While the article doesn't come out and say it, I'm guessing there will be serious total bandwidth limitations- since there are no tracking antennas, and since this is referred to as part of a satellite TV service, it must be from (and probably to) a satellite in geosync orbit. All traffic is probably going to just one satellite. Once a lot of people sign up for this- popularity will kill it- since I'm sure that they can sign people up to long term contracts faster than they can put more transponders up.
You obviously don't know what you're talking about... The average transponder Bandwidth on the newer GEO satellites is 72 MHz, using 8PSK modulation with FEC 2/3 Viterbi or 216/205 Reed-Solomon encoding gives you roughly 90 Mbps to use... most High-Speed VSAT (Very Small Aperture Terminal) systems allocate bandwidth in two chunks, one outbound carrier (From the hub) and several inbound carriers (From the station). Those are TDM/TDMA slotted carriers where you should contend for available slots (Slotted-ALOHA) which usually have an efficiency of 60-65%. If you split the transponder in two you have 45 Mbps downstream (Subject to a 85% efficiency limit)and several (Usually 2-4 Mbps) upstream carriers which serve a group of VSAT's.
Since you're using a multiple allocation mechanism you can effectively guarantee at least some upstream traffic per station using streams (Pre-assigned slots). Have too many users in your service, get another transponder, it's not expensive to implement since these kinds of systems have independent modems.
Plus there is the fact that anything beamed to you is probably also being beamed to everyone in a multiple state area around you. I sure hope they have some *strong* encryption built in. They ought to be doing something better than DES, no?
Oh brother... can you tell me how are you going to demodulate the signal you receive from the satellite? (QPSK and 8PSK are not easy to implement with a breadboard you know?, not taking into account the fact that you'll need to build a fairly big decoder for viterbi or Reed-Solomon plus the encryption on the timeslots). You will receive what's intended for you and only you... you will be expecting some downstream timeslots and your IDU (Indoor Unit) will only decode those.
I hate to say it, but the microsoft backed Teledesic system is a much better system (many satellites in low earth orbit- kinda like a cellular system) the satellites are closer and cover smaller areas, so the amount of bandwidth/satellite is much smaller.
Of course you don't mention that these satellites are smaller, have less transponders, drift like mad and have poor signal footprints (The amount of power you receive down in earth). Cellular systems only work for small amounts of data (Look at how painful it has been to get GPRS working, not to mention 3G). Of course, fixed cell systems like LMDS do provide better bandwidth but they use modified versions of ATM, Frame Relay or PPP in order to dynamically allocate bandwidth and those aren't good when you have 500 users downloading pr0n at the same time.
If you need more proof that these kinds of systems work (Only for mail and browsing, trust me) look for DVB (Digital Video Broadcast) services, or check out systems from Hughes Network Systems or STMI.
Only in the cities they serve. Time Warner's Roadrunner is the other monopoly. Except that DSL is becoming more and more available. Which is competition for both TW and ATT.
If that's not monopoly on cable services I don't know what it is.
So they really don't have a "monopoly in the US". Just in certain cities.
Actually the US has a duopoly which is indeed (In economic and service terms) far worse than a monopoly (Why open us, we have competition from the other guys).
Judging on the same basis intel should be sued by Zilog because of the same argument. As a fact... the only thing AMD and Cyrix haven't been able to perfectly emulate have been the many design flaws (FDIV bug, instability at more than 1 GHz, an MTH that doesn't work, etc. etc.). As a matter of fact, AMD and Cyrix designs are derivative compatible works that the only thing they share in common with Intel is the Instruction Set (And not even so because AMD uses a RISC engine that decodes and translates the IA-32 instructions into RISC86), and trust me it took them a really long time to get the designs to work (K6 while not a bad processor wasn't exactly a champion).
Oh, yeah another thing... AMD doesn't make north+south bridges as Intel do, and while te original Athlon platform was unstable I betcha a PIII in an CC820 motherboard with SDRAM is FAR FAR worse...
Actually, that's the reason behind simulators, Intel did that with the IA-64 but only to some parties. AMD actually wants to do three things, a) test the architecture and develop compile optimizations, b)gauge the response of the community towards the processor, c)Piss off Intel:)
Seems to me this "report" serves more to justify M$ inocence than to provide actual figures into how many apps are available. The problem with it is the way it presents the facts, stating that there are no entry barriers for other OSes, but it misses several important points.
Drivers: Not an application per se but when you consider that in the Windows world most drivers come with tuning applications attached, whereas in the Alternative OS world if you have a binary driver you have to thank god six times in Swahili while you dance around a candle or something...
Office Applications: Oh yeah, we've got StarOffice, Abi, KOffice and their ilk, but since M$ changes file formats every three days you're facing incompatible/unformatteable files.
Vertical Applications: He's not going to find those on Amazon for sure but there are at least 70.000 custom vertical apps developed around the world. The cost of migrating those pieces of often klunky and badly documented code? Better not talk about it.
Games: The turning point of the Windows world, right now Alternative OS users are for the better part left out in the cold, it's changing but slowly. And if I count the number of games released every month around the world for Windows PC's only (Including stuff like card and hunting cames) you have around 100 games out per month.
Multimedia: Windows have at least 8 media players, 10 DVD players and at least 100 different media utilities not including music or specific software like codecs. Where are those for BeOS or OpenBSD?
Only there in those categories, you've got a 50:1 relationship between Windows apps and other OS apps.
He also talks about M$ not enforcing monopolistic pricing... WTF? does he at least compare prices in the Windows world?
For Office 2000 Professional you pay $240, for Lotus SmartSuite Millenium you pay $89 for the same exact features (And no annoying talking clip), for Corel is around $120... so in average M$ charges 200% more than their competition, and why do they remain the first choice? FUD, strong arming of OEM's and a lock in file formats.
And let's not talk about server software... sheesh...
In the end what could've been an interesting report (Despite the pro M$ bias) ends up being a big FUD spitting piece. It might raise some interesting points like the real availability of shelf windows apps but the narrow minded view enforced by the author really kills all chances of enlightenment or at least food for thought.
Yes, one problem solved... But... Can it take down a Flock of Seagulls(TM)? You know those critters are hard to hit (They've been a moving target since the early 80's)and they tend to sing nasty space age love songs which can interfere with the lasers.
I've worked (and keep working:) in telecoms all my life and after reading most of the comments I have some comments.
Are you sure that they are capping bandwidth or are they congested? In the telco industry to oversubscribe (I've seen rates from 1/2 to 1/512) a service is a common practice (You usually take into account the bandwidth you have and assign a worst case scenario in which a percentage of all subscribers transmit and receive simultaneously). The bad part is that most network engineers don't take several factors into account like exactly what applications are being used. It's a lot different to oversubscribe if the final app is web browsing (Where you know that most users go to only five or six pages and you have proxies around) or Napster/FTP (Where you're going to have your bandwidth maxed for periods of time).
Another issue, how good is their connection to the Internet backbone? How many NAP points do they have? Usually having one huge pipe to one provider (as in bbn, uunet or digex) is not enough, you need two or three in order to have some traffic balancing. Also most IEX points are congested and unfortunately most of the content seems to be placed in 2-3 major providers.
So IMNHO I think this is a sign of poor network design as much as malice from their marketing dept.
Exactly, so why is Moody (A guy with an obvious pro MS bias stated in all he has written previously)telling that because Bugtraq lists 122 bugs Linux is inferior to NT?
Most NT bugs on display are real showstoppers, and mostly can be classified into the System DoS or file-read/gain-root varieties. Linux bugs OTOH tend to be mostly into the Service DoS category. Also it should be pointed out that if there's a buffer overflow and root compromise in for example gpm or XFree86 that's not a Linux bug.
In the strict sense Linux is only the kernel nothing else...
Another solution for your comms problems might be getting an INMARSAT ( www.inmarsat.org) certified phone (Such as the sets made by Nera or NEC) that can give you up to 64Kbps links. They connect to your computer through the serial port so they are Linux friendly and they can be used as voice or data devices (Although not simultaneously)and they work all around the globe provided that you have line of sight (Not that hard to get BTW)
The only cons are: -Expensive: expect a set to be around $2000 and the service isn't cheap either. -Bulky: The sets are roughly the same size as the notebook you'll be carrying and equally heavy (Around 5 lbs.)
Worth a check if you're so desperate. If you can live with less speed and spotty coverage (With roaming charges) go for a GSM-900 phone, or even a Nokia Communicator.
What I'm wondering is, is this going to turn out to be another "exploit the GPL, violate GPL, IPO it for all they can squeeze, then get out" scenario. A certain Linux distributor comes to mind. . .
Two things there, first you can mix and match the GPL with other licenses including non-open ones (We had that discussion over the LinuxCE mailing list with some insightful responses, check the digest in http://www.linuxce.org ) as long as you satisfy all requirements, for the GPL it is to release the source with the binaries of all GPL derived work.
The second thing comes not from Caldera but from harware manufacturers themselves, in a world where profit margins are reducing each day, IP (Intellectual Property in case anyone didn't get it:) is the only asset a corporation has to sustain a certain market lead and recover the investment through licensing fees, most of those companies won't be pleased if their IP and trade secrets are published in source code form. I know most of us wouldn't take advantadge of a situation like that, but struggling companies would because when you're hungry (For success, money or food) good intentions don't matter...
Caldera (Which I don't use, I'm a happy SuSE user) has been good to the community just like Red Hat, Cygnus, Terra Soft or SuSE have been... only problem here is that they need to be profitable or else they won't exist, and THEY ARE a necessity in order to get the application base to grow with quality products... it had to happen, someday the cool idealism of the Linux community had to face the cold face of corporate consumerism, and until now the results have been positive even with the presence of such sour grapes like LinuxOne...
Function vectors!=self modifying code but code that patches on the fly (i.e. TCP/IP stacks for non reentrant OSes). That would also exclude DLL's and other dynamically linked libraries.
BTW. What happens when two threads executing on two different processors try to optimize the same piece of code? Without proper synchronisation mechanisms (such as write-through or cache snoops) this is very dangerous. Work-arounds, suche as semaphore will most probably cause a performace penalty.
You have to decide what's mode important, in high speed CPUs where the CPU goes 10x as fast as the memory bus write throughs are a bigger performance hits than semaphores, in massively parallel architectures it could be even worse because another processor might be locking access to the RAM or I/O. OTOH, semaphores can be a performance hit if htey're not persistent. That brings you to the following problem, do you run a single kernel image that controls all processes or do you run several kernels with interprocess communication between the kernel images and the threads?
For example the INMOS Transputer solved the problem through interprocess communication using tuple-spaces in OCCAM, some HA *nixes use semaphoring between multiple kernels, so you have a performance hit which is offset by the more robust setup. And.... BTW, IMHO SMP without parallel I/O and storage offsets the benefits of having several processors except for computational intensive operations that don't depend heavily on stored data sets.
Coping with Asymmetrical Multiprocessing? Wow, I haven't thought about this in a long while... there's only one answer there... sepparate instruction/data stores and a shared memory space for communications... either that or forget about the whole thing:)
All systems are inherently inefficient, what you do to make that inefficiency not noticed is what counts.
Nevertheless I'd like to point out that JIT is based a somewhat dangerous technique: a program that alters (its own) code. I believe this technique was used in the eighties to scare off hackers by making code incomprehensible and hard to disassemble until the program was actually running. Also (even on a good old 6502 processor) it's possible to make some speed improvents peeking and poking into the code you're actually executing.
Not at all true... self modifying code is still around not only on JIT compilers but in interpreted languages and even device drivers (Or how do you think some device drivers adjust in real time to hardware changes?). And it's not a dangerous technique in itself, but (As usual) when executed improperly it can be a wild beast (As most people discovered when thy installed their first Cyrix and AMD 486 processors and discovered most x86 code wouldn't run properly because of cache integrity issues)
When compilers for Microcomputers got faster and most processor architectures (known as Harvard architecture, I believe) explicitly require a division of RW and RO memory segments, self-modifying code was abandoned...
First of all, Harvard machines require separate instruction and data pipelines and memory spaces, and 99% of the CPU's on the market (General, Embedded, etc.) are Von Neumann machines which use a single space to store instruction and data. The thing is, newer CPU's (Even if they're Von Neumann desings)have separate caches for data and instructions and most self modifying code violates the cache integrity (See above) because the code is modified in the cache but never stored back into memory (Unless you use a write-through cache design which is impractical from the performace point of view).
BUT (And that's a big BUT:) if your JIT VM forces periodic cache writes to RAM (Not every time but enough times to ensure some sort of coherence if the cache and RAM are out of sync) you lose a bit of performance but gain a stabler setup.
Disassemblers for JIT won't be as complex as a JIT assembler just because when you disassemble a piece of code you treat is under the black box principle (What goes in and what goes out) in order to derive the fundamental principles and algorithms, which can be implemented in 1E999 different ways, so if you disassembled the less optimal morph, bad luck... disassemble another test run and see if you get a better implementation of the algorithm. Or you could use the aforementioned principle and implement your own algorithm.
JIT code/compilers, BTW are also being tested to produce self modifying chips (ASIC's and FPGA's) under VHDL/Verilog using also NNets or Gen. Algorithms to obtaing a first implementation and then using the JIT to optimize it...
Really interesting stuff (The Transmeta Crusoe) and my bet is that soon other companies will follow altough not for the same reasons as Transmeta:)
After having my HP-Omnigo die a couple of years ago I wanted to buy a PDA... I was looking at the available choices back then:
-PSION: No retailers to be seen, would've been my choice based on their OS alone. -Apple Newton: Loved it, hated it's price, could've carried it on my bag with no problems, damn you Apple for letting it die... -Palm: WHAT??? Only 512K of RAM? WTF? Graffitti? (It was included on the Omnigo too so I was comfortable with it) -WindowsCE: Hated them, but had better hardware and a lot of potential. I even admit that WinCE wasn't too bad compared to Windoze...
So considering my primary uses (Note Taker, PIM and Terminal -- YES!!! You can program a cisco router with an H/PC and still look totally k3wl in the process) I decided to buy a WinCE machine (A cassiopeia A-20) until it died of abuse last year, then the Palm was a better choice (PSION wasn't anywhere to be seen in the retail stores) but damn its price... it was excessive... if Handspring would've been out then... things could've been different.
have no idea what WinCE devices are like compared to Palms on a hardware level, but I would bet that Palms are more reliable
Not a matter of reliability but of horsepower... WinCE Machines have better hardware because they have to run that bloated piece of crap called WinCE at a decent speed... most have >60MHz RISC processors (MIPS R3000/4000 variants, ARM?StrongARM and Hitachi SHx), color displays and lots of nifty stuff... only problem??? The OS:(
That's why I'm working to port Linux to my Philips Nino 210 because the HW is beautyful and strong and the posibilities of running a decent OS on it (And be able to synch it with Linux) are great...
And Linux on WinCE machines has more potential than on the Dreamcast... just ask all the WinCE companies that have quit the market recently:)
Mr. Katz portrays himself as an outcast (This in a public forum where most participants describe themeselves as outcasts) because he uses a mainstream application. If he thinks the amount of bad rep he has been getting of late comes from using MS Word, then try Lotus WordPro or Corel WordPerfect and see what happens...
It's not a matter of not being in the "in" crowd, is a matter of articulation. You can't expect a 100% approval rate when you voice your opinions out loud even if you're qualified as an expert. But what worries me is the fact that even though I don't usually agree with his points of view I sit back and think about it... Perhaps life taught me to look at criticism as a constructive thing no matter how much bile is thrown in it, but I can see that Mr. Katz has no regard whatsoever about other people's opinions.
So a guy told you to go fsck and die? Good for him, he's the one making an ass of itself, he could turn off reading your articles, but he choose not to and he shouldn't complain. But I find amusing that you try to grab some of the spotlight by making the "threat" public and write a series of articles about it... not very convincing IMHO.
The world doesn't revolve around your ego John (Thank goodness) but this public forum has been doing it on a regular basis. I've never wrote anything about your articles but sheesh, this time it looks more like a desperate call for airtime than a sincere attempt to discuss your problems.
If you need inspiration to write an article then take a stroll on the park or voice a worthy cause, don't take a cheap shot to cover a deadline. There are better subjects to discuss than a 1337 h4x0r d00d telling you to go away...
ZoeSch
Re:A good argument for moderating articles, this i
on
Planet Gattaca
·
· Score: 1
I'm going to say only one thing...
The universe has no morals, only laws that are not even remotely understood, we (Humans) are the ones that happily attach any morality to whatever happens and we can't explain.
I for once (I have a very strong reason for this) will like to see if over time someone clones/makes a human being and it ends up with feelings and emotions (Please don't give me any of the psychological crap about why we feel beforehand) then I believe most questions will already be answered.
In two millenia of sience and religion we still make mistakes and some can be costly for the people that are just standing by unaware of what we do. And that happens because we THINK we know some answers when we really don't know a damn thing.
Weird as it might sound I believe what Intel tried to patent was a way to do a distributed network between a PC an some devices (Computers, toasters, coconut monkeys, etc.. That might be a little more coherent considering Intel's latest shopping spree on appliance device building blocks (With their PC centric view).
I believe so because any person who has done serious work with distributed computing knows that hub's aren't supposed to originate transactions but to route them (As in receiving a start signal and sending the appropiate chunk of data or passing the request to another node).
BTW, SETI@Home is IMHO more a client-server oriented application than a real distributed network because all data is still centralized.
Thx for clarifying it... I thought the licencse clearly stated that downloads are free but that media redistribution required licensing fees... oh well...
As for downloading WP, nah I'll pass... I use SOffice (I paid for it, I mean $35 bucks to let those guys eat for a week... it was worth it:)
Yeah the article was good for clarifying points to the zealots out there... I can't run FreeBSD because most of my hardware isn't properly supported but OTOH I have to admit it kicks ass... Maybe one day I'll switch from Linux to OpenBSD at home...
The only thing lacking in the article was pointing out the amount of BSD derivative tools that have been ported to other OSes including Linux...
Hrrrmf... why wasting flame wars over Unix/Linux/xBSD/BeOS users?
/me thinks about playing a while with his Apple//c
The problem there is convention, Moto and Apple talk about different things when they mention G4, for Apple is their Line of PPC-740 machines. For Moto is a different version of the PPC-740 whith a MCM packaging (The 740 is a direct descendant of the Power4)
Damn! yeah sue me guys I loved computer architecture @ college:)
I guess you never attended a Computer Architecture class???
Polluted RISC??? err didn't you mean PA-RISC? hacked together? Err, again... look at the Pentium class processors for clues about how to keep hacking the same old architecture until it looks like they put it together glueing pieces of Cadence designs with magic glue...
Array of small RISC chips==Very huge board...
Come on, I don't have a G3/G4 but the architecture is CLEARLY superior to the x86 architecture...
Slashdot Mirror
Go to one of the Registries...
/20 but you need to fulfill certain prerequisites
If you're in the US:
www.arin.net
If you're in Asia or Australia
www.apnic.net
If you're in Europe
www.ripe.net
Be sure to read the guidelines... the least you can aquire is a
ZoeSch
Well if he cost him Oregano, he should've gone for parsley...
Sorry couldn't resist
Hmmm... Gore 182... now that's a Punk band that'll bore you to death...
While the article doesn't come out and say it, I'm guessing there will be serious total bandwidth limitations- since there are no tracking antennas, and since this is referred to as part of a satellite TV service, it must be from (and probably to) a satellite in geosync orbit. All traffic is probably going to just one satellite. Once a lot of people sign up for this- popularity will kill it- since I'm sure that they can sign people up to long term contracts faster than they can put more transponders up.
You obviously don't know what you're talking about... The average transponder Bandwidth on the newer GEO satellites is 72 MHz, using 8PSK modulation with FEC 2/3 Viterbi or 216/205 Reed-Solomon encoding gives you roughly 90 Mbps to use... most High-Speed VSAT (Very Small Aperture Terminal) systems allocate bandwidth in two chunks, one outbound carrier (From the hub) and several inbound carriers (From the station). Those are TDM/TDMA slotted carriers where you should contend for available slots (Slotted-ALOHA) which usually have an efficiency of 60-65%. If you split the transponder in two you have 45 Mbps downstream (Subject to a 85% efficiency limit)and several (Usually 2-4 Mbps) upstream carriers which serve a group of VSAT's.
Since you're using a multiple allocation mechanism you can effectively guarantee at least some upstream traffic per station using streams (Pre-assigned slots). Have too many users in your service, get another transponder, it's not expensive to implement since these kinds of systems have independent modems.
Plus there is the fact that anything beamed to you is probably also being beamed to everyone in a multiple state area around you. I sure hope they have some *strong* encryption built in. They ought to be doing something better than DES, no?
Oh brother... can you tell me how are you going to demodulate the signal you receive from the satellite? (QPSK and 8PSK are not easy to implement with a breadboard you know?, not taking into account the fact that you'll need to build a fairly big decoder for viterbi or Reed-Solomon plus the encryption on the timeslots). You will receive what's intended for you and only you... you will be expecting some downstream timeslots and your IDU (Indoor Unit) will only decode those.
I hate to say it, but the microsoft backed Teledesic system is a much better system (many satellites in low earth orbit- kinda like a cellular system) the satellites are closer and cover smaller areas, so the amount of bandwidth/satellite is much smaller.
Of course you don't mention that these satellites are smaller, have less transponders, drift like mad and have poor signal footprints (The amount of power you receive down in earth). Cellular systems only work for small amounts of data (Look at how painful it has been to get GPRS working, not to mention 3G). Of course, fixed cell systems like LMDS do provide better bandwidth but they use modified versions of ATM, Frame Relay or PPP in order to dynamically allocate bandwidth and those aren't good when you have 500 users downloading pr0n at the same time.
If you need more proof that these kinds of systems work (Only for mail and browsing, trust me) look for DVB (Digital Video Broadcast) services, or check out systems from Hughes Network Systems or STMI.
ZoeSch
Only in the cities they serve. Time Warner's Roadrunner is the other monopoly. Except that DSL is becoming more and more available. Which is competition for both TW and ATT.
If that's not monopoly on cable services I don't know what it is.
So they really don't have a "monopoly in the US". Just in certain cities.
Actually the US has a duopoly which is indeed (In economic and service terms) far worse than a monopoly (Why open us, we have competition from the other guys).
ZoeSch
Judging on the same basis intel should be sued by Zilog because of the same argument. As a fact... the only thing AMD and Cyrix haven't been able to perfectly emulate have been the many design flaws (FDIV bug, instability at more than 1 GHz, an MTH that doesn't work, etc. etc.). As a matter of fact, AMD and Cyrix designs are derivative compatible works that the only thing they share in common with Intel is the Instruction Set (And not even so because AMD uses a RISC engine that decodes and translates the IA-32 instructions into RISC86), and trust me it took them a really long time to get the designs to work (K6 while not a bad processor wasn't exactly a champion).
Oh, yeah another thing... AMD doesn't make north+south bridges as Intel do, and while te original Athlon platform was unstable I betcha a PIII in an CC820 motherboard with SDRAM is FAR FAR worse...
ZoeSch
Actually, that's the reason behind simulators, Intel did that with the IA-64 but only to some parties. AMD actually wants to do three things, a) test the architecture and develop compile optimizations, b)gauge the response of the community towards the processor, c)Piss off Intel :)
:)
Imagine if it works
ZoeSch
Seems to me this "report" serves more to justify M$ inocence than to provide actual figures into how many apps are available. The problem with it is the way it presents the facts, stating that there are no entry barriers for other OSes, but it misses several important points.
Drivers: Not an application per se but when you consider that in the Windows world most drivers come with tuning applications attached, whereas in the Alternative OS world if you have a binary driver you have to thank god six times in Swahili while you dance around a candle or something...
Office Applications: Oh yeah, we've got StarOffice, Abi, KOffice and their ilk, but since M$ changes file formats every three days you're facing incompatible/unformatteable files.
Vertical Applications: He's not going to find those on Amazon for sure but there are at least 70.000 custom vertical apps developed around the world. The cost of migrating those pieces of often klunky and badly documented code? Better not talk about it.
Games: The turning point of the Windows world, right now Alternative OS users are for the better part left out in the cold, it's changing but slowly. And if I count the number of games released every month around the world for Windows PC's only (Including stuff like card and hunting cames) you have around 100 games out per month.
Multimedia: Windows have at least 8 media players, 10 DVD players and at least 100 different media utilities not including music or specific software like codecs. Where are those for BeOS or OpenBSD?
Only there in those categories, you've got a 50:1 relationship between Windows apps and other OS apps.
He also talks about M$ not enforcing monopolistic pricing... WTF? does he at least compare prices in the Windows world?
For Office 2000 Professional you pay $240, for Lotus SmartSuite Millenium you pay $89 for the same exact features (And no annoying talking clip), for Corel is around $120... so in average M$ charges 200% more than their competition, and why do they remain the first choice? FUD, strong arming of OEM's and a lock in file formats.
And let's not talk about server software... sheesh...
In the end what could've been an interesting report (Despite the pro M$ bias) ends up being a big FUD spitting piece. It might raise some interesting points like the real availability of shelf windows apps but the narrow minded view enforced by the author really kills all chances of enlightenment or at least food for thought.
ZoeSch
Yes, one problem solved... But... Can it take down a Flock of Seagulls(TM)? You know those critters are hard to hit (They've been a moving target since the early 80's)and they tend to sing nasty space age love songs which can interfere with the lasers.
I've worked (and keep working :) in telecoms all my life and after reading most of the comments I have some comments.
Are you sure that they are capping bandwidth or are they congested? In the telco industry to oversubscribe (I've seen rates from 1/2 to 1/512) a service is a common practice (You usually take into account the bandwidth you have and assign a worst case scenario in which a percentage of all subscribers transmit and receive simultaneously). The bad part is that most network engineers don't take several factors into account like exactly what applications are being used. It's a lot different to oversubscribe if the final app is web browsing (Where you know that most users go to only five or six pages and you have proxies around) or Napster/FTP (Where you're going to have your bandwidth maxed for periods of time).
Another issue, how good is their connection to the Internet backbone? How many NAP points do they have? Usually having one huge pipe to one provider (as in bbn, uunet or digex) is not enough, you need two or three in order to have some traffic balancing. Also most IEX points are congested and unfortunately most of the content seems to be placed in 2-3 major providers.
So IMNHO I think this is a sign of poor network design as much as malice from their marketing dept.
ZoeSch
Exactly, so why is Moody (A guy with an obvious pro MS bias stated in all he has written previously)telling that because Bugtraq lists 122 bugs Linux is inferior to NT?
Most NT bugs on display are real showstoppers, and mostly can be classified into the System DoS or file-read/gain-root varieties. Linux bugs OTOH tend to be mostly into the Service DoS category. Also it should be pointed out that if there's a buffer overflow and root compromise in for example gpm or XFree86 that's not a Linux bug.
In the strict sense Linux is only the kernel nothing else...
ZoeSch
Another solution for your comms problems might be getting an INMARSAT ( www.inmarsat.org) certified phone (Such as the sets made by Nera or NEC) that can give you up to 64Kbps links. They connect to your computer through the serial port so they are Linux friendly and they can be used as voice or data devices (Although not simultaneously)and they work all around the globe provided that you have line of sight (Not that hard to get BTW)
The only cons are:
-Expensive: expect a set to be around $2000 and the service isn't cheap either.
-Bulky: The sets are roughly the same size as the notebook you'll be carrying and equally heavy (Around 5 lbs.)
Worth a check if you're so desperate. If you can live with less speed and spotty coverage (With roaming charges) go for a GSM-900 phone, or even a Nokia Communicator.
ZoeSch
What I'm wondering is, is this going to turn out to be another "exploit the GPL, violate GPL, IPO it for all they can squeeze, then get out" scenario. A certain Linux distributor comes to mind. . .
:) is the only asset a corporation has to sustain a certain market lead and recover the investment through licensing fees, most of those companies won't be pleased if their IP and trade secrets are published in source code form. I know most of us wouldn't take advantadge of a situation like that, but struggling companies would because when you're hungry (For success, money or food) good intentions don't matter...
Two things there, first you can mix and match the GPL with other licenses including non-open ones (We had that discussion over the LinuxCE mailing list with some insightful responses, check the digest in http://www.linuxce.org ) as long as you satisfy all requirements, for the GPL it is to release the source with the binaries of all GPL derived work.
The second thing comes not from Caldera but from harware manufacturers themselves, in a world where profit margins are reducing each day, IP (Intellectual Property in case anyone didn't get it
Caldera (Which I don't use, I'm a happy SuSE user) has been good to the community just like Red Hat, Cygnus, Terra Soft or SuSE have been... only problem here is that they need to be profitable or else they won't exist, and THEY ARE a necessity in order to get the application base to grow with quality products... it had to happen, someday the cool idealism of the Linux community had to face the cold face of corporate consumerism, and until now the results have been positive even with the presence of such sour grapes like LinuxOne...
ZoeSch
Function vectors!=self modifying code but code that patches on the fly (i.e. TCP/IP stacks for non reentrant OSes). That would also exclude DLL's and other dynamically linked libraries.
:)
BTW. What happens when two threads executing on two different processors try to optimize the same piece of code? Without proper synchronisation mechanisms (such as write-through or cache snoops) this is very dangerous. Work-arounds, suche as semaphore will most probably cause a performace penalty.
You have to decide what's mode important, in high speed CPUs where the CPU goes 10x as fast as the memory bus write throughs are a bigger performance hits than semaphores, in massively parallel architectures it could be even worse because another processor might be locking access to the RAM or I/O. OTOH, semaphores can be a performance hit if htey're not persistent. That brings you to the following problem, do you run a single kernel image that controls all processes or do you run several kernels with interprocess communication between the kernel images and the threads?
For example the INMOS Transputer solved the problem through interprocess communication using tuple-spaces in OCCAM, some HA *nixes use semaphoring between multiple kernels, so you have a performance hit which is offset by the more robust setup. And.... BTW, IMHO SMP without parallel I/O and storage offsets the benefits of having several processors except for computational intensive operations that don't depend heavily on stored data sets.
Coping with Asymmetrical Multiprocessing? Wow, I haven't thought about this in a long while... there's only one answer there... sepparate instruction/data stores and a shared memory space for communications... either that or forget about the whole thing
All systems are inherently inefficient, what you do to make that inefficiency not noticed is what counts.
ZoeSch
Nevertheless I'd like to point out that JIT is based a somewhat dangerous technique: a program that alters (its own) code. I believe this technique was used in the eighties to scare off hackers by making code incomprehensible and hard to disassemble until the program was actually running. Also (even on a good old 6502 processor) it's possible to make some speed improvents peeking and poking into the code you're actually executing.
:) if your JIT VM forces periodic cache writes to RAM (Not every time but enough times to ensure some sort of coherence if the cache and RAM are out of sync) you lose a bit of performance but gain a stabler setup.
:)
Not at all true... self modifying code is still around not only on JIT compilers but in interpreted languages and even device drivers (Or how do you think some device drivers adjust in real time to hardware changes?). And it's not a dangerous technique in itself, but (As usual) when executed improperly it can be a wild beast (As most people discovered when thy installed their first Cyrix and AMD 486 processors and discovered most x86 code wouldn't run properly because of cache integrity issues)
When compilers for Microcomputers got faster and most processor architectures (known as Harvard architecture, I believe) explicitly require a division of RW and RO memory segments, self-modifying code was abandoned...
First of all, Harvard machines require separate instruction and data pipelines and memory spaces, and 99% of the CPU's on the market (General, Embedded, etc.) are Von Neumann machines which use a single space to store instruction and data. The thing is, newer CPU's (Even if they're Von Neumann desings)have separate caches for data and instructions and most self modifying code violates the cache integrity (See above) because the code is modified in the cache but never stored back into memory (Unless you use a write-through cache design which is impractical from the performace point of view).
BUT (And that's a big BUT
Disassemblers for JIT won't be as complex as a JIT assembler just because when you disassemble a piece of code you treat is under the black box principle (What goes in and what goes out) in order to derive the fundamental principles and algorithms, which can be implemented in 1E999 different ways, so if you disassembled the less optimal morph, bad luck... disassemble another test run and see if you get a better implementation of the algorithm. Or you could use the aforementioned principle and implement your own algorithm.
JIT code/compilers, BTW are also being tested to produce self modifying chips (ASIC's and FPGA's) under VHDL/Verilog using also NNets or Gen. Algorithms to obtaing a first implementation and then using the JIT to optimize it...
Really interesting stuff (The Transmeta Crusoe) and my bet is that soon other companies will follow altough not for the same reasons as Transmeta
ZoeSch
BZZT Wrong Answer!!!
:(
:)
After having my HP-Omnigo die a couple of years ago I wanted to buy a PDA... I was looking at the available choices back then:
-PSION: No retailers to be seen, would've been my choice based on their OS alone.
-Apple Newton: Loved it, hated it's price, could've carried it on my bag with no problems, damn you Apple for letting it die...
-Palm: WHAT??? Only 512K of RAM? WTF? Graffitti? (It was included on the Omnigo too so I was comfortable with it)
-WindowsCE: Hated them, but had better hardware and a lot of potential. I even admit that WinCE wasn't too bad compared to Windoze...
So considering my primary uses (Note Taker, PIM and Terminal -- YES!!! You can program a cisco router with an H/PC and still look totally k3wl in the process) I decided to buy a WinCE machine (A cassiopeia A-20) until it died of abuse last year, then the Palm was a better choice (PSION wasn't anywhere to be seen in the retail stores) but damn its price... it was excessive... if Handspring would've been out then... things could've been different.
have no idea what WinCE devices are like compared to Palms on a hardware level, but I would bet that Palms are more reliable
Not a matter of reliability but of horsepower... WinCE Machines have better hardware because they have to run that bloated piece of crap called WinCE at a decent speed... most have >60MHz RISC processors (MIPS R3000/4000 variants, ARM?StrongARM and Hitachi SHx), color displays and lots of nifty stuff... only problem??? The OS
That's why I'm working to port Linux to my Philips Nino 210 because the HW is beautyful and strong and the posibilities of running a decent OS on it (And be able to synch it with Linux) are great...
And Linux on WinCE machines has more potential than on the Dreamcast... just ask all the WinCE companies that have quit the market recently
ZoeSch
True egolatrism nonetheless...
Mr. Katz portrays himself as an outcast (This in a public forum where most participants describe themeselves as outcasts) because he uses a mainstream application. If he thinks the amount of bad rep he has been getting of late comes from using MS Word, then try Lotus WordPro or Corel WordPerfect and see what happens...
It's not a matter of not being in the "in" crowd, is a matter of articulation. You can't expect a 100% approval rate when you voice your opinions out loud even if you're qualified as an expert. But what worries me is the fact that even though I don't usually agree with his points of view I sit back and think about it... Perhaps life taught me to look at criticism as a constructive thing no matter how much bile is thrown in it, but I can see that Mr. Katz has no regard whatsoever about other people's opinions.
So a guy told you to go fsck and die? Good for him, he's the one making an ass of itself, he could turn off reading your articles, but he choose not to and he shouldn't complain. But I find amusing that you try to grab some of the spotlight by making the "threat" public and write a series of articles about it... not very convincing IMHO.
The world doesn't revolve around your ego John (Thank goodness) but this public forum has been doing it on a regular basis. I've never wrote anything about your articles but sheesh, this time it looks more like a desperate call for airtime than a sincere attempt to discuss your problems.
If you need inspiration to write an article then take a stroll on the park or voice a worthy cause, don't take a cheap shot to cover a deadline. There are better subjects to discuss than a 1337 h4x0r d00d telling you to go away...
ZoeSch
I'm going to say only one thing...
The universe has no morals, only laws that are not even remotely understood, we (Humans) are the ones that happily attach any morality to whatever happens and we can't explain.
I for once (I have a very strong reason for this) will like to see if over time someone clones/makes a human being and it ends up with feelings and emotions (Please don't give me any of the psychological crap about why we feel beforehand) then I believe most questions will already be answered.
In two millenia of sience and religion we still make mistakes and some can be costly for the people that are just standing by unaware of what we do. And that happens because we THINK we know some answers when we really don't know a damn thing.
Weird as it might sound I believe what Intel tried to patent was a way to do a distributed network between a PC an some devices (Computers, toasters, coconut monkeys, etc.. That might be a little more coherent considering Intel's latest shopping spree on appliance device building blocks (With their PC centric view).
I believe so because any person who has done serious work with distributed computing knows that hub's aren't supposed to originate transactions but to route them (As in receiving a start signal and sending the appropiate chunk of data or passing the request to another node).
BTW, SETI@Home is IMHO more a client-server oriented application than a real distributed network because all data is still centralized.
ZoeSch
Thx for clarifying it... I thought the licencse clearly stated that downloads are free but that media redistribution required licensing fees... oh well...
:)
As for downloading WP, nah I'll pass... I use SOffice (I paid for it, I mean $35 bucks to let those guys eat for a week... it was worth it
ZoeSch
He's selling software like WordPerfect, Perfect Backup and other non GPL products for ridiculous amounts of money like $5 or summat.
Last time I checked WP for Linux was around $60 for commercial users...
Sounds like a god damned pirate that's trying to get some free publicity...
ZoeSch
Yeah the article was good for clarifying points to the zealots out there... I can't run FreeBSD because most of my hardware isn't properly supported but OTOH I have to admit it kicks ass... Maybe one day I'll switch from Linux to OpenBSD at home...
//c
The only thing lacking in the article was pointing out the amount of BSD derivative tools that have been ported to other OSes including Linux...
Hrrrmf... why wasting flame wars over Unix/Linux/xBSD/BeOS users?
/me thinks about playing a while with his Apple
ZoeSch
Flame on bro!
:)
The problem there is convention, Moto and Apple talk about different things when they mention G4, for Apple is their Line of PPC-740 machines. For Moto is a different version of the PPC-740 whith a MCM packaging (The 740 is a direct descendant of the Power4)
Damn! yeah sue me guys I loved computer architecture @ college
ZoeSch
I guess you never attended a Computer Architecture class???
Polluted RISC??? err didn't you mean PA-RISC? hacked together? Err, again... look at the Pentium class processors for clues about how to keep hacking the same old architecture until it looks like they put it together glueing pieces of Cadence designs with magic glue...
Array of small RISC chips==Very huge board...
Come on, I don't have a G3/G4 but the architecture is CLEARLY superior to the x86 architecture...
ZoeSch
Considering that Laibach is more than an Industrial band :)
The idea of the Kunst is REALY old... they even had a tour of the Kunst a couple of years ago... yawn!!!
OTOH I like Laibach music and even their ideology... Nevertheless the whole idea of a virtual nation is laughable...