Re:What I've always wondered ...
on
Intel's Big Chip
·
· Score: 1
..... and the answer is compatability with existing pentiom instruction set.
Q. Why is th epetium chip so much bigger then sparc/powerPC.
A. compatability with 386 instruction set.
Q. Why was the 386 32 bit architecture so awful?
A. Compatability with 16 bit 286 architecture.
Q. Why was the 286 instruction set so ugly?
A. Because is was compatable with 8 bit 8080 instruction set.
Q. Where did that wierd 8080 register architecture come from.
A. They copied it from the 4 bit 4040 chip.
So there you have it; Intel have spent five years not delivering a chip that has twice the transistor count and half the processing power of its nearest rival so it can emulate a 4 bit processor in hardware.
Q. Why?
A. The worlds best selling operating system was written for an 8080 chip, and is now so complicated that they cannot port the code to any other architecture.
Most people seem to be missing the main finding from the test:: IBM's JRE VM runs nearly as well as the compiled java. So if you want better performance get a better JVM? (( I am sure that Suns HotSpot would post similar results )).
Just out of curiousity I compared the Java "primes" program with an almost identical C program on a Sun box and got the following results:-
This is a bit like comaparing a fleet of ford escorts with a PeterBuilt.
You can buy a lot of low end fords for the same price as a 50 ton truck, but all the family saloons put together will not transport 50 tons.
The current mainframe technoligy "IBMs zSeries" is extremely powerfull and so highly configureable that each "mainframe" is effectively a "one-off" with differnt numbers of CPUs, memory size, channels(IO buses) and interconnects, it is very diffcult to compare with other architectures.
Some years ago IBM posted the higest TCP/A benchmark ever using a cluster of mainframes and the venerable IMS database software, this configuration was some 5 times faster than the previous fastest benchmark using Oracle and DEC alpha cluster.
Given that the alpha chip hasn't moved on much since then and that a 1gz+ pentium is roughly equvalent to a compaq alpha chip this differential still holds.
On the other hand you don't really get much of a zSeries box for $1,000,000. But the power consumption and aircon required for 400 Intel processors has got to eat up some of the differential.
Berkley did not invent the internet, they did not even invent TCP/IP.
They (Bill Joy really) coded an implementation of the TCP/IP protocol which was already defined and implemented on several other systems.
They did add the standard sockets/inetd interface. And the TCP/IP stack as coded is the basis of nearly all current UNIX and all Windows implementations of TCP/IP.
However this happened largely because the code was free (as in beer), if Berkley had tried to charge for thier TCP/IP stack and patent thier sockets implmentation then SUN, Microsoft et all would probably have written thier own version rather than get into a contractual relationship for a fundamental part of thier systems.
This principal applies to almost any part of the internet as it now exits. Free and open software gets used because it is cheap, easy to improve and easy to standardise. Proprietry software is avoided because of expense, vendor lock in, difficulties in standards setting etc.etc.
As Mike Berniers Lee said "If we had charged for Mosaic nobody would have used it".
Right. There is no particular reason why 64 bit processors should be faster than 32 bit processors ( or indeed why 16 bit processors should be faster than 8 bit processors!).
However there is a definite corelation between applications which require fast processing and applications which require lots of in memory data. (Large databases and really good graphic games are two that spring to mind). So chip manufacturers have tended to increase speed and addressability in parallel.
But then again if you cannot run a decent desktop in 1GB of memory just call yourself "Bill Gates".
Most of the mid to high end non Windows servers now uses 64 bit chips. They are nice and they are fast, but very few ship with > 2GB of memory.
PS. earlier on there was a comment whinging about SUN using a "special" compiler to increase thier SPEC scores. The compiler was not that "special" it just optimised for the parrellel instruction capability of the UltraSparc III architecture and will be included in the next release of the Sun compiler. So if you want really fast applications on the Sparc III you will have to buy Sun's compiler along with the truley awful "Forte" IDE.
Granularity -- i.e. you accidentally deleted "xmascard.lst" how do you know which tape it was on and how long will it take to scan several tapes to find your 2K file?
How can you be sure the tape is readable. In my experience something like 5% of backup tapes cannot be read by any drive except the one that wrote the tape.
No standard format, no backward compatabilty, short market life for thevarioustecnoligies. You may be very smug that you have an "acme" format backup stored in a safe deposit box, but after your house is flooded (mentioning the f*r* word upsets some people) you discover "acme" drives are no longer sold anymore, you just have to hope someone is trying to unload a working drive on e-bay.
Expense -- the drives, the tapes etc. will all cost several times as much as a couple of external hard drives.
Sweden probably had the best shot at demacratic socialism, but they had the sense to leave volvo, saab, Nobel etc as private companies.
One of the major companies that is still in government ownership is the maker of "Absolut" vodka, so any proponent of state ownership can get blind drunk with a clear conscience.
What really surpises me is that anybody can still take the concept of Marxist Economics seriously. We are talking about a guy who organised his own personal finances so badly that one of his children died of malnutrition.
It was mentioned in the article that the court officere managed to access and update information stored on an IBM mainframe system.
This is not Windows, but an operating system that has taken security very seriously for well over twenty years.
Any organisation abd/or profesional that configures one of these machines badly enough to allow data to be accessed with a trivial hack is definately negligent in my opinion.
There is a standards commitee under the unberlla of the Open Systems Group.
Microsoft was a member of this group but I don't think they were very active. If I remeber correctly the OO academics ( Booch et all. were the driving force behind this one, and, Sun were the first to implement ).
As for COM ( or OLE, or DCOM or any of the other names this dogs breakfast has had). It hardly a standard as the interface varies with every release, and, there is no oppertunity for anyone but Redmond to provide input.
Architecture is definately the best tech job there is.
You get to play with all the new toys. (err.. Evaluate the latest technologies.)
You get taken to all the best restaraunts and bars by suppliers ( I mean establish fruitful working relationships with potential vendors).
Best of all you get to swan around, showing pretty slides with lots of boxes and look important.(... comunicate new directions in technoligies to the companies descision makers.)
And if you can make as a Consultant Architect you pick up lots and lots of money, and, move on before somebody has to put your wierd and wonderful ideas into practice. ((.. constantly bring your in depth experience to new projects and customers.)
PS. I hope no one out there works out my real identity, or I may have to go back to doing a real job!
Yep, its the humble oxytetracycline antibiotic (usually prescribed for step throat and othe rmonir infections).
I know this because every few years my wife gets an urge to see the Taj Mahal by moonlight.
This trip has been cancelled many times because of political urest, war, famine, and one year bubonic plague -- which was not such a big deal but we were advised not to travel if you were pregnant or likely to be pregnant as oxytetracacline has adverse effects on the developing feotus.
The "java platform" is not just experience and folk wosdom.
It is a standard well defined set of facilities (JVM, comiler, JAR files et. etc. ) which a java platform must support, plus, at a higer level of protocols for defining things like javabeans and EJBs. These are not vague fuzzy conventions: a platform either supports EJBs or it doesn't, an IDE can support java beans or it cannot.
This is very different from say "C" where you have the language and a set of library functions which are very precisely defined and thats it.
Programming C/C++ for OS/390 CICS has very little in common with coding MFC programs for windows apart from the language itself. Programming Java usings EJBs is pretty much the same for beans residing on OS/390 cics as it is for "COM+" NT based beans.
Atricle doesn't do wxWindows justice.
on
New Perl GUI
·
· Score: 2, Informative
Having played around with wxPython, I can honestly say I will never go back to Tk or any other GUI interface.
What the article doesn't empahsise enough is that this is a cross platform GUI. The same code will work on win32, nearly all *nixes, plus apple and some wierd paltforms to boot.
The other thing the article doesn't impress strongly enough is how good the underlying C++ wxWindows library is. This is C++ GUI Nirvana, OO from the bottom up, crossplatform, native look and feel, a really useful set of builtin widgets, plus, perhaps most importantly a really well though out and easily extandable class structure.
From now on I will use nothing else, I may even give up programming curses.
Not to mention some architectural gotchas unique to Windoze.
The NT kernal uses a message passing architecture, and, interupts are translated into messages on an internal queue, which, are then picked up by the appropraite aplication(s). This works fine (better than fine in fact) for a single processer running a GUI application.
However only one processor can be used to manage the actual message queues. In a multiprocessor machine NT splits the queue into 2 queues one for software generated interupts and one for hardware interupts. Processor 0 handles the software interupts and the higest numbered CPU handels the hardware interupts.
So it doesn't matter how many processors you add to your system, a sungle processor must handle all the IO and network interupts.
For a client and IO intensive application like a mail server this rapidly becomes the main bottleneck on performance.
Hence the pretty much standard recomendation of a two processor machine to handle mail. You can buy an eight processor machine but two processors will end up doing all the work and the other six will sit there twiddling thier virtual thumbs.
Both SUN and IBM are currently shipping chips that outperform the Itanium (which is very much sample ware) on every real world benchmark.
Furthermore the latest generation of the Power4 chip (Shipping in October) comes with two processers with a builtin shared L3 cache on a single chip!
In addittion it features several other speed enhancing innivations.
SUN are rummored to have something similar in mind for thier next generation Sparcs.
Basically INTEL are dying on the alter of "386" compatability. In trying to make a 64 bit chip compatable at instruction level to the worlds worse 16 bit instruction set INTEL have set themselves an impossable task.
Re:Where You Are Wrong
on
HP Buys Compaq
·
· Score: 2, Insightful
The reason Oracle (and any other producer of benchmarked software) port to the Alpha platform early is that it gives the highest benchmark scores.
Probably 50% of sales for the Alpha/Tru64 paltform are benchmark related. ( As opposed to 100% of NT alpha sales).
The reasons HPUX is one of the last ports to be made are:-
a. Most unix professionals have come to hate HP/UX.
b. Its actually three ports you need to make, one for 32 bit HP/UX 7 and earlier, one for 32 bit HP?UX HP/UX 8 and later and one for 64bit HP/UX 8 and later.
c. Nobody is buying new HP boxes they are slow, overpriced and have a very peculiar unix running on them. The market leaders are Sun with thier very basic very standard Solaris OS, or, IBM with thier very reliable easy to use full featured AIX OS -- both produce well made very fast boxes.
Re:Only one merger away from..
on
HP Buys Compaq
·
· Score: 1
Buying double glazing, timesharing an overpriced holidy apartment, or, even contributing to the Veterins Association wasn't what I had in mind when I installed my phone.
But will those b*st*rd* ( and b*tch*s) stop calling me, or, even get of the line --- NO.
Well most people aren't using computers just because they like hacking; they want there computer to send mail, write letters, surf the web, play quake & etc.
There is no reason why linux shouldn't be as easy to install as windows, other, than the general "it's cool to hack" ethos which means the linux community doesn't even see the need for point and click installtion.
If you want to see linux running on the average users desk top then there has to be some way of getting it there that doesn't involve reading a two inch thick manual and learning bash shell scripting.
Slashdot Mirror
Q. Why is th epetium chip so much bigger then sparc/powerPC.
A. compatability with 386 instruction set.
Q. Why was the 386 32 bit architecture so awful?
A. Compatability with 16 bit 286 architecture.
Q. Why was the 286 instruction set so ugly?
A. Because is was compatable with 8 bit 8080 instruction set.
Q. Where did that wierd 8080 register architecture come from.
A. They copied it from the 4 bit 4040 chip.
So there you have it; Intel have spent five years not delivering a chip that has twice the transistor count and half the processing power of its nearest rival so it can emulate a 4 bit processor in hardware.
Q. Why?
A. The worlds best selling operating system was written for an 8080 chip, and is now so complicated that they cannot port the code to any other architecture.
Most people seem to be missing the main finding from the test :: IBM's JRE VM runs nearly as well as the compiled java. So if you want better performance get a better JVM? (( I am sure that Suns HotSpot would post similar results )).
Just out of curiousity I compared the Java "primes" program with an almost identical C program on a Sun box and got the following results:-
Suns Java VM -- 31 seconds.
C program -- 23 seconds.
Optimesed C -- 17 seconds.
This is a bit like comaparing a fleet of ford escorts with a PeterBuilt.
You can buy a lot of low end fords for the same price as a 50 ton truck, but all the family saloons put together will not transport 50 tons.
The current mainframe technoligy "IBMs zSeries" is extremely powerfull and so highly configureable that each "mainframe" is effectively a "one-off" with differnt numbers of CPUs, memory size, channels(IO buses) and interconnects, it is very diffcult to compare with other architectures.
Some years ago IBM posted the higest TCP/A benchmark ever using a cluster of mainframes and the venerable IMS database software, this configuration was some 5 times faster than the previous fastest benchmark using Oracle and DEC alpha cluster.
Given that the alpha chip hasn't moved on much since then and that a 1gz+ pentium is roughly equvalent to a compaq alpha chip this differential still holds.
On the other hand you don't really get much of a zSeries box for $1,000,000. But the power consumption and aircon required for 400 Intel processors has got to eat up some of the differential.
Berkley did not invent the internet, they did not even invent TCP/IP.
They (Bill Joy really) coded an implementation of the TCP/IP protocol which was already defined and implemented on several other systems.
They did add the standard sockets/inetd interface. And the TCP/IP stack as coded is the basis of nearly all current UNIX and all Windows implementations of TCP/IP.
However this happened largely because the code was free (as in beer), if Berkley had tried to charge for thier TCP/IP stack and patent thier sockets implmentation then SUN, Microsoft et all would probably have written thier own version rather than get into a contractual relationship for a fundamental part of thier systems.
This principal applies to almost any part of the internet as it now exits. Free and open software gets used because it is cheap, easy to improve and easy to standardise. Proprietry software is avoided because of expense, vendor lock in, difficulties in standards setting etc.etc.
As Mike Berniers Lee said "If we had charged for Mosaic nobody would have used it".
We were refering to addressability here. I seem to remember that even the huble pdp-11 could handle 16 bit intigers!
Right. There is no particular reason why 64 bit processors should be faster than 32 bit processors ( or indeed why 16 bit processors should be faster than 8 bit processors!).
However there is a definite corelation between applications which require fast processing and applications which require lots of in memory data. (Large databases and really good graphic games are two that spring to mind). So chip manufacturers have tended to increase speed and addressability in parallel.
But then again if you cannot run a decent desktop in 1GB of memory just call yourself "Bill Gates".
Most of the mid to high end non Windows servers now uses 64 bit chips. They are nice and they are fast, but very few ship with > 2GB of memory.
PS. earlier on there was a comment whinging about SUN using a "special" compiler to increase thier SPEC scores. The compiler was not that "special" it just optimised for the parrellel instruction capability of the UltraSparc III architecture and will be included in the next release of the Sun compiler. So if you want really fast applications on the Sparc III you will have to buy Sun's compiler along with the truley awful "Forte" IDE.
MySQL strikes me as not much use for double entry bookkeeping if you cannot guarentee both entries got to the database.
As for it being an excellent persistance layer -- well if you cannot guarentee that the whole object was stored its not.
Don't listen to him. Tapes always let you down!
Some of the problems with tape are:---
Granularity -- i.e. you accidentally deleted "xmascard.lst" how do you know which tape it was on and how long will it take to scan several tapes to find your 2K file?
How can you be sure the tape is readable. In my experience something like 5% of backup tapes cannot be read by any drive except the one that wrote the tape.
No standard format, no backward compatabilty, short market life for thevarioustecnoligies. You may be very smug that you have an "acme" format backup stored in a safe deposit box, but after your house is flooded (mentioning the f*r* word upsets some people) you discover "acme" drives are no longer sold anymore, you just have to hope someone is trying to unload a working drive on e-bay.
Expense -- the drives, the tapes etc. will all cost several times as much as a couple of external hard drives.
Consume less beer ?????????
The guy is crazy!
Research has consitently found that consuming significantly less oxygen is 100% effective in stopping wieght gain.
Sweden probably had the best shot at demacratic socialism, but they had the sense to leave volvo, saab, Nobel etc as private companies.
One of the major companies that is still in government ownership is the maker of "Absolut" vodka, so any proponent of state ownership can get blind drunk with a clear conscience.
What really surpises me is that anybody can still take the concept of Marxist Economics seriously. We are talking about a guy who organised his own personal finances so badly that one of his children died of malnutrition.
It was mentioned in the article that the court officere managed to access and update information stored on an IBM mainframe system.
This is not Windows, but an operating system that has taken security very seriously for well over twenty years.
Any organisation abd/or profesional that configures one of these machines badly enough to allow data to be accessed with a trivial hack is definately negligent in my opinion.
Don't forget about halfpennies and farthings (1/4 penny) coins as well
CORBA is an open standard.
There is a standards commitee under the unberlla of the Open Systems Group.
Microsoft was a member of this group but I don't think they were very active. If I remeber correctly the OO academics ( Booch et all. were the driving force behind this one, and, Sun were the first to implement ).
As for COM ( or OLE, or DCOM or any of the other names this dogs breakfast has had). It hardly a standard as the interface varies with every release, and, there is no oppertunity for anyone but Redmond to provide input.
Architecture is definately the best tech job there is.
You get to play with all the new toys. (err.. Evaluate the latest technologies.)
You get taken to all the best restaraunts and bars by suppliers ( I mean establish fruitful working relationships with potential vendors).
Best of all you get to swan around, showing pretty slides with lots of boxes and look important.( ... comunicate new directions in technoligies to the companies descision makers.)
And if you can make as a Consultant Architect you pick up lots and lots of money, and, move on before somebody has to put your wierd and wonderful ideas into practice. (( .. constantly bring your in depth experience to new projects and customers.)
PS. I hope no one out there works out my real identity, or I may have to go back to doing a real job!
This is so obviously a defensive patent to protect there Websphere and Domino range of products from SM,BEA,Broadvision et all.
I mean has anyone actually had a polite letter from an IBM lawer recently?
Yep, its the humble oxytetracycline antibiotic (usually prescribed for step throat and othe rmonir infections).
I know this because every few years my wife gets an urge to see the Taj Mahal by moonlight.
This trip has been cancelled many times because of political urest, war, famine, and one year bubonic plague -- which was not such a big deal but we were advised not to travel if you were pregnant or likely to be pregnant as oxytetracacline has adverse effects on the developing feotus.
The "java platform" is not just experience and folk wosdom.
It is a standard well defined set of facilities (JVM, comiler, JAR files et. etc. ) which a java platform must support, plus, at a higer level of protocols for defining things like javabeans and EJBs. These are not vague fuzzy conventions: a platform either supports EJBs or it doesn't, an IDE can support java beans or it cannot.
This is very different from say "C" where you have the language and a set of library functions which are very precisely defined and thats it.
Programming C/C++ for OS/390 CICS has very little in common with coding MFC programs for windows apart from the language itself. Programming Java usings EJBs is pretty much the same for beans residing on OS/390 cics as it is for "COM+" NT based beans.
Having played around with wxPython, I can honestly say I will never go back to Tk or any other GUI interface.
What the article doesn't empahsise enough is that this is a cross platform GUI. The same code will work on win32, nearly all *nixes, plus apple and some wierd paltforms to boot.
The other thing the article doesn't impress strongly enough is how good the underlying C++ wxWindows library is. This is C++ GUI Nirvana, OO from the bottom up, crossplatform, native look and feel, a really useful set of builtin widgets, plus, perhaps most importantly a really well though out and easily extandable class structure.
From now on I will use nothing else, I may even give up programming curses.
I think people are still in "Office" mode thinking, and, want an integrated wordprocessor, spreadsheet and moni-database package.
Its the spreadsheet capabiliy that makes people choose StarOrifce (which I have never taken too) over an AbiWord, GnuCalc and MySql combo.
Not to mention some architectural gotchas unique to Windoze.
The NT kernal uses a message passing architecture, and, interupts are translated into messages on an internal queue, which, are then picked up by the appropraite aplication(s). This works fine (better than fine in fact) for a single processer running a GUI application.
However only one processor can be used to manage the actual message queues. In a multiprocessor machine NT splits the queue into 2 queues one for software generated interupts and one for hardware interupts. Processor 0 handles the software interupts and the higest numbered CPU handels the hardware interupts.
So it doesn't matter how many processors you add to your system, a sungle processor must handle all the IO and network interupts.
For a client and IO intensive application like a mail server this rapidly becomes the main bottleneck on performance.
Hence the pretty much standard recomendation of a two processor machine to handle mail. You can buy an eight processor machine but two processors will end up doing all the work and the other six will sit there twiddling thier virtual thumbs.
Both SUN and IBM are currently shipping chips that outperform the Itanium (which is very much sample ware) on every real world benchmark.
Furthermore the latest generation of the Power4 chip (Shipping in October) comes with two processers with a builtin shared L3 cache on a single chip!
In addittion it features several other speed enhancing innivations.
SUN are rummored to have something similar in mind for thier next generation Sparcs.
Basically INTEL are dying on the alter of "386" compatability. In trying to make a 64 bit chip compatable at instruction level to the worlds worse 16 bit instruction set INTEL have set themselves an impossable task.
The reason Oracle (and any other producer of benchmarked software) port to the Alpha platform early is that it gives the highest benchmark scores.
Probably 50% of sales for the Alpha/Tru64 paltform are benchmark related. ( As opposed to 100% of NT alpha sales).
The reasons HPUX is one of the last ports to be made are:-
a. Most unix professionals have come to hate HP/UX.
b. Its actually three ports you need to make, one for 32 bit HP/UX 7 and earlier, one for 32 bit HP?UX HP/UX 8 and later and one for 64bit HP/UX 8 and later.
c. Nobody is buying new HP boxes they are slow, overpriced and have a very peculiar unix running on them. The market leaders are Sun with thier very basic very standard Solaris OS, or, IBM with thier very reliable easy to use full featured AIX OS -- both produce well made very fast boxes.
Hotpaq Crashhard ?
The "bomb" in dot.bomb
Buying double glazing, timesharing an overpriced holidy apartment, or, even contributing to the Veterins Association wasn't what I had in mind when I installed my phone.
But will those b*st*rd* ( and b*tch*s) stop calling me, or, even get of the line --- NO.
Can I sue them for misuse of resources?
Well most people aren't using computers just because they like hacking; they want there computer to send mail, write letters, surf the web, play quake & etc.
There is no reason why linux shouldn't be as easy to install as windows, other, than the general "it's cool to hack" ethos which means the linux community doesn't even see the need for point and click installtion.
If you want to see linux running on the average users desk top then there has to be some way of getting it there that doesn't involve reading a two inch thick manual and learning bash shell scripting.