Letters about all sorts of things

Of course, you know this means WAR

Wendy,

We are starting an international effort to save the public domain. Using a combination of mailing list and wiki, we will try to organize the different forms of resistance. See attached article for our call to action. The effort is already endorsed on Larry Lessig's weblog: here.

Do you think you could promote this a little? I would also be very pleased if you chose to participate. The relevant URL is here, which also links to the wiki page.

Thanks,

Erik Moeller
http://www.humanist.de/erik
Editor of: http://www.violence.de

----->

I'm writing in response to your editorial at the Inquirer regarding the (ridiculous) Supreme Court decision in Eldred.

One of your suggestions is to get copyright holders to place into the public domain those works that are no longer selling. While I understand the basis for your suggestion, in the long term it's a very bad idea.

There are many who mistakenly believe that as long as a copyrighted work is making money it should not enter the pubic domain. Or to put it another way, only worthless works should enter the public domain. Your push to have out of print books, music, and films placed in the public domain only feeds into that erroneous thinking.

I think we should demand nothing less than ALL eligible works to be placed into the public domain. Imagine if your argument was used in the civil rights movement: "We don't want equal rights for all, just for those light skinned Negroes." That obviously was not the course to take, and neither is yours.

Steve Newlin
Email address supplied

Intel's thermal plans chill fan makers

Hi MM

Here is what I would to if I was a small heatsink/fan maker. I would tell Intel to blow it out their arse. Then I would take each new model and make sure every well read hardware testing web site got a free one. If it cools as well as I think it does the tests will all prove me correct and the product will sell.

Let's face it, most of these small specialty heatsink companies don't sign contracts to make 100,000 units for Dell. They sell to the overclockers and we don't even look at a list like AMD's Approved list. I buy a heatsink/fan by what I read on the hardware test sites I trust not by what the CPU maker tells me to buy.

Ken in Kanada
Email address supplied

Firm to develop solar charger for mobile phones

Nikkei.net reports of invention of a wheel.

"A REPORT ON NIKKEI.NET said that a Japanese company has invented a charger for mobile phones (cellphones) that uses a Siemens solar panel."

I tried Google:

http://www.google.com/search?q=nokia+1610+solar+charge&sourceid=opera&num=0&ie=utf-8&oe=utf-8 and in a link, not so far from the top:

http://press.nokia.com/PR/199703/775637_5.html

"The NonStop Nokia 1610 Plus Features the Unique Light Battery with Solar Panels - Nokia Introduces the Most Mobile Phone Under the Sun 03-20-1997

Nokia Mobile Phones today launched the Nokia 1610 plus, an improved version of the popular Nokia 1610 digital consumer phone. A stylish, high-quality yet affordable handset, the Nokia 1610 plus boasts the longest operation times available in its segment. The product features an innovative new power option, the slim and lightweight battery with solar panels. The phone will be available in Asia from April. "

It seems the Finland-originated company Nokia did this waay back in 1997, and IMHO better. It's nice that you reported this five years later, as a new innovation.

Jussi Heino
Email address supplied

[New innovation? Kill the word innovation we say. It means new... Ed.]

Where's the software on 64-bit X86 CPUs?

I just love to read the hyperbole laden Itanium missives from the marketing departments at Intel, HP, etc. You are right, it's a lot of smoke and not much fire and the things are soooo expensive that none of these companies will ever get an ROI. The Itanium is about to be supplanted by a vastly superior economic model. The economics will override any real (or more likely imagined) technical superiority.

AMD on the other hand, is a different story. If their backwards compatible Sledgehammer CPU (what were they thinking when they renamed it!) gets anywhere even remotely close to the performance level of Itanium, PPC64 or Alpha, the game will be over and Intel will be forced to compete with their own equivalent (Yamhill/Potomac?). Of course, Intel will do their own different X86-64 extensions, poisoning the well of AMD's architecture and forcing software developers. to choose. It will temporarily make a complete hash of the 64 bit commodity market. Yet, we will overcome, because we will have too. We will be forced to change by a huge shift in cost/benefit ratios. AMD will have kick started the transition and all of our development efforts will follow.

This is a common theme in disruptive technologies. In this case we are getting large jumps in performance with minimal or no cost premium. In some areas we will be getting improvement by a full order of magnitude! When such changes occur, it is inevitable that enterprising, creative people will take advantage of the improvements and develop products which get snapped up simply for their impressive economic benefit.

Put simply, the benefits of 64 bit are far too great to ignore and although the enthusiast market will be a rapid adopter (so users can frag each other ever more quickly) the business market is where the real action will be. We are devolving from a PC centric computing model to a thin client model where the PC, if used at all, is less and less it's own platform and more and more an extension of the networked client server environment. It makes no sense to have all these fat, brittle, maintenance intensive, virus culturing, power hungry PC's scattered about when powerful server centric computing provides so many cost benefits.

Using thin clients that are merely display and I/O devices is a great way for corporations to regain control over costs, security, stability, etc. and impose some corporate will on how their systems are used. Of course, ever more advanced applications will be further enabled by these systems. How advanced and what makes them so? The virtual elimination of non-OO technologies will be afforded, nay, even dictated by these systems. The advent of commodity 64 bit servers will be a watershed transition point for software, networking and systems.

So what has commodity x86-64 have to do with it? Let me count the ways

· Power and capacity. Raw MIPS and bandwidth. Huge memory capacity.
· The sheer computational power to support more users and the memory and addressing capacity to run hundrds, even thousands of huge, huge apps in enormous memory spaces.
· The ease of supporting existing 16 and 32 bit apps while simultaneously allowing for deployment of large memory model apps in a linear address space (which makes coding vastly easier). Complex data structures spanning gigabytes of memory wil become common.
· Greater support for object oriented applications which tend to place a premium on processor efficiency and memory subsystems. This allows for greater shift of the burden from application designer to the "machine" and allows developers to work with ever higher levels of abstraction with correspondingly higher productivity.
· Development of new applications that were simply too difficult or expensive previously. Data mining for the masses comes to mind. Ever more detailed and sophisticated virtual reality is another key application.

Of course, this will only work if these chips truly become commodity, i.e. if they (and the systems wrapped around them) have low costs relative to today's expensive servers. That is why many of use are waiting, waiting, waiting, upons pins and needles for the first fruits of AMD's effort to make themselves known. We want to know the performance figures, the costs, the configurations, etc. so we can start doing price/benefit and reliability estimates for ourselves and our clients.

We do not want to pay $2,000 to $4,000 per CPU (the way Intel has raped us) but we will put up with bare bones systems that cost that much, if they have sufficient price/performance increases to justify the paradigm shift. We want to have the choice between a lowest possible cost compute server and a rugged, reliable, redundantly supported application server or dbms server. We need to see the spread from motherboard makers such as Tyan to the systems companies such as Newisys so we can make wise economic choices to meet our unique needs.

We want to see improvements in the underlying OS's, primarily Linux and BSD so that our new efforts can avoid the huge trap of development and deployment within the ever rapacious Microsoft OS and yawning .Net trap. You can bet there will be rapid advances in this area with the HT linked AMD cpu's needing a some serious tuneup work in the scheduler and thread affinity management area (I proffered some questions on this issue a few months ago at Van's site just before it folded - the link is still active). The kernel hackers are just itching to make these systems sizzle.

What will we get for our efforts? Lots of benefits flow from commoditization of such power:

· Servers that handle enormous workloads with ease. When every business has a $5,000-$10,000 server with anywhere from 4GB to 16GB of memory, in-memory DBMS will become the norm and this will allow incredibly powerful OO-RDBMS to flourish. Currently these things are performance pigs at any reasonable size but that will quickly change with larger memory models. Eliminating performance robbing disk I/O will provide for several orders of magnitude performance improvement in these "natural" OO DBMSs.
· The above will help enormously in reducing or eliminating the "impedance mismatch" between OO applications and the currently common and "affordable" RDBMS. The amount of effort taken to map between high level and intuitive object oriented abstractions and their underlying storage representation will be dramatically reduced.

· Given a shift away from traditional relational DBMS towards OO-RDBMS, we will now be able to develop OO style collaborative and/or federated applications. These kinds of apps will be their own stateful entities but with the added OO style advantage of exposed interfaces, i.e. they will be able to coordinate their knowledge and talk to each other. Current system are simply too limited in performance to do this on any commercial scale but with an order of magnitude performance increase they become practical, even a necessity. Think EJB's on steroids.

· With lots of server power on tap the thin client can take off. Current servers can easily support anywhere from dozens of thin client users to nearly a hundred per system (I know they can because we build and maintain such configurations for our clients). Now replace those servers with units which provide nearly an order of magnitude of performance increase and you get $10,000 servers which support hundreds of users running thousands of apps.

· Alternatively, keep the current user count from the above example but deploy more sophisticated apps and gain higher productivity. OO is the key to it all and OO takes CPU snort and memory to run properly. See the notion of collaborative apps above. They eat a lot of MIPS and have lots of memory manipulations due to their highly abstracted OO architecture but they are worth it.

· Even the home market will gain. As the technology matures and power consumption is reduced, the central server in the home will always be on and serving the myriad of thin, application specific devices scattered around the home. The TV will be connected (Sony's vision more or less) as will every room (including the bath). The kid's bedroom and study room stations are just more clients as are the touch displays in the kitchen cabinets, the workstation node in the garage and the Wi-Fi linked vehicles. These clients can get skinny and cheap and low power because they don't need any moving parts since their entire computational load is handled by that server that has MIPS to spare (which is a monster performer by today's standards).

Well, of course there's a LOT more but your other respondents will fill you in better than my fevered imagination can. Just remember that such a transition is an eventuality. It cannot be stopped (but perhaps altered a bit - see Intel above) or even delayed very long. The transition is inevitable, as will be the subsequent transition to either 128 bit or 256 bit computing (or some completely new paradigm such as quantum or dataflow ...).

The economics of this type of change are simply too great to be stopped. We want that power, we need that power in order to advance. It is the computational power, the capacity to change the way we do things, the sheer economic benefit that will force this change, much to the chagrin of Intel and HP with Itanium.

Let the tidal wave of change come a crashing!

Spencer M Kittelson
Email address supplied

Maxtor readies Serial ATA drives

Hi Mike,

I was curious if you had any information regarding when manufacturers plan to introduce Serial ATA CD-RWs and DVD drives. There are loads of motherboards and controller cards available for SATA, and I've read that hard drive makers have been talking about releasing (although they don't seem to be in any hurry) SATA drives sometime this year. However, searching for any mention of SATA at the Yamaha Multimedia, Liteon IT, and Pioneer Electronics sites yield absolutely zero results.

This is particularly frustrating as it implies that we will probably have to wait a great deal longer for a SATA-only system. With the finalizing of the SATA specification on August 29, 2001, I would have expected that it would be six months at most until we began to see drives based on the new spec. Now, nearly SEVENTEEN months later, I'm still waiting to see any sort of SATA drive on the shelves.

The economy isn't the only reason that computers are selling poorly these days.

Regards,

John Lahtinen PS. Please feel free to print any/all of this email and include my name. Thanks!

Intel's Yamhill still lives, swimming in the Potomac

Mr. Magee,

We (former Yosemite guys) are having a good laugh over your Yosemite reference here (at Intel Santa Clara) this morning. The Yosemite project got cancelled in Q2'00! Seems that HP folks in Oz land are really behind the times. Over two years behind even after accounting for the June '02 date on the slide set.

Regarding 64b addressing, follow the socket address pin counts and you'll be able to get these 'Yamhill' rumors more in line with reality.

Dave
Full name and email supplied

HP gets knickers in twist over SAP benchmarks

Hang on a minute! You say "Compaq unfortunately forced Microsoft to

stop developing Windows for Alpha some years ago." That isn't exactly what happened and anyone involved should know better.

Microsoft owned the code but a large part of the development was funded by Compaq and performed by Compaq (DEC) engineers. So when Compaq cut the engineers and funding, Microsoft quit developing the OS. And Microsoft got to keep all the 64 bit code. (What a deal! nearly free code)

I still don't know where you get that Compaq _forced_ Microsoft to stop development. Just sounds like flame bait to me.

Luke Fowler
Email address supplied

[See this and Compaq responds to Register Alpha NT stories Ed.]

------>

If I'm not mistaken, the Windows "NT" kernel guys continue to develop on Alpha boxes because it's a good validation platform for them to keep the kernel clean for multi-platform use, if they should ever need to do multi-platform again (and XP embedded sounds like such a beast).

I don't remember where I read it, but chances are good it was The Inq or El Reg.

I believe the "NT" kernel guys continue to do limited Alpha development specifically for this purpose. Hammer is not really a different architecture, being just an extension, and Itanic is kind of "half-different" since it still shares a lot of design paradigms that Intel likes to use, including the PeeCee platform.

Name, email address supplied

----->

No processor has made a dent in x86. None at all. Many many have failed.

Power PC failed, Alpha failed, Motorola failed several times, Apple failed as did TI and others.

Ask yourself why. I will tell you why.

Companies hate incompatibility. They got burned a lot in the early PC era. Platforms changed many times and each time they had to buy all new equipment. Finally x86 caught on and now they havn't had to throw away equipment and software. They are not about to ride that pony again.

Not if they have any choice. Intel thought they could force them onto a new platform with Itanic. Well even Intel can not do this thanks to AMD.

Companies will jump on the Hammer bandwagon for one simple reason. They don't have to throw anything away. Everything will still work. They will have a smooth migration to 64 bits and can fall back to 32 bits any time they want to. Even if they discover a year from now the new 64 bit software is flawed they can eaisly fall back to the proven 32 bit applications..

Many CEO's and CIO's have lost their jobs after getting their companies to jump onto a new platform. These new platforms often failed. There was no falback position and the companies were royally screwed. IU have worked at many that invested millions upon millions trying to convert from mainframes to Networks. They failed miserably until the networks finally became mature enough around 1998. They all understand that Itanic is a no fall back position and few will be willing to accept that.

American Power Conversion recalls two UPSs

Reading the INQUIRER saved my life

Dear Mike & Co.

Usually, I read the Inquirer for it's snappy patter, and to partake of the rumour mill.

But today reading the Inquirer might well have saved my life!!

You had a story on the recall by APC of some of their UPC models, for a potential fire-hazard.

Your story about the APC recall made me, after some ho-hum procrastinating, get out of my chair to check my UPS unit to see whether it was one of the models affected.

Guess what, it was!

If I had not read the Inquirer, who knows when I would have gotten any notice of the recall?

So, now you can say that the Inquirer is a life-saver!

It's required reading, ... making the world a safer place for geeks everywhere.

Many thanks, and best of luck

Al Urquidez
Email address supplied

------>

Dear Inq,

I'm lucky enough to have not one, but both of the models of UPSs listed.

I'm using a CS 500, and the CS 350 I had previously is at my parents. Both are probably in use >60 hours a week if only to power speakers. Both were purchased at a popular US retailer. The CS 500 is in the recall and I"ll find out this afternoon about the other. Let's hope they at least will just ship a new one flat out, so I don't have to go through the hassle of UPSing [funny, ha ha :-P] the other one (two?) back.

Thanks for the heads up,

Paul
Email address supplied

------>

Thanks for the article on APC's unit recall. I would not have known otherwise.

Regards,

Allan Robbins
Email address supplied