I run a lot of old SPARCs, mostly SPARCStation 5's. I run primarily Solaris and OpenBSD on them, depending on purpose... However, I ran two SPARCs on Linux for about 6 months as production servers. (shell, mail, dns, web, ftp, some sql)
At the time, three main choices existed: RedHat, SuSE, and Debian. Redhat installed rather easily, but it completely sucked, and a lot of things were just broken. In fact, the machines locked after less than two days of testing. Lather, rinse, repeat - same results for 2 weeks.
I tried for like 12 hours to get SuSE to install, and I just couldn't hack it. I'm _very_ familiar with both Sun hardware and Linux, so I was lead to believe that it was just bad packaging. I gave up on SuSE.
I installed Debian... The first two installs weren't quite right (user error!)... but after I got things in order, I installed Debian on these machines, put them through testing hell for a week, and put 'em up. They ran pretty well for being 110mhz:) The Linux kernel itself isn't very mature for SPARCs, and its process creation time is AWFUL. These machines managed to handle about 200 users for 6 months, before the load just got too high. However, kernel-level resource problems required that I rebooted the machines about once every 60 days.
As far as SPARCs go, Solaris has very little competition. NetBSD and OpenBSD both run a bit faster, but file I/O is noticeably slower. However, *BSD is insanely stable on SPARCs (I've got a server with over 280 days uptime on SPARC/OpenBSD). Linux doesn't really hack it yet. Perhaps I'll get bored in the next few months and try out Slackware's SPARC distro... If anyone can do it right, it's Slackware:)
RedHat is FAR too close to Microsoft, in happy-friendly GUI ways, basic utilities crash constantly ways, and ebrace-and-extend ways.
1. Happy-friendly ways. chkconfig, linuxconf, rpm. The ideas are nice. they're very nice ideas. However, they're botched in a number of ways i'll soon get to. KDE, GNOME, whatever is available in whatever default installation of RedHat is nice and quite useable. Just like Microsoft. That is certainly admirable. Doesn't make it good, though.
2. Basic utilities crash. Have any of you done an RH 6.2 FTP install? Hell, even a custom CD install can easily pull a segfault out of RedHat. netconfig on 5.x and 6.1 will segfault if it doesn't get specific PTR lookup responses. Even worse than that, it'll segfault in the middle, so you don't know if it's changed your sysconfig settings, your live IP address/subnet/default route, or what... and if they're out of sync, don't expect netconfig to work once more to fix it. It'll just segfault some more and leave you hanging unless you can vi/etc/sysconfig/network/*, ifconfig, and route.
3. Embrace and extend. RedHat has positioned itself to be the high-market-share distro. However, RedHat intentionally releases broken standards (RH 7's egcs for one?) and moves things around in such a way that if a software developer writes a program to be installed on a RedHat Linux system, it won't install on any non-RedHat-based distribution. If it does install, the crazy egcs release will keep it from running on the new machine. RedHat often screws with things like init scripts just enough to make them UNIX-like, but to break POSIX standards. What pisses me off about RedHat is how deliberate their embrace-and-extend design policies are.
I don't recommend RedHat for anything, because learning the quirks of RedHat puts users into bad practices of using their proprietary tools, or expecting the proprietary behavior of their tools to be standard cross-platform. It's sort of like how a lot of linux distros have a 'route' command that, for some reason, won't accept 'route add -net default' (which is standard across UNIX) but will accept 'route add default gw'... annoying.
And why do users use SysV-style Linux distros, and still use ps -ax? why not ps -ef?:)
The government has turned down the prospects of creating a counter-worm, but any decently-experienced assembly programmer with sockets experience could just disassemble the current worm, make a number of changes, and release the worm on a time-skewed box. A really crafty assembly programmer could even keep the binary size of the worm the same, so in case the worm has some self-check mechanism, it won't notice any difference. I personally wouldn't mind seeing this fire fought with fire - Let the anti-worm run its course for a month, and then have it destroy itself. That would wipe out the vast majority of the code-red virus.
It is a really rash and dangerous tactic, but considering the scenario that a number of people are expecting from this worm, are there really any other effective options?
No kidding! The FF-style storylines are normally very complex and very deep... In a game, you've got 20-60 hours of gameplay in which you get to mill over and decipher the storyline, so it makes a bit more sense. Sadly, they had to try to cram an FF-quality storyline into a 2-hour movie, and I think that was its greatest weakness.... time.
The storyline, of course, borrowed from a bunch of story ideas in the past (Gaya vs. FF7's lifestream?, phantoms vs. Jenova?), and kept with the fairly common mystical/religious undertones that help make the FF series what it is. For someone like Katz (a person who hasn't played any of the games, and is, well, as silly as he is), the story line is either going to seem overly complicated (and thus underexplained) or extremely hokey and childish.
Every FF story is a challenge to read. It is also a challenge to the reader, regarding how one lives their life. I know I'm probably reading too much into this, but the point is that the FF storylines aren't supposed to be just entertaining like Angelina Jolie's tits in Tomb Raider... the storylines of FF games (and movie) make you think. a lot.
Precisely it: Jack of all trades, master of none. However, I am thoroughly aware of what I don't know.
I'm no DB genius. I'm hardly even decent at it. I was just given that responsibility because I could get it done without constantly trashing junking and re-doing the databases and/or servers. That doesn't make me special - it just makes me, 'essentially... the DBA...'.
Hell, i'm not excellent at any of the specialist tasks that I do. I may be able to write mediocre code in a few languages, but that doesn't make me the prime candidate for a project lead. And I know that. And I never even hinted that I should be considered for such things.
In case you didn't read too well, I didn't make any 'wild statements to boost my own ego'. I stated what it is I do. I didn't even say that I do it well. I just said that I do it. And I, too, let my record speak for itself. Do you think a college-dropout geek is going to get a decent job without a decent (though short) track record of accomplishments? I sure don't.
Please, don't confuse acknowledging that I know something (or even a good deal about something) with knowing everything.
I'm the stereotypical candidate for prima donna syndrome: a few days shy of 21, dropped out of the engineering program at a state University because it was unchallenging and mediocre on its very best days, and dove into the IT field. I'm a Unix Sysadmin for a little company with scrambling and confused management - a glorified dot-com.
Since it's a small company, I'm essentially also the DBA, network admin, Cisco guru, neurotic PERL geek, and so on. I get frustrated quickly with people. I word my sentences carefully to provide the most clear and concise meaning possible (management calls these 'very curt' responses!), and attempt to usher the questioning coworker out of my line of attention as quickly as possible. I tell programmers that their idea on implementing this or that is "like an ostrich - it's got wings, but there's no way in hell it's gonna fly." I'm a young, cocky asshole.
But WHY? It seems that no one has asked WHY we prima donna types are this way. My explaination is that I'm a die-hard perfectionist. I'm very interested in the architecture of things... both concepts and actual structures. I'm big on using available standards, or creating thoroughly documented standards if necessary. I'm big on harmony. I don't like solutions that plug one hole in a leaking boat, just so the water can come in through another large crack. I'm a die-hard perfectionist. Though I'm more than willing to throw really bastardized hacks into place when they do not create new problems.
If a programmer who specializes in socket programming comes to me with an idea on how to do task X, and I can think of multiple more efficient and more effective ways to do task X (NOTE: I am NOT a socket programmer, nor a specialist at socket programming), I will point the weaknesses that I see in their idea and offer the ideas that came to mind. I'm always constructive and ALWAYS offer alternative solutions, though my thoroughly-learned tendency to be concise with my words sets them on the wrong foot. Half an hour later, my supervisor calls me in and asks me about the 'incident'. He's actually quite understanding and open-minded. I love explaining my reasoning to him - he remembers it and often uses it productively. I explain my reasoning, he's happy, and I'm back to work. However, upper management (2 people, it's a small company) slowly builds an image of me being unfriendly and not helpful. Bad situation for me.
Management looks at things in terms of investment, risk, and a few other things that I'm not overly attentive of. Technical people often look at things in terms of efficiency and merit of design. However, only a small percentage of techies I know also disassemble ideas and concepts into security and liability to their company. Well, then again, most techies probably work in an environment where the management (at least) has liability already covered before ideas/problems/customers get down to them.
The merits of design are not the merits of finance and profit. The two sides oftentimes dislike thinking about the other's point of view, or are unfamiliar/afraid of it to some degree.
The bottom line: I am a prima donna because my point of view of any given situation is very different than management's point of view. I am not excessively willing to look at their point of view, and likewise they are not excessively willing to look at mine. I accept that and try my best to work with them on sharing our points of view. However, a 100% technically-oriented company cannot survive with a 0% technically-oriented management running the show. The components that make the company work aren't going to be properly filled in the right proportions. There's only so far I can stretch to make such things work... after that point, I'm called a prima donna and management holds their noses high. That's fine by me - my self esteem isn't hurt by other peoples' opinions (that i consider misconceptions)... that sort of behavior would not allow me to function well in the technical manner that serves my employer.
It's a problem of point of view causing frustration.
My first thought was that most people living in the projects of your nearby large-city wouldn't have such snazzy GPS systems. Thus, everytime you wanted to take the interstate into downtown, the computer/cellphone/whatever would beep and tell you to take Martin Luther King Blvd under the bridge because it'd think there were fewer cars down that way.
That reminds me about the wonderful directions MapQuest once gave me that took me to the middle of the projects in Atlanta, when I wanted to get to Emory University. Trying to get out of the projects in an unfamiliar city at 4am with 3 geeks in a pretty green saturn isn't exactly my idea of fun (well, it's funny now, but wasn't while it was happening!:)
Such communal reinforcement of the underdog is what makes a free-enterprise democracy stand out from the crowd. The US market (where these companies are based, located, run from, and count their revenue quarterly) is a self-proclaimed free enterprise, but is so politely tuned, chopped into portions, and regulated with as many bad as there are good laws that any G.E.D. graduate can see the US market is as much socialist as it is democratic.
Supporting an underdog is the necessary social response in order to maintain an open, competitive market. There is no doubt that Intel has a very heavy hand in the commodity CPU industry, and the acquisition of Alpha technology by one of Intel's competitors would help level the playing field. This sort of competition would benefit the customers, and the market, much more than Intel gobbling this up just to ensure that no one else can give them a scare in their niche market.
Remember that when DEC sold off their assets, they sold a LOT of microprocessor technology to Intel, including the StrongARM, their chipsets, their embedded controllers, and their NICs. They certainly had enough of a relationship with Intel that they could have worked a deal on the Alpha technology if they wanted to. Instead, DEC sold the Alpha processor to Compaq, under the hope that they would make it grow and keep the very high-quality processor in the market.
It certainly looks like Compaq has been slowly killing the Alpha by tort, and has started parceling off bits and pieces of the technology. First, the EV6 bus got licensed to AMD for the Athlon. Then it got licensed out to all those motherboard makers. Now they're selling off some architectural information and compilers for the Alpha to Intel. I guess the one glimmer of hope for all of us who tend to like the underdog is that Compaq has disposed of assets both to AMD and Intel, holding some sort of neutral position. That does leave some hope of avoiding a more-or-less Intel-only future.
Partner Policies of exclusivity
on
Ask Robert Young
·
· Score: 2
Many hardware vendors are starting to claim "support" for Linux, but are only creating binary-only, 386-compiled RPM driver packages, resulting in vendor-distribution lock-in for their customers. This is being done almost exclusively with RedHat Linux, and leaves a very bad taste in customer's mouths when they realize they have been burned.
An example culprit is Dell. The "Dell" RAID controller is a modified Adaptec RAID controller which is not supported under the kernel, but can be made to work under (only) RedHat 5.x/6.x while running a specific kernel version (the latest release was only for 2.2.14 kernels). I entirely understand that these driver choices are made entirely by the vendor (Dell) and are their responsibility, but a larger problem remains.
Dell offers "Linux" packaged on some of their hardware, and claim that they are compatible with "Linux" with that hardware.
Of course, much of their hardware is only "compatible" with RedHat Linux, and only under specific RPM releases for specific default RedHat kernels (such as the RAID situation just mentioned).
This hardware vendor trend has caused a sort of "embrace and extend" situation involving hard-compiled drivers packaged via RPM. I have to work with three "Linux-compatible" servers at work that are NOT Linux kernel compatible. I am forced to run RedHat 6.1 with the 2.2.14 kernel, because it is NOT possible to upgrade to a newer kernel, or a newer distribution. It certainly leaves the impression that RedHat is trying to push other Distributions out of the market, while locking-in their customers to quickly outdated (and vulnerability-ridden) software.
How does RedHat stand on the endorsement/promotion/disapproval of such actions by OEM partners, such as Dell?
How do you feel about the indirect damages upon the rest of the Linux community, due to RedHat's stance on this issue?
My father was a blood spatter analyst for the FDLE (Florida Dept. of Law Enforcement) for some years before becoming the chief forensic investigator for this district's medical examiner's office. I've often tried to help him (since ~1987) try to implement viable visualization tools for crime scenes, courtroom examples, etc. The complications on the technical side are very difficult, but what's most difficult is trying to match the exacting scientific rigor of forensics.
In order to build an accurate animated example, you need a CAD-level rendering engine that will allow you near-perfect specification of distance and relation. Secondly, this rendering engine has to be capable of making 2-D 'snapshots' at different depths on the axis you are intersecting, in order to make a graphical cross-section of the event. This is necessary for all sorts of fun stuff like determining how far an assailant's knife is from the victim at the point when she hits him over the head with a frying pan, etc..
Next, you need isometric 3D-like point of view for mathematically-useful 3D representations. Lastly, you need 1-point, 2-point, and 3-point perspective for the sake of providing clear visuals to the non-mathgeeks in the jury.
Sure, if you're DAMN good with some of the CAD and rendering tools out there, it CAN be done, but the problem is that it can't be done by most, if any, qualified and apt forensic investigators, who need to ensure the scientific accuracy and validity of such renderings.
A number of companies had come to my father and offered tools that were easy-to-use and supposedly mathematically precise. However, these tools ended up being very shoddy walk-through demos at best, and did not allow for any sort of scientifically-reasonable use. Essentially, none of the tools out there were capable of meeting the technical precision required for forensic investication and still be user-friendly enough for a reasonably computer-smart person to build in less than a month's time, just for one scene.
The big boys (FBI, etc) have the money to hire full-time ubergeek engineers to develop engineer-level-accuracy worlds for their displays, but the sort of money and expertise needed just isn't available to the medium-to-large forensic offices (serving 1.5 million people).
I'd love to see such tools, and look forward to playing with them in the future. I had a hell of a lot of fun with AutoCAD and MicroStation all those years ago!
Verisign acquired NSI... Also, the NSI organization has been kept complete. The delegation of authority has been passed to Verisign directly (and NOT NSI, which is a separate functioning body)... this means that they will be given a chance to take their expertise from the NSI people and rebuild it from scratch in this excercise... so, it's not exactly selling out to the enemy. It's actually a chance for Verisign (who is admittedly evil as hell) and the recently-purchased NSI (who hold the Webster's Dictionary definition for the term "evil as hell") to redeem themselves.
This is probably the most humble sort of action we will get to see from ICANN. It's about as close to an admittance that they need someone with more experience to handle the major chunk (.com,.net) of internet domain names. Now if they'd only take this humility to the next level, jump off a cliff like the bunch of lemmings they act like, and get replaced by some sane policymakers regarding international domains... THEN the domain name fiasco would have some progress..
With the horrible management of NSI, i've oftentimes been very curious about the usefulness of a.net-vs-.org domain, as no prerequisites were upheld in the domain name selection process. It's excellent to see ideas actually flowing through some important peoples' heads regarding the maintenance of quality in the existing TLD's. They seem to get ignored for the sake of discussing shiny new TLD's far too often.
I'd be really interested to see a legal case where the parent company of a website was sued for nested pop-up ads and banner ads, which were unwillingly using a customer's bandwidth for unauthorized purposes (aka a DoS attack)... it'd be an excellent legal precedent for the sake of upholding a customer's right to control what data comes across their connection.
Since this is Slashdot, and this community is rather defensive, if not tyrranical, about proper diction, it is necessary for me to point out your misuse of the word 'ethics' and recommend the replacement of 'morals.' Ethics are of a personal nature - a governing body of conscience of a smaller scale (personal) than morals, which affect a (sub)culture.
My ethics happen to conflict with a number of ideas you've taken for granted in your argument. I, for one, believe that the Internet is a medium for progressive scalable communication. It is meant for distributing (and re-distributing) information. As it was created, engineered, and still largely funded by the U.S. Government (i.e. tax dollars), I believe that U.S. citizens have a right to use the Government-created and still much-subsidized network without the hassles and frustrations as banner ads, popup ads, etc. I believe that it is in the best interest of the U.S. Government to extend this flexible resource to other Governments, which may choose to participate.
In this perspective, I see little difference between the Internet and the national library infrastructure. Books are shared between libraries and information is made available to the user. This is equally subsidized by the U.S. Government, and suffers from no equivalent of banner ads or popups.
My morals tend to agree with you, that for the sake of the existing culture, circumventing the ads that pay for services that are free (to the user, as in beer) and which are enjoyed and used. My morals take note of the counterproductive actions being taken, and I dislike the means in which the banner-ad problem has been dealt with. Nonetheless, these moral evaluations do not override my ethical judgement that advertising of this sort has no place in a still-largely government subsidized network infrastructure.
That particular drawing had an end goal of showing the specific mathematical proportions of body parts. It illustrated in a clear, easy-to-understand manner the mathematical relationships that exist in the skeletal structure of humans. Yes, it was secondarily art, but the art was only a means of easily conveying the ideas in a media form that many could understand.
I'm by no means saying that a brilliant scientist's work is not art, nor he an artist. The piece of work was indeed art, but was not primarily or predominantly art. That is one often-found difference between analytical art (such as elegant code, an impressive bridge, or blueprints for a house) and 'pure' art (painting, sculpture, music).
You are speaking of enabling bridging, which is neither a firewall nor a router.
Had what you described been true, there would have been no need for Microsoft to create Microsoft Proxy Server (version 1 OR version 2!), nor MS NAT, as implemented in Windows 2000 server.
Clarity helps a discussion, and OpenBSD really does have a kick-ass NAT implementation. It's not touched by any MS product. I've used them all.:)
If anyone here considers the following a flame, they need to dislodge their head from their colon... I grew up in a cop-family.
Police are often very excitable people. In an environment (such as a small town or an University) with little action, such a grossly hypothetical possible infraction as this clan-gamer site thing can easily lead to a few officers jumping the gun, leaving their supervisors in a position where they have to go ahead to save face ("umm.. yeah! that's what i, er, told my officers to do, sir!"), even after initial procedures and civil liberties have been violated.
In such a small environment, there's rarely a proper procedure or infrastructure to investigate allegations of illegal activity. This leads to cops jumping up, getting some absolutely asinine warrant (thanks to the equally clueless judge or justice who approved the warrant!), and a clan-gamer gets their room raided.
It's a matter of idiots, acting outside of their responsible authority. Idiots. lots of them. in unison and cheerful coordination.
Heh, it's also humourous that the scheduled meeting to ask the kids about this website was just cancelled. I've had this happen to me before, regarding educational facilities and technical goings-on, and then just rashly cancel a program they don't understand that's already half-started (such as a high school starting an animation class and needing networking equipment). The intellectually out-of-touch often act rashly to 'protect' their perceived authorities and possible loss of authority.
The police were idiots. they were poorly tamed and irresponsible idiots. they did a piss-poor job. they'll get slapped on the wrist and the whole cycle will repeat again somewhere else. The worst thing is that all of this confusion could so easily be solved with a minimal application of common sense.
The other solution would have been for them to actually read the webpage, and perhaps do a Search-Engine search for terminology they didn't understand. Good luck getting people to read these days, though..
The split was kinda funky. The way I understand it is that all chips/ASICs/blahblah (such as 21152 PCI bridge, 2114x-series network interfaces, StrongARM processors, etc) were bought by Intel, while Compaq bought only the Alpha architecture stuffs. There may be some other considerations, but that much i know is for sure.
Last I checked, the ARM series of embedded chips was designed by DEC, and purchased by Intel.... making Intel (of course, what's new *snicker*) the primary chip manufacturer in this realm.
It still pisses me off that Intel bought a perfectly good DEC, and has since buried many of their better technologies away from public use.
</flame>
Comparing sources to packages is absurd, asinine, and just a flat out troll.
Citing only a Debian-friendly example for how to get proprietary packages on Debian GNU/Linux only also serves no benefit, especially since it's an off-topic note anyways.
Also,/stand/sysinstall's ability to install specific packages is IRRELEVANT! It's a system installer, not an exclusive package management system. Also, if you've ever worked with a 'BSD.tgz "package", it's a very simple thing... "tar -zxfC / blah.tgz"... just unpack it to the root directory, and it's installed. Also, any time the ports package is used to compile something, it builds a.tgz so that you can use it in the future to install the package on that (or other) systems. THAT is smart.
Want to know the best part? I can't stand FreeBSD. It drives me nuts. However, a lot of the things in FreeBSD are much more sane than typical Linux implementations. Mostly because "typical" Linux gets translated to Debian and RedHat... and although I respect Debian, It does some peculiar and proprietary stuff.
Slackware Linux is a happy medium. It is structured BSD-style, is 100% bad-linux-implementation-compatible, and otherwise could easily be confused as being FreeBSD to the casual onlooker. It is also the only distribution I've found that works reliably and cleanly with other Debian/FreeBSD/OpenBSD/Solaris systems without any heavy tweaking. Of course, some tweaking is nice for marginal performance increases, but the point is that it's compatible and a quick performer by default.
Yep, OpenBSD has a very thorough out-of-the-box NAT firewalling solution. It's extremely powerful (and in my experience far superior to linux 2.2 ipmasq)... I use it at home and work with OLD Sparcs (25-40mhz) and still achieve near 10mbps on all 10mbps interfaces.
Iptables under the 2.4 kernel has impressed me. I use it for a test lab at work, and the first thing I noticed was that it allows active FTP through its implementation of NAT. It's probably due to some funky extension they've added to FTP standards, i'm not sure, nor good enough of a programmer to figure it out from the code... Nonetheless, they've finally made a decent product that starts comparing to the cleanliness and quality of BSD's NAT.
The other BSDs also use the same BSD NAT, however, my last experience of using NAT on FreeBSD (back on 3.3) was that it was quite erratic in behavior compared to the same setup using OpenBSD. I might have been royally screwing something up, too:)
Correction: you work for money. He obviously doesn't work only for money, else he would have already booked out the door and landed another job.
There is more to employment than just money. Some people take certain jobs or work in certain fields for idealistic purposes as well.
Nevertheless, this situation sounds like no matter which side of the coin is chosen, departure is necessary. You're definitely dead-on about the shakiness of the company if one employee leaves.
The design conceptuality is the biggest factor here... SPARC processors are built with a common, strict instruction set, and all performance enhancements are done at the hardware level (anything that is beyond the user-definable assembly-like language). Intel adds extensions to each new processor they come out with. Comparing apples to apples doesn't exactly work, when it comes to FPU/ALU/FLOPS/etc... here's why.
Take a Pentium Pro 200 and benchmark it on the exact same software with a PIII-1ghz. Theoretically, there should be a five-fold scaling difference in performance. However, applications that have not had additional optimization code added in for the newer PIII architecture do NOT get a five-fold increase in performance. Expect to see something along the lines of three-fold.
Applications that _have_ been optimized for the newer architecture will blow away the old chips. For example, look at Quake III on a Pentium Pro 200 with a PCI Video card, and then try the same benchmarks with all the same variables with a PIII-1ghz. the difference is more like 20-25fold than it is five-fold.
Either way, it's a lose-lose situation for Intel. If you buy a fast, expensive server, it won't be scalably faster than your old server when using the same software. Also, in another year, when SSE3 and MMX revision 234023 come out on Intel's next chip (or at least, next microcode release), your now-old server won't be optimized for the code in new software and will run it poorly.
With Sun's SPARC architecture, this stuff doesn't happen. Of course, it did from SPARC to UltraSPARC, but that's a HUGE architectural change - that's expected. Just like going from a PIII to an IA64-based chip, when they come out. Bottom line is that you're playing guessing games and software-matching-tag to figure out if an Intel-based server upgrade will really help you, where in the Sun market, your upgrade will be matched equally and as-expected with greater performance.
If you really *like* tweaking and modifying and optimizing software at the instruction-level on your servers, hey, I won't stop you... I know that I've got way too many servers to manage to possibly consider that as a reasonable option in the enterprise.
On the nitpicking technical side, integrated network controllers ARE PCI devices. They are connected directly to the PCI bus and use a local PCI assignment. They communicate to the processor through the same PCI bridge as a PCI-Card. They *are* PCI devices. Thus, there is no performance advantage in having an integrated network controller.
A similar situation goes for integrated video. Either you have a third-party video subsystem soldered on the board (best circumstance), which uses the PCI or AGP bus directly, or you have an on-chipset implmentation of onboard video. The on-chipset implementations tie into the PCI or AGP bus as well, but to keep costs down, the chipset manufacturers build a truly inferior video system. Either they build it with some stupid blocking-mode so you can't hit the PCI bus while updating video (This includes the Intel i8xx chipsets w/ integrated video) or it has no hardware-acceleration at all for line-drawing, let alone mpeg video at anything above 16-bit color.
The reason I dislike integrated features is because they are repeatedly implemented poorly, unless you spend two to three times as much for a superior motherboard with superior parts. Even still, your flexibility and ability to choose is limited. Personally, I'd spend twice as much for a machine with NO integrated 3com network controller, because I don't want to use a 3com controller.
Frankly, integrated video solutions still suck. I'm typing this on an i815-based system with integrated video (thankfully it was disable-able and there's an AGP slot) and integrated sound, both of which use fully-blocking I/O and literally HALT the system if I try to use them to any degree. Can't play an MP3 and edit text files with the onboard sound, because the file I/O screws with the sound's precious I/O stream, which thus screws up the video's I/O, causing me to have what appears to be bad video RAM due to a shitty chipset.
If you have the choice of buying integrated, don't. period. I don't give a damn if it's for an appliance, a toaster, or a damn chicken coop. There's a reason this CuMine 533EB i'm on is not even HALF the speed of my PII-350 at home - integrated features and flat-out inefficiency.
(I was a production manager for a computer manufacturer, so don't flame me unless you've got some real-world experience with this stuff. thanks.)
Slashdot Mirror
I run a lot of old SPARCs, mostly SPARCStation 5's. I run primarily Solaris and OpenBSD on them, depending on purpose... However, I ran two SPARCs on Linux for about 6 months as production servers. (shell, mail, dns, web, ftp, some sql)
:) The Linux kernel itself isn't very mature for SPARCs, and its process creation time is AWFUL. These machines managed to handle about 200 users for 6 months, before the load just got too high. However, kernel-level resource problems required that I rebooted the machines about once every 60 days.
:)
At the time, three main choices existed: RedHat, SuSE, and Debian. Redhat installed rather easily, but it completely sucked, and a lot of things were just broken. In fact, the machines locked after less than two days of testing. Lather, rinse, repeat - same results for 2 weeks.
I tried for like 12 hours to get SuSE to install, and I just couldn't hack it. I'm _very_ familiar with both Sun hardware and Linux, so I was lead to believe that it was just bad packaging. I gave up on SuSE.
I installed Debian... The first two installs weren't quite right (user error!)... but after I got things in order, I installed Debian on these machines, put them through testing hell for a week, and put 'em up. They ran pretty well for being 110mhz
As far as SPARCs go, Solaris has very little competition. NetBSD and OpenBSD both run a bit faster, but file I/O is noticeably slower. However, *BSD is insanely stable on SPARCs (I've got a server with over 280 days uptime on SPARC/OpenBSD). Linux doesn't really hack it yet. Perhaps I'll get bored in the next few months and try out Slackware's SPARC distro... If anyone can do it right, it's Slackware
RedHat is FAR too close to Microsoft, in happy-friendly GUI ways, basic utilities crash constantly ways, and ebrace-and-extend ways.
/etc/sysconfig/network/*, ifconfig, and route.
:)
1. Happy-friendly ways. chkconfig, linuxconf, rpm. The ideas are nice. they're very nice ideas. However, they're botched in a number of ways i'll soon get to. KDE, GNOME, whatever is available in whatever default installation of RedHat is nice and quite useable. Just like Microsoft. That is certainly admirable. Doesn't make it good, though.
2. Basic utilities crash. Have any of you done an RH 6.2 FTP install? Hell, even a custom CD install can easily pull a segfault out of RedHat. netconfig on 5.x and 6.1 will segfault if it doesn't get specific PTR lookup responses. Even worse than that, it'll segfault in the middle, so you don't know if it's changed your sysconfig settings, your live IP address/subnet/default route, or what... and if they're out of sync, don't expect netconfig to work once more to fix it. It'll just segfault some more and leave you hanging unless you can vi
3. Embrace and extend. RedHat has positioned itself to be the high-market-share distro. However, RedHat intentionally releases broken standards (RH 7's egcs for one?) and moves things around in such a way that if a software developer writes a program to be installed on a RedHat Linux system, it won't install on any non-RedHat-based distribution. If it does install, the crazy egcs release will keep it from running on the new machine. RedHat often screws with things like init scripts just enough to make them UNIX-like, but to break POSIX standards. What pisses me off about RedHat is how deliberate their embrace-and-extend design policies are.
I don't recommend RedHat for anything, because learning the quirks of RedHat puts users into bad practices of using their proprietary tools, or expecting the proprietary behavior of their tools to be standard cross-platform. It's sort of like how a lot of linux distros have a 'route' command that, for some reason, won't accept 'route add -net default' (which is standard across UNIX) but will accept 'route add default gw'... annoying.
And why do users use SysV-style Linux distros, and still use ps -ax? why not ps -ef?
The government has turned down the prospects of creating a counter-worm, but any decently-experienced assembly programmer with sockets experience could just disassemble the current worm, make a number of changes, and release the worm on a time-skewed box. A really crafty assembly programmer could even keep the binary size of the worm the same, so in case the worm has some self-check mechanism, it won't notice any difference. I personally wouldn't mind seeing this fire fought with fire - Let the anti-worm run its course for a month, and then have it destroy itself. That would wipe out the vast majority of the code-red virus.
It is a really rash and dangerous tactic, but considering the scenario that a number of people are expecting from this worm, are there really any other effective options?
The corrected URL:
& startpos=242423
http://www.angio.net/pi/bigpi.cgi?UsrQuery=424242
Searching for a 6-digit, not an 8-digit, string...
No kidding! The FF-style storylines are normally very complex and very deep... In a game, you've got 20-60 hours of gameplay in which you get to mill over and decipher the storyline, so it makes a bit more sense. Sadly, they had to try to cram an FF-quality storyline into a 2-hour movie, and I think that was its greatest weakness.... time.
The storyline, of course, borrowed from a bunch of story ideas in the past (Gaya vs. FF7's lifestream?, phantoms vs. Jenova?), and kept with the fairly common mystical/religious undertones that help make the FF series what it is. For someone like Katz (a person who hasn't played any of the games, and is, well, as silly as he is), the story line is either going to seem overly complicated (and thus underexplained) or extremely hokey and childish.
Every FF story is a challenge to read. It is also a challenge to the reader, regarding how one lives their life. I know I'm probably reading too much into this, but the point is that the FF storylines aren't supposed to be just entertaining like Angelina Jolie's tits in Tomb Raider... the storylines of FF games (and movie) make you think. a lot.
Precisely it: Jack of all trades, master of none. However, I am thoroughly aware of what I don't know.
... the DBA...'.
I'm no DB genius. I'm hardly even decent at it. I was just given that responsibility because I could get it done without constantly trashing junking and re-doing the databases and/or servers. That doesn't make me special - it just makes me, 'essentially
Hell, i'm not excellent at any of the specialist tasks that I do. I may be able to write mediocre code in a few languages, but that doesn't make me the prime candidate for a project lead. And I know that. And I never even hinted that I should be considered for such things.
In case you didn't read too well, I didn't make any 'wild statements to boost my own ego'. I stated what it is I do. I didn't even say that I do it well. I just said that I do it. And I, too, let my record speak for itself. Do you think a college-dropout geek is going to get a decent job without a decent (though short) track record of accomplishments? I sure don't.
Please, don't confuse acknowledging that I know something (or even a good deal about something) with knowing everything.
I'm the stereotypical candidate for prima donna syndrome: a few days shy of 21, dropped out of the engineering program at a state University because it was unchallenging and mediocre on its very best days, and dove into the IT field. I'm a Unix Sysadmin for a little company with scrambling and confused management - a glorified dot-com.
Since it's a small company, I'm essentially also the DBA, network admin, Cisco guru, neurotic PERL geek, and so on. I get frustrated quickly with people. I word my sentences carefully to provide the most clear and concise meaning possible (management calls these 'very curt' responses!), and attempt to usher the questioning coworker out of my line of attention as quickly as possible. I tell programmers that their idea on implementing this or that is "like an ostrich - it's got wings, but there's no way in hell it's gonna fly." I'm a young, cocky asshole.
But WHY? It seems that no one has asked WHY we prima donna types are this way. My explaination is that I'm a die-hard perfectionist. I'm very interested in the architecture of things... both concepts and actual structures. I'm big on using available standards, or creating thoroughly documented standards if necessary. I'm big on harmony. I don't like solutions that plug one hole in a leaking boat, just so the water can come in through another large crack. I'm a die-hard perfectionist. Though I'm more than willing to throw really bastardized hacks into place when they do not create new problems.
If a programmer who specializes in socket programming comes to me with an idea on how to do task X, and I can think of multiple more efficient and more effective ways to do task X (NOTE: I am NOT a socket programmer, nor a specialist at socket programming), I will point the weaknesses that I see in their idea and offer the ideas that came to mind. I'm always constructive and ALWAYS offer alternative solutions, though my thoroughly-learned tendency to be concise with my words sets them on the wrong foot. Half an hour later, my supervisor calls me in and asks me about the 'incident'. He's actually quite understanding and open-minded. I love explaining my reasoning to him - he remembers it and often uses it productively. I explain my reasoning, he's happy, and I'm back to work. However, upper management (2 people, it's a small company) slowly builds an image of me being unfriendly and not helpful. Bad situation for me.
Management looks at things in terms of investment, risk, and a few other things that I'm not overly attentive of. Technical people often look at things in terms of efficiency and merit of design. However, only a small percentage of techies I know also disassemble ideas and concepts into security and liability to their company. Well, then again, most techies probably work in an environment where the management (at least) has liability already covered before ideas/problems/customers get down to them.
The merits of design are not the merits of finance and profit. The two sides oftentimes dislike thinking about the other's point of view, or are unfamiliar/afraid of it to some degree.
The bottom line: I am a prima donna because my point of view of any given situation is very different than management's point of view. I am not excessively willing to look at their point of view, and likewise they are not excessively willing to look at mine. I accept that and try my best to work with them on sharing our points of view. However, a 100% technically-oriented company cannot survive with a 0% technically-oriented management running the show. The components that make the company work aren't going to be properly filled in the right proportions. There's only so far I can stretch to make such things work... after that point, I'm called a prima donna and management holds their noses high. That's fine by me - my self esteem isn't hurt by other peoples' opinions (that i consider misconceptions)... that sort of behavior would not allow me to function well in the technical manner that serves my employer.
It's a problem of point of view causing frustration.
My first thought was that most people living in the projects of your nearby large-city wouldn't have such snazzy GPS systems. Thus, everytime you wanted to take the interstate into downtown, the computer/cellphone/whatever would beep and tell you to take Martin Luther King Blvd under the bridge because it'd think there were fewer cars down that way.
:)
That reminds me about the wonderful directions MapQuest once gave me that took me to the middle of the projects in Atlanta, when I wanted to get to Emory University. Trying to get out of the projects in an unfamiliar city at 4am with 3 geeks in a pretty green saturn isn't exactly my idea of fun (well, it's funny now, but wasn't while it was happening!
Such communal reinforcement of the underdog is what makes a free-enterprise democracy stand out from the crowd. The US market (where these companies are based, located, run from, and count their revenue quarterly) is a self-proclaimed free enterprise, but is so politely tuned, chopped into portions, and regulated with as many bad as there are good laws that any G.E.D. graduate can see the US market is as much socialist as it is democratic.
Supporting an underdog is the necessary social response in order to maintain an open, competitive market. There is no doubt that Intel has a very heavy hand in the commodity CPU industry, and the acquisition of Alpha technology by one of Intel's competitors would help level the playing field. This sort of competition would benefit the customers, and the market, much more than Intel gobbling this up just to ensure that no one else can give them a scare in their niche market.
Remember that when DEC sold off their assets, they sold a LOT of microprocessor technology to Intel, including the StrongARM, their chipsets, their embedded controllers, and their NICs. They certainly had enough of a relationship with Intel that they could have worked a deal on the Alpha technology if they wanted to. Instead, DEC sold the Alpha processor to Compaq, under the hope that they would make it grow and keep the very high-quality processor in the market.
It certainly looks like Compaq has been slowly killing the Alpha by tort, and has started parceling off bits and pieces of the technology. First, the EV6 bus got licensed to AMD for the Athlon. Then it got licensed out to all those motherboard makers. Now they're selling off some architectural information and compilers for the Alpha to Intel. I guess the one glimmer of hope for all of us who tend to like the underdog is that Compaq has disposed of assets both to AMD and Intel, holding some sort of neutral position. That does leave some hope of avoiding a more-or-less Intel-only future.
Many hardware vendors are starting to claim "support" for Linux, but are only creating binary-only, 386-compiled RPM driver packages, resulting in vendor-distribution lock-in for their customers. This is being done almost exclusively with RedHat Linux, and leaves a very bad taste in customer's mouths when they realize they have been burned.
An example culprit is Dell. The "Dell" RAID controller is a modified Adaptec RAID controller which is not supported under the kernel, but can be made to work under (only) RedHat 5.x/6.x while running a specific kernel version (the latest release was only for 2.2.14 kernels). I entirely understand that these driver choices are made entirely by the vendor (Dell) and are their responsibility, but a larger problem remains.
Dell offers "Linux" packaged on some of their hardware, and claim that they are compatible with "Linux" with that hardware. Of course, much of their hardware is only "compatible" with RedHat Linux, and only under specific RPM releases for specific default RedHat kernels (such as the RAID situation just mentioned). This hardware vendor trend has caused a sort of "embrace and extend" situation involving hard-compiled drivers packaged via RPM. I have to work with three "Linux-compatible" servers at work that are NOT Linux kernel compatible. I am forced to run RedHat 6.1 with the 2.2.14 kernel, because it is NOT possible to upgrade to a newer kernel, or a newer distribution. It certainly leaves the impression that RedHat is trying to push other Distributions out of the market, while locking-in their customers to quickly outdated (and vulnerability-ridden) software.
How does RedHat stand on the endorsement/promotion/disapproval of such actions by OEM partners, such as Dell?
How do you feel about the indirect damages upon the rest of the Linux community, due to RedHat's stance on this issue?
My father was a blood spatter analyst for the FDLE (Florida Dept. of Law Enforcement) for some years before becoming the chief forensic investigator for this district's medical examiner's office. I've often tried to help him (since ~1987) try to implement viable visualization tools for crime scenes, courtroom examples, etc. The complications on the technical side are very difficult, but what's most difficult is trying to match the exacting scientific rigor of forensics.
In order to build an accurate animated example, you need a CAD-level rendering engine that will allow you near-perfect specification of distance and relation. Secondly, this rendering engine has to be capable of making 2-D 'snapshots' at different depths on the axis you are intersecting, in order to make a graphical cross-section of the event. This is necessary for all sorts of fun stuff like determining how far an assailant's knife is from the victim at the point when she hits him over the head with a frying pan, etc.. Next, you need isometric 3D-like point of view for mathematically-useful 3D representations. Lastly, you need 1-point, 2-point, and 3-point perspective for the sake of providing clear visuals to the non-mathgeeks in the jury.
Sure, if you're DAMN good with some of the CAD and rendering tools out there, it CAN be done, but the problem is that it can't be done by most, if any, qualified and apt forensic investigators, who need to ensure the scientific accuracy and validity of such renderings.
A number of companies had come to my father and offered tools that were easy-to-use and supposedly mathematically precise. However, these tools ended up being very shoddy walk-through demos at best, and did not allow for any sort of scientifically-reasonable use. Essentially, none of the tools out there were capable of meeting the technical precision required for forensic investication and still be user-friendly enough for a reasonably computer-smart person to build in less than a month's time, just for one scene.
The big boys (FBI, etc) have the money to hire full-time ubergeek engineers to develop engineer-level-accuracy worlds for their displays, but the sort of money and expertise needed just isn't available to the medium-to-large forensic offices (serving 1.5 million people).
I'd love to see such tools, and look forward to playing with them in the future. I had a hell of a lot of fun with AutoCAD and MicroStation all those years ago!
Verisign acquired NSI... Also, the NSI organization has been kept complete. The delegation of authority has been passed to Verisign directly (and NOT NSI, which is a separate functioning body)... this means that they will be given a chance to take their expertise from the NSI people and rebuild it from scratch in this excercise... so, it's not exactly selling out to the enemy. It's actually a chance for Verisign (who is admittedly evil as hell) and the recently-purchased NSI (who hold the Webster's Dictionary definition for the term "evil as hell") to redeem themselves.
I hate 'em both, but i'm optimistic.
This is probably the most humble sort of action we will get to see from ICANN. It's about as close to an admittance that they need someone with more experience to handle the major chunk (.com, .net) of internet domain names. Now if they'd only take this humility to the next level, jump off a cliff like the bunch of lemmings they act like, and get replaced by some sane policymakers regarding international domains... THEN the domain name fiasco would have some progress..
.net-vs-.org domain, as no prerequisites were upheld in the domain name selection process. It's excellent to see ideas actually flowing through some important peoples' heads regarding the maintenance of quality in the existing TLD's. They seem to get ignored for the sake of discussing shiny new TLD's far too often.
With the horrible management of NSI, i've oftentimes been very curious about the usefulness of a
I'd be really interested to see a legal case where the parent company of a website was sued for nested pop-up ads and banner ads, which were unwillingly using a customer's bandwidth for unauthorized purposes (aka a DoS attack)... it'd be an excellent legal precedent for the sake of upholding a customer's right to control what data comes across their connection.
Anyone seen anything like this happen yet?
Since this is Slashdot, and this community is rather defensive, if not tyrranical, about proper diction, it is necessary for me to point out your misuse of the word 'ethics' and recommend the replacement of 'morals.' Ethics are of a personal nature - a governing body of conscience of a smaller scale (personal) than morals, which affect a (sub)culture.
My ethics happen to conflict with a number of ideas you've taken for granted in your argument. I, for one, believe that the Internet is a medium for progressive scalable communication. It is meant for distributing (and re-distributing) information. As it was created, engineered, and still largely funded by the U.S. Government (i.e. tax dollars), I believe that U.S. citizens have a right to use the Government-created and still much-subsidized network without the hassles and frustrations as banner ads, popup ads, etc. I believe that it is in the best interest of the U.S. Government to extend this flexible resource to other Governments, which may choose to participate.
In this perspective, I see little difference between the Internet and the national library infrastructure. Books are shared between libraries and information is made available to the user. This is equally subsidized by the U.S. Government, and suffers from no equivalent of banner ads or popups.
My morals tend to agree with you, that for the sake of the existing culture, circumventing the ads that pay for services that are free (to the user, as in beer) and which are enjoyed and used. My morals take note of the counterproductive actions being taken, and I dislike the means in which the banner-ad problem has been dealt with. Nonetheless, these moral evaluations do not override my ethical judgement that advertising of this sort has no place in a still-largely government subsidized network infrastructure.
That particular drawing had an end goal of showing the specific mathematical proportions of body parts. It illustrated in a clear, easy-to-understand manner the mathematical relationships that exist in the skeletal structure of humans. Yes, it was secondarily art, but the art was only a means of easily conveying the ideas in a media form that many could understand.
I'm by no means saying that a brilliant scientist's work is not art, nor he an artist. The piece of work was indeed art, but was not primarily or predominantly art. That is one often-found difference between analytical art (such as elegant code, an impressive bridge, or blueprints for a house) and 'pure' art (painting, sculpture, music).
You are speaking of enabling bridging, which is neither a firewall nor a router.
:)
Had what you described been true, there would have been no need for Microsoft to create Microsoft Proxy Server (version 1 OR version 2!), nor MS NAT, as implemented in Windows 2000 server.
Clarity helps a discussion, and OpenBSD really does have a kick-ass NAT implementation. It's not touched by any MS product. I've used them all.
If anyone here considers the following a flame, they need to dislodge their head from their colon... I grew up in a cop-family.
Police are often very excitable people. In an environment (such as a small town or an University) with little action, such a grossly hypothetical possible infraction as this clan-gamer site thing can easily lead to a few officers jumping the gun, leaving their supervisors in a position where they have to go ahead to save face ("umm.. yeah! that's what i, er, told my officers to do, sir!"), even after initial procedures and civil liberties have been violated.
In such a small environment, there's rarely a proper procedure or infrastructure to investigate allegations of illegal activity. This leads to cops jumping up, getting some absolutely asinine warrant (thanks to the equally clueless judge or justice who approved the warrant!), and a clan-gamer gets their room raided.
It's a matter of idiots, acting outside of their responsible authority. Idiots. lots of them. in unison and cheerful coordination.
Heh, it's also humourous that the scheduled meeting to ask the kids about this website was just cancelled. I've had this happen to me before, regarding educational facilities and technical goings-on, and then just rashly cancel a program they don't understand that's already half-started (such as a high school starting an animation class and needing networking equipment). The intellectually out-of-touch often act rashly to 'protect' their perceived authorities and possible loss of authority.
The police were idiots. they were poorly tamed and irresponsible idiots. they did a piss-poor job. they'll get slapped on the wrist and the whole cycle will repeat again somewhere else. The worst thing is that all of this confusion could so easily be solved with a minimal application of common sense.
The other solution would have been for them to actually read the webpage, and perhaps do a Search-Engine search for terminology they didn't understand. Good luck getting people to read these days, though..
The split was kinda funky. The way I understand it is that all chips/ASICs/blahblah (such as 21152 PCI bridge, 2114x-series network interfaces, StrongARM processors, etc) were bought by Intel, while Compaq bought only the Alpha architecture stuffs. There may be some other considerations, but that much i know is for sure.
Last I checked, the ARM series of embedded chips was designed by DEC, and purchased by Intel.... making Intel (of course, what's new *snicker*) the primary chip manufacturer in this realm.
It still pisses me off that Intel bought a perfectly good DEC, and has since buried many of their better technologies away from public use. </flame>
Comparing sources to packages is absurd, asinine, and just a flat out troll.
/stand/sysinstall's ability to install specific packages is IRRELEVANT! It's a system installer, not an exclusive package management system. Also, if you've ever worked with a 'BSD .tgz "package", it's a very simple thing... "tar -zxfC / blah.tgz" ... just unpack it to the root directory, and it's installed. Also, any time the ports package is used to compile something, it builds a .tgz so that you can use it in the future to install the package on that (or other) systems. THAT is smart.
Citing only a Debian-friendly example for how to get proprietary packages on Debian GNU/Linux only also serves no benefit, especially since it's an off-topic note anyways.
Also,
Want to know the best part? I can't stand FreeBSD. It drives me nuts. However, a lot of the things in FreeBSD are much more sane than typical Linux implementations. Mostly because "typical" Linux gets translated to Debian and RedHat... and although I respect Debian, It does some peculiar and proprietary stuff.
Slackware Linux is a happy medium. It is structured BSD-style, is 100% bad-linux-implementation-compatible, and otherwise could easily be confused as being FreeBSD to the casual onlooker. It is also the only distribution I've found that works reliably and cleanly with other Debian/FreeBSD/OpenBSD/Solaris systems without any heavy tweaking. Of course, some tweaking is nice for marginal performance increases, but the point is that it's compatible and a quick performer by default.
Following logical standards is a Good Thing (tm).
Yep, OpenBSD has a very thorough out-of-the-box NAT firewalling solution. It's extremely powerful (and in my experience far superior to linux 2.2 ipmasq)... I use it at home and work with OLD Sparcs (25-40mhz) and still achieve near 10mbps on all 10mbps interfaces.
:)
Iptables under the 2.4 kernel has impressed me. I use it for a test lab at work, and the first thing I noticed was that it allows active FTP through its implementation of NAT. It's probably due to some funky extension they've added to FTP standards, i'm not sure, nor good enough of a programmer to figure it out from the code... Nonetheless, they've finally made a decent product that starts comparing to the cleanliness and quality of BSD's NAT.
The other BSDs also use the same BSD NAT, however, my last experience of using NAT on FreeBSD (back on 3.3) was that it was quite erratic in behavior compared to the same setup using OpenBSD. I might have been royally screwing something up, too
Correction: you work for money. He obviously doesn't work only for money, else he would have already booked out the door and landed another job.
There is more to employment than just money. Some people take certain jobs or work in certain fields for idealistic purposes as well.
Nevertheless, this situation sounds like no matter which side of the coin is chosen, departure is necessary. You're definitely dead-on about the shakiness of the company if one employee leaves.
The design conceptuality is the biggest factor here... SPARC processors are built with a common, strict instruction set, and all performance enhancements are done at the hardware level (anything that is beyond the user-definable assembly-like language). Intel adds extensions to each new processor they come out with. Comparing apples to apples doesn't exactly work, when it comes to FPU/ALU/FLOPS/etc... here's why.
Take a Pentium Pro 200 and benchmark it on the exact same software with a PIII-1ghz. Theoretically, there should be a five-fold scaling difference in performance. However, applications that have not had additional optimization code added in for the newer PIII architecture do NOT get a five-fold increase in performance. Expect to see something along the lines of three-fold.
Applications that _have_ been optimized for the newer architecture will blow away the old chips. For example, look at Quake III on a Pentium Pro 200 with a PCI Video card, and then try the same benchmarks with all the same variables with a PIII-1ghz. the difference is more like 20-25fold than it is five-fold.
Either way, it's a lose-lose situation for Intel. If you buy a fast, expensive server, it won't be scalably faster than your old server when using the same software. Also, in another year, when SSE3 and MMX revision 234023 come out on Intel's next chip (or at least, next microcode release), your now-old server won't be optimized for the code in new software and will run it poorly.
With Sun's SPARC architecture, this stuff doesn't happen. Of course, it did from SPARC to UltraSPARC, but that's a HUGE architectural change - that's expected. Just like going from a PIII to an IA64-based chip, when they come out. Bottom line is that you're playing guessing games and software-matching-tag to figure out if an Intel-based server upgrade will really help you, where in the Sun market, your upgrade will be matched equally and as-expected with greater performance.
If you really *like* tweaking and modifying and optimizing software at the instruction-level on your servers, hey, I won't stop you... I know that I've got way too many servers to manage to possibly consider that as a reasonable option in the enterprise.
On the nitpicking technical side, integrated network controllers ARE PCI devices. They are connected directly to the PCI bus and use a local PCI assignment. They communicate to the processor through the same PCI bridge as a PCI-Card. They *are* PCI devices. Thus, there is no performance advantage in having an integrated network controller.
A similar situation goes for integrated video. Either you have a third-party video subsystem soldered on the board (best circumstance), which uses the PCI or AGP bus directly, or you have an on-chipset implmentation of onboard video. The on-chipset implementations tie into the PCI or AGP bus as well, but to keep costs down, the chipset manufacturers build a truly inferior video system. Either they build it with some stupid blocking-mode so you can't hit the PCI bus while updating video (This includes the Intel i8xx chipsets w/ integrated video) or it has no hardware-acceleration at all for line-drawing, let alone mpeg video at anything above 16-bit color.
The reason I dislike integrated features is because they are repeatedly implemented poorly, unless you spend two to three times as much for a superior motherboard with superior parts. Even still, your flexibility and ability to choose is limited. Personally, I'd spend twice as much for a machine with NO integrated 3com network controller, because I don't want to use a 3com controller.
Frankly, integrated video solutions still suck. I'm typing this on an i815-based system with integrated video (thankfully it was disable-able and there's an AGP slot) and integrated sound, both of which use fully-blocking I/O and literally HALT the system if I try to use them to any degree. Can't play an MP3 and edit text files with the onboard sound, because the file I/O screws with the sound's precious I/O stream, which thus screws up the video's I/O, causing me to have what appears to be bad video RAM due to a shitty chipset.
If you have the choice of buying integrated, don't. period. I don't give a damn if it's for an appliance, a toaster, or a damn chicken coop. There's a reason this CuMine 533EB i'm on is not even HALF the speed of my PII-350 at home - integrated features and flat-out inefficiency.
(I was a production manager for a computer manufacturer, so don't flame me unless you've got some real-world experience with this stuff. thanks.)