Has the whole Unix world gone stark staring bonkers?

Is it just me, or this all just slightly mad ?

If you had said to me 5 years ago that the bastions of proprietary systems Novell and IBM were the good guys in a fight against a Unix vendor like SCO I would have laughed in your face. To me this opens an interesting question. It is entirely possible that SCO is correct in saying some of its code migrated into open source products. Surely it is just as likely that it migrated back ?

I have included a piece of TheInq website in this letter., can you spot it ? Even in plain text from a known source ? Fancy doing it with 300,000 lines of C++ against a shared code base in the millions ?

A someone who used to make money doing source reviews for both IBM and Microsoft and the first Microsoft Intel Xenix in the 1980s. (for very good money he says modestly,) I can state with authority that there is no possibility of reliably catching this. All developers get fragments from their peers. To integrate it into code means changes to variable names etc that mask its ancestry, not through dishonesty, but merely to make it fit in. So it just depends on the risk that a company is prepared to take and how much effort it wants to put into software testing.

As you are no doubt aware, nearly all operating system code is C and C++, so fragments of open source may well be in Windows and the proprietary Unixes.

Given that much will be under GPL or equivalent, I predict that very soon there will be retaliation. In this, SCO is much more vulnerable. If you destroy a distie like (say) Red Hat, then this is sad, but open source will survive, since it can easily find a way to hardware. On the other hand SCO is very vulnerable to Anton Pillar and equivalent orders.

Dominic Connor
Email address compris

Nvidia did well, not badly
Nvidia cries foul as hand caught in cookie jar

I sense bias. I also sense that the author decides to attack the people supporting nVidia's actions instead of actually attacking the specifics of the facts. No where did I read that the changes made were so drammatic as to drastically alter the image in 3dmark. No facts were presented in this editorial about how drammatically different they were. If it took the developer edition and examining the image from multiple angles to see the difference then I think someone's pulling gnat-shit out of pepper.

I have to hand it to nVidia for taking the initiative in manipulating their driver so that it could find these inefficiencies (at the slight expense of image exactness). I know you might just say I don't know what I'm talking about. But frankly I haven't read that anyone has stated that they know what they are talking about and have presented it cogently with enough detail to sway my thinking.

If you are going to present facts you better damn well present them in detail. To conjure hatred because you don't agree with them is silly. And since you consider this cheating, how much cheating is too much cheating? If ATI was cheating (only a little) and nVidia was cheating a bunch more, it is still cheating.

If 3dmark puts in such generic routines to render and test score such hardware, obviously with ideologies from generations ago, and the manufacturers of the drivers and cards conclude that generic routines which are not optimized for their cards is wrong, then I put more stock in the video card manufacturers way of doing things then in 3dmarks'. Had nVidia or ATI skipped portions of the tests, had they inserted copies of the images in place of the rendered ones, had they vastly subjugated the test results, I would say they are cheating. But to optimize a series of routines at the driver level where they are able to generate essentially the same image, not noticeable by most human eyes (which is the case with graphics cards, no?) then I'd say that bitching about it makes you loose faith in my eyes.

Let's get real. Start attacking the other portions of the computing industry, not just the big guys going head to head, about how they have for years exchanged quality for quantity or scheduling.

Jim Blaich
Email address supplied

Journalists suck, not Bluetush
Basically, Bluetooth sucks

dear mike,

i must slightly disagree on your article "Basically, Bluetooth sucks", some time ago, there was a small workshop at my working place for IT and smal-to-medium-size companies to encourage the uptake of IT in smaller companies. i guess there were something like 40-50 people around carrying stuff like laptops, PDA's and mobiles.

wondering how many people actually have bluetooth switched on, i searched with my mobile and found 7 devices (three mobiles, three PDA's and one laptop). not too bad for only 50 IT oriented people in a small workshop in belgium, compared to the 1000 geeks at IDF...

so i think bluetooth is getting quite widespread and reasonable percentage of users people actually have it switched on! whether they actually use it (i don't realy see any 'killer' application), or just haven't found out how to switch it off (bluetooth and userfriendliness don't always go in hand) is another matter.

not to mention bluetooth's rather battery-draining effect... so i think Bluetooth doesn't suck (except maybe for usefullness, userfriendliness and energy efficiency ;) ).

best regards,

dimitri
Email address supplied

Each time you can't find anyone in one of those trade shows, there could be plenty of people gaming with each other, filing copy via dialup or GPRS, web-browsing on PDAs, chatting on their handsfree earpieces and suchlike.

That's because bluetooth has two modes of behaviour. Discoverable, when you need to introduce devices together that haven't seen each other before, or bonded, where two devices know of each other and are happy to chat to each other.

So unless someone was setting something up right as you scanned for discoverable devices, of course you won't see many, if any. That's the way it works, for goodness' sake.

I remember the times in the past when you knew what the hell you wrote about. Crap like this doesn't do you any favours.

Alan Fleming
Email address supplied

MicroAnvika at the bottom of Tottenham Court Road is a good place to try... I picked up 1 other phone and several laptops all live.

Richard
Email address supplied

Hmm, I have a perfectly good Bluetooth car handsfree, since handsfree is the law in the Empire State; it mutes the stereo and pauses the CD like a normal handsfree. My Bluetooth headset also works great, even though its brand (Motorola) differs from my cellphone's brand (Ericsson). Both are voice-dial compatible, though Ericsson's voice dial in handset leaves a lot to be desired.

Additionally, Apple's really sewn up Bluetooth good. Even with a dongle for my "legacy" titanium powerbook, it is invaluable for syncing all my phone numbers to my cell (T68i), syncing my addressbook so I don't have to bother with a sweaty Palm, and even for little things like controlling iTunes (and even the mouse pointer!) with Sony Ericsson Clicker.

For the newer Apple stuff with integrated BT, Microsoft's bluetooth mouse & keyboard are, yes, compatible. So, if you want a wireless keyboard & mouse without dongles (particularly nice on a laptop), bluetooth is the way. At least on IBM and newer Apple boxes AFAIK: I do not recognize any junk that isn't IBM or Apple (or, in a pinch, Sony).

Granted, Nokia has crap for Bluetooth: on the limited selection of phones that implements it, they carefully avoid fully supporting standard profiles so you have to buy Nokia bluetooth accessories. Simple solution: avoid Nokia.

C'mon dude, you can do better than THAT.. Being a bluetooth playa hata is SO 8 months ago...

Cheers
Bluetooth Boy
Email address supplied

In my experience, Bluetooth sucks because it's making my hair go grey and fall out. I have so many problems actually getting the thing to work properly, ie with my PC and Ericsson T39m. Most of the time, a cable would probably be far simpler, easier, faster and more reliable.

Mathew Wilson
Email address supplied

Hi Mike,

As I see it, Bluetooth today is a very good replacement for wires. I used a "Fujitsu-Siemens Celsius H Mobile" laptop last year. This series of laptops have a very thin (ca. 0,5 cm.) removable keyboard, using Bluetooth. It's a very large and heavy workstation-laptop, so the hadny keyboard is a very good thing during presentations, and when using the laptop in meetings. This keyboard also have a small "mouse-pad", so I could control everything from this.

The people I borrowed the laptop from, didn't know about the removable keyboard, so when I removed it they were kind of shocked. But after I had showed them how to "drive" a preseantation from this keyboard, they were extremely pleased.

A wi-fi link would problably not be a good idea for this use. The range is too long and so is uses to much energy. And other kinds of wireless standards seems to miss the security needed. (I will never use a wireless keyboard unless the uses Bluetooth.) Wi-FI simply cannot do this. (How can you log the keyboard into the PC?)

Bluetooth on phones on the other hand is a dud so far. But that is not a new thing in telecom. WAP and 3G seems to be going down the drain too.:) But I am looking forward to getting a GPRS phone one day, since I like "always-on" connections.

Vegar Kleppe
Email address supplied

Vole always recommended separate Firewalls
Microsoft pulls security patch for Windows XP

From the get-go, Microsoft has always said that their OS firewall is meant only to provide a rudimentary level of protection. I detected a hint of sarcasm in your article, which is very characteristic of the OSC. Thing is Microsoft has ALWAYS recommended a separate firewall approach for those people serious about securing their network.

Big hole, worth noting… but the way you made it sound was that this firewall was supposed to be fully functional - when admittedly Microsoft said it was not.

-- Ryan C. Lambert
Email address supplied

AMD no Milli Vanilli no more
Benchmarking: Is the Pentium 4 unassailable?

I enjoyed article for perspective. It was, however, inaccurate to say that the K6 was a copycat CPU. Not since the days of the AMD 486 has any AMD CPU been an Intel knock-off. The K5 and K6 were independent, original designs that plugged into the socket 7 infrastructure.

If you remember, the K5 was late to market on account of the independent development cycle. It got stuck in the pipe too long. The K6 design, ironically, did not originate with AMD or Intel. AMD bought the design and the company that designed it because it was so good and could be brought to market quickly. The 486's, however, were fair game for the claim that AMD was "the Milli Vanilli of semiconductor manufacturers." Thank you Dr. Grove.

Mark Ellsworth
Email address supplied

Descartes, like Cabell, is spinning in his grave
Delayed write error is ghost in the machine

Mike,

Thanks for bringing this issue to light. However, all my systems which have this problem experience this when attempting to save a file to a server.

Just as others noticed though, the issue is sporadic - the system will have been working for a good while, then start to exhibit this phenomenon - then subside for a while.

Dropping the UDMA BIOS level to '0' eliminates the problem for me however. The other potential cure - disabling disk drives cache - doesn't.

Cheers,

Steve Hopper
University of California, San Diego
Email address supplied

Tweedledee, Tweedledum... Alice, Alice, WTF is Alice?
Benchmarking: Is the Pentium 4 unassailable?

Dear Mike, Jack,

Thanks for the interesting article that tries to portray the competitive situation in x86 cpus from a new perspective. You point out that "There is a significant number of programming areas where SSE2 cannot boost performance very much"

In fact, I would like to turn this phrase around and say: There is only a very limited number of programming areas where SSE2 can be utilised to speed up an operation (mostly audio/video).

So the whole issue of the rise and rise of benchmarking results for the P4 are not because all general purpose software is now SSE2 optimised - most of it cannot be.

That only leaves us with the "leaning on benchmark programmers" issue. As you will recall from the rumpus about BapCo, not only were the weights of sections changed (i.e. Video encoding now more important than for example gaming) but even within these sections certain (SSE2-friendly and bandwidth.intensive) tasks were raised while others were disregarded.

As a computational physicist in a university environment, where a non-uniform network of PCs is used for computations, I was sorely disappointed by the introduction of the P4.

It has been known from its very introduction but seems to be mentioned less often these days, so let me say it once more

*** The general floating point unit (x87) of a P4 is only about half as fast as that of its predecessor P3! ***

Intel has taken a general problem-solving machine (the cpu-in-a-computer) and turned it into a not-so general one (for Video and Audio but not much else).

In this situation, we in physics are very happy to have the opportunity to choose the AMD platform for our computations because we cannot afford to

a) waste 50% of our computational time in x87 tasks or
b) recompile all code just for a few newly-purchased P4s just in case it turns out 5% of the problem can be SSE optimised.

...and for all who think having fast Video and Audio is good enough, think again - right now, Video and Audio compression algos are based on FFT, usually on small arrays (blocks), which current versions of SSE are geared towards.

With the upcoming crop of wavelet - based compression algos (vis JPEG2000 and according motion techniques), this optimisation will break down. So then we're back to using x87.

These algos deliver far superior quality (no more blocky artifacts in your action scenes, only a slight blurring maybe if you turn down the bandwidth), so you'll want to have them.

But I guess the introduction of these long overdue compression algos will not exactly be assisted by Intel. Neither by AMD, as soon as they have SSE2 bragging rights as well.

Which brings us back to the question why monopolies - or oligopolies - stifle innovation and are bad for their customers.

Regards,

Andreas
Email address supplied

Only some comments and arguments to your article, you might consider:

i) It is NOT possible to patent instructions (At least in the US) . This is very important to know and to understand, because it has a great impact on the computer industry.

ii) All you can do, and people do it, is to patent some mehacnsims youz need for a certain instruction

iii) Many computer instruction specifics _have_ been published and patented in the 50ies, 60ies and 70ies.

iv) Most patents in the area of MMX are held by Hewlett Packard, specifically by Ruby Lee (see her Princeton homepage for more details)

v) IBM makes 2-3 times more (!) patents in the computzer field than Intel. Of course it holds most historic patents.

vi) AMD files more patents than Intel and has patent exchange agreements with other companies like IBM.

What this means is, that if Intel refuses AMD to license a certain patent in one field, AMD could still use it because of some agreements with IBM, or even worse, AMD could refuse a patent to Intel in some other field that hurts.

vii) This is the reason why companies are afraid of companies that sell only _patents_ NOT products (RAMBUS)....

SONIC Blues Ex-CTO Andy Wolfe refered these companies as "rogue states"

viii) I know it is commonplace to bash Intel technology, mostly because of its "stupid" marketing, but behind that marketing is a formidable engineering team, and a management with a strong commitment to (costly) engineering (something that we in germany / europe have lost). I have visited several computer architecture conferences, and have met plenty of (brilliant) Intel engineers. I met a few engineers from other copmpanies, such as SUN, DEC, ARM, IBM,... I never met anybody from AMD.

You will not find any publication in the area of computer architecture from AMD.

regards,

Name, email address supplied

OpenVMS on Itanic not all there
OpenVMS safe in HP's hands, says Shannon

Hello Mike,

Shannon spread the meaning that OpenVMS is one of the best OS and will be supported on the Itanium platform. This is undoubted.

But OpenVMS for Itanium will not be a full version of the OS. The port is far away from the current version and the reliability, scalability and the support of applications. It will be a VMS 1.0 version for Itanium and not a Version 7 as for Alpha. Many layered products will be cancelled and many developers left the ship hiring to other OS development projects. New projects and developer are not interested on OpenVMS, have a look onto the sales documents.

The new HP is busy with laying off the good old DEC crews. So OpenVMS engineers are rare in many countries. The support will be organized by an "competence center" of the last engineers within all countries. A "competence center" is always the last resort and step and not the kind of support of a living products like HPUX or Windows.

The management at HP doesn't care about Compaq products except the x86 computers. The goal is to be a new "Dell" like company, migrating all the unloved architectures like Alpha to the Itanium base. OS versions doesn't realy count. Have a look to Tru64 which is cancelled and customers must migrate to HPUX coming under the name "Enterprise unix". So the roadmap is to migrate the Alpha customers to Itanium and to minimize the development, costs and manpower to only one or two operating systems, preferrable Windows. Windows will be the final step as announced several times by Carly Fiorina on roadshows and briefings.

I'm wondering why OpenVMS and Alpha will not be transferred to a spin off to a the good old DEC-like company (maybe with the same name). The Itanium is far away to be the "Standard" CPU for 64-Bit applications. It is not cheap, not reliable and full of bugs and design faults. Opteron is the 64-Bit architecture and server plattform of the future within the x86 market, kicking Itanium from the map. The goal from HP/Intel to establish a low-cost cpu failed and costs are far away from PaRisc and Alpha cost-savings.

Jose Fernandez
Email address supplied

Name that Toon
Real Networks to compete with Apple iTunes

Not only is your title for your article misleading, but so is the small article. It is NOT 79 cents per song. It is 9.99 dollars monthly for the subscription and THEN .79 per song to be BURNED to a CD. Which means you have to use Real Networks proprietary CD burning software, which allows them to stream the song to your computer and then their software burns it to your CD in your CD-RW drive. The point: Apple's music store is still cheaper up to 50 songs each month, because you have to pay the regular subscription fee of 9.99 dollars a month for Rhapsody. So figure this: I buy no music with Apple's Music store one month and none with Rhapsody; I still pay 9.99 dollars that month for Rhapsody but zero dollars to Apple. I buy one song one month to both services; I pay 10.74 to Rhapsody and only .99 to Apple. This service still doesn't compete with Apple, it is still all subscription based and you can't even own your songs. Stop paying one month and IF you didn't pay per song to burn the music to a CD and you lose it all. The point of this email is to make sure you staff writers for your online news services clear your information up and quickly.

Please forward to whomever needs to read this for future articles.

Greye
Email address supplied