed to keep selling new PCs.

. png

On the other hand, Linden Labs is adding multi-threading to make Second Life work faster for those with multiple cores. Of course adding 20 cores won't make things 20 times faster — but it does make it possible to do ten things at once if you have enough memory and other resources.

Subject: Open-source project governance: the Debian Linux case

Ian Murdock interviewed

Ian Murdock: Debian "missing a big opportunity"

Debian is in a lot of ways process run amok. ... Sometimes you have to make decisions that are unpopular; that's what leaders do. What ends up happening in this committee mentality is that no leader feels empowered to make decisions unless everyone agrees with him. And since no one as the size of the organization grows ever agrees on anything, no decisions ever get made. ... I think the fundamental mistake was this adoption of a democratic process ... I think in some ways the people who were really pushing for pure democracy at Debian wanted to see this as a sort of social experiment - what happens when every decision is put up to a vote. You know, pure democracy... It looks a lot better on paper than it ends up in practice.

Is this perhaps the appearance "as farce" of pure democracy in history? As the French Revolution was its appearance "as tragedy"?

Rod Montgomery == monty@sprintmail.com

The old description of the Polish government in the monarchy where every major aristocrat had a veto was "despoty tempered by anarchy" and later "anarchy tempered by assassination." Unanimous consent is a great idea until something has to get done.

Subject: Free Trade, Ricardo, and the Future of Employment

You said:
"They also put the lie to Karl Marx's dictum that capitalism would result in the concentration of wealth with a small number getting richer and richer while most everyone else fell further and further behind."

I don't remember having read that in Karl Marx works. But anyway, that was what has effectively happened. Less than 10% of the world population has more than 90% of the world wealth. And the concentration has been growing faster in the last century.

We can argue if this is a problem caused by capitalism or lack of it.

Globalization, with the emphasis in free trade and free movement of production factors (including jobs), tends to be better in distribution of wealth.

A qualified senior network admin like me, in Uruguay (the country where I live) was able to earn around $ 800 a month, until a few years ago. That was a ridiculous amount, in comparison to what the companies are paying for the same job in the USA. The fact that globalization and Internet have brought a lot of those positions here, has been a blessing for us. We're earning double what we were, and companies are making a profit. That's pure capitalism at work, and MOST OF THE WORLD IS REALLY HAPPY WITH THIS TREND.

I think that this is an irresistible force. Corporations can't afford not to go where labor and material costs are lower.

I predict that, in 50 years, maybe less, the world will be a very different place. Salaries and other prices of the economy will irresistibly tend to be the same, no matter you were born in Bangkok, Montevideo or New York.

Ah, and 40 hours a week will be a thing of the past, undoubtedly. My grandchildren sons will be asking how were we able to tolerate that. Maybe not 4 hours a week, but 20 would not surprise me.

Daniel Cardozo

I had not thought that it was the business of the people of the United States to make the world happy at the expense of the American people, but I certainly understand your view.

I fear you have not read Marx very carefully, but you are hardly alone in that.

Peter Glaskowsky adds,

Mr. Cardozo writes: "Less than 10% of the world population has more than 90% of the world wealth."

The situation was dramatically less egalitarian 100 years ago, and even worse 1,000 years ago. Capitalism created the means by which the poor could generate more income than was necessary to live on and _keep_ that surplus.

. png

Here is a tale of computer woes, Dr. Pournelle.

Washington Times article (link may age badly)

"Everyone who ever sat at a keyboard knows that moment of terror with the realization that a single accidental keystroke has wiped out hours of work. Pity the technician for the Alaska Department of Revenue who wiped out a disk drive containing information for an account worth $38 billion.

"Performing routine maintenance, the technician deleted applicant information for an oil-funded account -- one of Alaska residents' biggest perks -- and mistakenly reformatted the backup drive, as well.

"There was still hope until the department discovered its third line of defense, backup tapes, were unreadable."

Charles Brumbelow

I would think the moral of this story somewhat obvious...

A tip on system maintenance.

This is if you don't have a spare Windows system to attach the crashed hard drive. You need to carry a Knoppix Linux boot CD (or DVD) with you. I have saved much user data and system settings by booting to Knoppix. It can read and write NTFS, after you change the properties to allow write-access to the partition. I'm not sure if the Niven system lost data, but I didn't read that you took that step.

-peter cantwell

This works, and was the subject of a column a few years ago. There was never a moment when we had any fear of Larry Niven's system losing any data. The problem was in the motherboard.

It started with this message:

Subject: Win Update KB929338 kills my computer confirmed

Dr. Pournelle,

I inadvertently confirmed that the windows update critical patch kills my Athlon64 desktop.

After I restored my computer and before I told windows to ignore the bad update, I accidentally shut down my computer with the "install updates then reboot" option, and it again killed my computer.

The update in question is KB929338.


Which generated an answer from security expert Rick Hellewell:

Dr. Pournelle:

Haven't seen any problems with that one.

But, your reader needs to supply more information. ("Doctor it hurts, fix it!" "What hurts?" "Fix it!")

When posting questions, it's important to be specific. "...kills my computer" doesn't give me any information to know where to start looking. I'd need to know the exact problem you have. You need to tell me where it hurts, when it hurts, what you were doing when it hurt, and facts about your system.

Sample: "I just applied the patch, restarted the computer, and during startup got an error message saying ..." "My computer starts OK, but when I use the xyz program and do the abc function, the error message says ..."

Without more specific information, there is little I can offer, other than "Try a Google search of your symptoms".

Regards, Rick Hellewell

In general that's good advice. If you need help, it pays to be specific. I do remind readers that this isn't a help service, and we look into problems only if it looks as if the problem may be of general interest to readers. This isn't surliness, just a lack of time.

That got a new letter:

Need more information

Dr. Pournelle,

Here's what happens.

I installed windows update KB929338, and when it said to reboot, I allowed it to reboot. The usual bios text came up, the computer checked to see if there is a boot cd in the cdrom drive as usual. The screen went blank as usual as it starts to boot from the hard drive, the hard drive light blinked once, and then the computer froze with a blank screen. The windows boot splash screen never gets displayed. Not only is the computer frozen, but instead of the usual 5 seconds press on the power button to get it to shut down, pressing the power button makes the computer power down immediately.

That is unusual for an ATX motherboard and power supply. To make sure it wasn't simply thinking hard, I let it sit for 15 minutes like that with no change.

If I press F8 at bootup, I get the usual windows boot options for safe mode, safe mode with network, logged, etc. which tells me the initial startup bootloader works and can access the hard drive. None of these boot options result in the computer booting. The safe modes spontaneously reboot while the various modules are scrolling past, and none of the other semi-normal boot options ever get to the windows bootup splash screen.

I booted from a Linux CD, mounted the windows NTFS volume, and searched for the boot log file. I think the boot sequence in logged boot mode aborts before the boot log file is created because I was unable to find the boot log file.

On the Microsoft public forums, there seem to be a few others with a similar problem with this update. No solution has been posted other than some people manage to get it to boot in safe mode and remove the update, some cannot remove the update and have to do a repair installation, and some like me restore from a backup using a third party utility.

I inadvertently updated the computer again by clicking the "update then shut down" button when shutting down my computer instead of telling it to ignore the update, and the results were identical (non booting computer, 3 hour ghost restore from my external USB hard drive).

I do not suspect mobo/cpu/memory, hard drive, MBR, or other boot sector failures because linux runs fine off of a CD, and pressing F8 gets me to the boot menu so the computer is able to locate and run the initial windows bootloader. Plus windows runs just fine after restoring the previous ghost image.

One troubleshooting step I did not try was to unplug every peripheral (usb devices, second monitor, etc) or open the computer and start removing optional stuff like my soundcard and non-boot hard drives. Back in the win9x days this would be a standard troubleshooting step but with a working image on my backup hard drive, I did not do this.

My system specs:
Windows XP Home current with all critical updates and service packs until KB929338
MSI K8N Neo 2 Platinum motherboard, socket 939, Nforce 3 chipset
Athlon64 X2 4400 cpu
2 1GB sticks of PC3200 (DDR400) DDR memory (gskill brand)
AGP Nvidia GeForce 6800GT video card
Creative Audigy 2 ZS sound card
2 IDE hard drives (primary ide master and slave)
1 DVD recorder (secondary ide master)
1 SATA hard drive (sata port 3, also my boot drive)
Floppy drive

I hope this is enough information. After another hour browsing Microsoft's support site, I still could not locate any way to report this problem to Microsoft that did not involve giving them money. I haven't had this serious of a problem with windows for about a decade and don't visit the Microsoft support site very often, so maybe there is an easy way to report problems that I just don't see. The public "support" forums seem to be full of people with problems and very few solutions. My posts on the windows update and windows help/support forums are buried below page 13 with no responses more informative than "try safe mode and remove the update".

The update appears to address a problem I have had very intermittently after letting windows run for days without rebooting (windows stop errors) so I would have liked to see if this update would fix the very rare BSODs I get, but in this specific case the cure is worse than the disease.


And more replies. Begin with Eric Pobirs:

Very good detail. Actually, the behavior of the ATX power switch can change depending on the system state and that includes a simple power cutoff in the early stages of startup. This suggests whatever is going wrong is early in the boot process before the mode change kicks in. Beyond that, I'd say there needs to be a survey of affected systems to get some idea of whether this is due to a specific item in those systems or is a condition that any system might replicate under the right circumstance. In other words, luck of the draw.

Seeing as a manual install appears to bypass the failure it's likely one of those conditional things. The only thing notable in his system that wasn't in any of the half-dozen places where I've installed the update successfully is the Audigy board. I'd be surprised if that itself is the culprit but a piece of software tied to it or any number of popular items that may have been running in the background but weren't part of the test suite at Microsoft.

And the probable cause from Mr. Hellewell:

Perhaps this will help your reader (who did an excellent job of providing complete information about the problem).

With a Microsoft support forum search of the KB article: A message in there seems to indicate that perhaps the update is not being installed correctly by auto updates due to your OEM-versioned OS.