since my mother taught me how to program in QBASIC when I was eleven
Wow--you have a cool mother.
To answer the question, I don't think it really matters much. While a degree from MIT would definitely put you above the rest, most of the rest doesn't have an "elite" degree either.
What really matters is that you can show you know your stuff, that you actually fo know your stuff, and that you get real-world experience.
When I first read the headline, I thought it was reporting a major data loss incident at Creative Labs. I thought, "Awww, that's too bad. Maybe they can use this as an opportunity to have competent software engineers rewrite their notoriously terrible drivers from scratch." Ah well, maybe next year.
Itanium is being pushed increasingly into higher end computers. You know why Itanium is important?
Power970 cpu limit: 2-4 cpus Opteron cpu limit: 8 cpus Itanium cpu limit: 512 cpus.
I'm afraid you are quite incorrect.
A system builder is not limited by the processor architecture in how many processors can be added to a single system. If a company were willing to throw enough money at it, you could have a 32-way i386. It would be rather inefficient, as the i386 is not designed to make SMP systems efficient and easy to implement, but it could be done.
The limiting factor is the interconnect logic between the CPUs (and in software land, the OS). The Opteron is in no way limited to 8-way systems; that is just the point at which a designer must add their own interconnect logic between the CPUs, because the 8xx series of Opterons "runs out of" Hypertransport links. In fact, Sun Microsystems and Serverworks are collaborating on the creation of a 16-way and 32-way Opteron chipset".
For example that 2nd ranked "top500" computer is a 20 machine Beowolf style cluster. Each machine has 512 cpus.
Further, I would argue that the "Top500" list is fairly meaningless, because the only test they use to measure "performance" is Linpack. All Linpack does is solve linear equations and linear least-squares problems. You'll notice that Xeon systems are much faster than Opterons in this test. This does not model the real world, where Opterons have a clear performance lead in almost everything you can throw at them.
There was a discussion about this on StorageReview.com about this time last year. You'll notice the last post in the thread is to an article describing the inadequacy of the current supercomputer benchmarks, and an announcement that they plan to completely overhaul the system by 2006. Not that you were defending the legitimacy of the Top500 list per se, but it was brought up in this thread, and provided an opportunity to bitch and moan about it without creating another boring message.:)
that the/. crowd really cares, as we all know that even if they do eventually choose to use AMDs in the desktops that they will contiunue to use the shittyest mobos possible, and we wont be able to overclock... HOW FUN!
Dell uses Intel motherboards. Intel motherboards are among the most reliable and trusted by experienced IT folks and, along with Intel chipsets, are the reason that many people stick with Intel even when AMD's processors are faster, better designed, lower temperature, and cheaper.
But of course, you cannot overclock most of them easily, ergo they are terribly low-quality parts, or as you put it, the "shittyest mobos possible."
I connect my laptop to my TV via S-video (no HDTV yet) and to the stereo with a conventional 1/8" jack. For movies, I mount a fileshare and stream DivX recordings of DVDs over wireless, with the laptop using the TV as a "second monitor". For audio, my main stereo is already connected to my desktop PC, which streams my FLAC and MP3 files from the same file server.
Er, oh yeah. Thanks for pointing that out. I knew that, so I don't know what I was smoking when I mentioned OpenVMS in my example of C software. Someone please mod this up +1 informative.
"...and he notes that the problem is largely with C/C++ and mostly because of the buffer overflow problems."
OpenBSD and OpenVMS are written in C. Qmail and djbdns are written in C. Is it difficult to prevent buffer overflows? If you are reading a string, either use a string class, or read only as many characters as the character array can store. (What a novel idea!) If you are writing a string, among other things, set the last possible character of that string to null, just in case. These are but single simplified examples, but it is not impossible by any means, or even all that difficult, to write solid code. Among other things, the problem is that it takes individual effort to make sure every static-sized buffer isn't abused. As Murphy would tell you, human error is bound to crop up--increasingly so as the complexity of the project increases. I believe there was a post on the formula for this not too long ago.
As to the solution, well, that's a tough one. Higher level languages (Java, C#) help reduce these problems (and help reduce performance as well), but are just a band-aid. Perhaps the Manhattan Project (no, not that one) will come up with something better.
Until then, try to avoid products which have proven themselves to be full of holes year after year, week after week. And no, this doesn't just include all Microsoft server software. BIND and Sendmail come to mind.
I probably should have drawn some sort of link between storage and Moore's law, as I can see that the FP post seems to imply that Moore's Lawbservation inherently, necessarily applies to storage, which it does not as you mentioned. This was not my intention.
This misses the point entirely. It doesn't matter that Moore happened to be talking about transistor count when he made his famous observation--the point is that the exponential (rather than linear) growth he observed applies to more than just transistor count. It is a famous observation that is easily recognized, so it was appropriate to compare the slowing of storage technology to the slowing of IC advancement. Besides, Slashdot submissions often need a little extra flair to be accepted for publication.
I submitted this article, and expressly used Moore's Lawbservation in terms of hard drive technology because like transistor count in ICs, hard drive storage increases exponentially rather than linearly. Who said it can be used ONLY to describe processors and the like? You?
If you would like to discuss how it relates to storage--and it does--feel free to post to the StorageReview forums, or email me through them (user name: Sivar).
And why not? While it was originally about transistor count and such, I see no reason why it cannot be applied to other aspects of computing hardware, or of any technology that improves in a exponential rather than linear fashion.
Motorola on the other hand designed their 68000 was designed to be a 32bit chip from the get go, which I believe was first introduced in 1979 or so, at least according to my data book titled "break away from the past". Makes you wonder why anyone thought it was a good idea to use the 8086 for the PC.
I've spoken to some of the people that made this decision for various companies (e.g. Raytheon). The general consensus was that the difference between the 68K and the x86 was "night and day", but that the Intel chips were available dirt cheap (for the time) in massive quantities, and that there were more developers that knew the architecture (at least in their specific situations). Thus, yet again, business logic overrode technical/engineering prowess, and most of us have been using the worst major CPU architecture ever designed because of it. (Not that I blame Intel, they didn't have any other architecture to learn their mistakes from).
I look forward to OpenBSD releases not because I use OpenBSD, or even that I am particularly interested in it -- it's the OpenBSD songs that I wait for. They are actually quite good. "The Legend of Puffy Hood" and "Puff the Barbarian" were particularly good, and I found that many non-geeks liked them for their music if nothing more (and they tend to appreciate the lyrics once they read the page I linked above).
Speaking of lyrics, if you read them, you will find that they are actually very clever political satire. The 3.5 song, and some of the earliest ones, are somewhat disappointing IMO, but the others are excellent (and 3.5 is as well if you can appreciate a parody a parody).
I highly recommend that all Slashdot readers take a look at the page and download the OpenBSD songs. You may even want to pick up an ISO of OpenBSD while you're there.
You are right in general, however, specifically on the Athlon64 in long mode, "long" is 64-bit (as is "long long"), just as int is 32-bit on Pentiums even though only 16 bits are guaranteed.
Hopefully they come up with something better than "long long long" before the advent of widely available 128-bit processors, eh?.
I noticed that the summary points out how the AMD64 is faster. Well, what were you expecting, its 64bit vs. 32bit.
If you want a fair comparison then compare the AMD 32bit processors against Intel 32bit processors.
And compare AMD64s against Intel's 64 bit chips.
Someone needs a gentle tap with a cluestick.
1) Being 64-bit does not necessarily improve performance and, in fact, can degrade performance when used on the VAST majority of applications that primarily use integer numbers of less than 4.3 billion (2^32 unsigned). Take a look at Solaris/SPARC64 for an example.
2) Even in applications that can make use of 64-bit integers, the AMD64 specification defines an "integer" as 32-bits. Software has to expressly use a "long" (or similar) to make use of the other half of the register size, and because on 95% of computers out there (read: vanilla x86 systems) a "long" is the same thing as an "int", this is done rarely at best.
3) Even if all software in the universe could get a staggering performance boost from 64-bit registers AND were instantly tuned to use them, it wouldn't matter because all of the software used to compare the Athlon64 to the Pentium IV is 32-bit software running on a 32-bit operating system, except in the occasional tests that are designed specifically to test the benefit of the Athlon64's 64-bit mode.
4) Even if every one of the professional review sites were manned by biased or clueless authors (generally true of Tom's Hardware and GamePC (and any review website run by your average l33t w4r3z d00d or non-technical game enthusiast), though the former appears to be improving), the 10% average gain when compiling software to use the 64-bit extentions of the Athlon64 is nowhere near the actual performance gain, in 32-bit software, that the Athlon64 has over the Pentium IV in most games and a number of other applications.
5) Even if the performance gain of 64-bit mode was greater by far than it is now, the bulk of the performance improvement in most software is from a: the integrated memory controller (which is also used in 32-bit mode), and b: the fact that the number of general-purpose registers has doubled from 8 to 16, greatly reducing the amount of register variable swapping needed. Again, most apps simply do not care if they can fit huge numbers in a register, because they do not need them.
If only the search tool would support OpenOffice documents, it would be more useful to many people. Surely the zip'd XML dormat is easier to figure out than the intentionally-difficult-to-parse Offiec format. [Hint for google employees]
I had opened the link in a new tab and was about to read it, and then I caught the "I use Mandrake, so that makes me suave and sophisticated." part, which instantly removed any faith I may have had in the test.;-)
Surprisingly, that review seemed to be unbiased and made few if any rediculous claims based on miniscule performance differences. If they keep this up, I might start reading them again.
Slashdot Mirror
Why not...Oh, I don't know... Use Internet explorer? Build a plugin for the proprietary content.
Or use Netscape for that matter; they do own the company.
Talk about lack of focus.
Why is it that so many failing companies insist on doing things the hard, expensive way? Or did I answer my own question?
Reaching 1.0 is a big deal, because certain PHBs will not allow the installation of "beta" products.
I agree it is kind of irritating when every dot release of every major OSS project is announced though.
To answer the question, I don't think it really matters much. While a degree from MIT would definitely put you above the rest, most of the rest doesn't have an "elite" degree either.
What really matters is that you can show you know your stuff, that you actually fo know your stuff, and that you get real-world experience.
You're both correct: It's So, you're busy playing Half-Life Two too...
When I first read the headline, I thought it was reporting a major data loss incident at Creative Labs.
I thought, "Awww, that's too bad. Maybe they can use this as an opportunity to have competent software engineers rewrite their notoriously terrible drivers from scratch." Ah well, maybe next year.
A system builder is not limited by the processor architecture in how many processors can be added to a single system. If a company were willing to throw enough money at it, you could have a 32-way i386. It would be rather inefficient, as the i386 is not designed to make SMP systems efficient and easy to implement, but it could be done.
The limiting factor is the interconnect logic between the CPUs (and in software land, the OS).
The Opteron is in no way limited to 8-way systems; that is just the point at which a designer must add their own interconnect logic between the CPUs, because the 8xx series of Opterons "runs out of" Hypertransport links.
In fact, Sun Microsystems and Serverworks are collaborating on the creation of a 16-way and 32-way Opteron chipset".
Further, I would argue that the "Top500" list is fairly meaningless, because the only test they use to measure "performance" is Linpack. All Linpack does is solve linear equations and linear least-squares problems. You'll notice that Xeon systems are much faster than Opterons in this test. This does not model the real world, where Opterons have a clear performance lead in almost everything you can throw at them.
There was a discussion about this on StorageReview.com about this time last year. You'll notice the last post in the thread is to an article describing the inadequacy of the current supercomputer benchmarks, and an announcement that they plan to completely overhaul the system by 2006.
Not that you were defending the legitimacy of the Top500 list per se, but it was brought up in this thread, and provided an opportunity to bitch and moan about it without creating another boring message.
But of course, you cannot overclock most of them easily, ergo they are terribly low-quality parts, or as you put it, the "shittyest mobos possible."
I connect my laptop to my TV via S-video (no HDTV yet) and to the stereo with a conventional 1/8" jack.
For movies, I mount a fileshare and stream DivX recordings of DVDs over wireless, with the laptop using the TV as a "second monitor".
For audio, my main stereo is already connected to my desktop PC, which streams my FLAC and MP3 files from the same file server.
Er, oh yeah. Thanks for pointing that out. I knew that, so I don't know what I was smoking when I mentioned OpenVMS in my example of C software.
Someone please mod this up +1 informative.
OpenBSD and OpenVMS are written in C. Qmail and djbdns are written in C.
Is it difficult to prevent buffer overflows? If you are reading a string, either use a string class, or read only as many characters as the character array can store. (What a novel idea!) If you are writing a string, among other things, set the last possible character of that string to null, just in case.
These are but single simplified examples, but it is not impossible by any means, or even all that difficult, to write solid code.
Among other things, the problem is that it takes individual effort to make sure every static-sized buffer isn't abused. As Murphy would tell you, human error is bound to crop up--increasingly so as the complexity of the project increases. I believe there was a post on the formula for this not too long ago.
As to the solution, well, that's a tough one. Higher level languages (Java, C#) help reduce these problems (and help reduce performance as well), but are just a band-aid. Perhaps the Manhattan Project (no, not that one) will come up with something better.
Until then, try to avoid products which have proven themselves to be full of holes year after year, week after week. And no, this doesn't just include all Microsoft server software. BIND and Sendmail come to mind.
I probably should have drawn some sort of link between storage and Moore's law, as I can see that the FP post seems to imply that Moore's Lawbservation inherently, necessarily applies to storage, which it does not as you mentioned. This was not my intention.
Oops.
This misses the point entirely. It doesn't matter that Moore happened to be talking about transistor count when he made his famous observation--the point is that the exponential (rather than linear) growth he observed applies to more than just transistor count. It is a famous observation that is easily recognized, so it was appropriate to compare the slowing of storage technology to the slowing of IC advancement.
Besides, Slashdot submissions often need a little extra flair to be accepted for publication.
I submitted this article, and expressly used Moore's Lawbservation in terms of hard drive technology because like transistor count in ICs, hard drive storage increases exponentially rather than linearly. Who said it can be used ONLY to describe processors and the like? You?
If you would like to discuss how it relates to storage--and it does--feel free to post to the StorageReview forums, or email me through them (user name: Sivar).
And why not? While it was originally about transistor count and such, I see no reason why it cannot be applied to other aspects of computing hardware, or of any technology that improves in a exponential rather than linear fashion.
I've spoken to some of the people that made this decision for various companies (e.g. Raytheon). The general consensus was that the difference between the 68K and the x86 was "night and day", but that the Intel chips were available dirt cheap (for the time) in massive quantities, and that there were more developers that knew the architecture (at least in their specific situations). Thus, yet again, business logic overrode technical/engineering prowess, and most of us have been using the worst major CPU architecture ever designed because of it.
(Not that I blame Intel, they didn't have any other architecture to learn their mistakes from).
I look forward to OpenBSD releases not because I use OpenBSD, or even that I am particularly interested in it -- it's the OpenBSD songs that I wait for. They are actually quite good.
"The Legend of Puffy Hood" and "Puff the Barbarian" were particularly good, and I found that many non-geeks liked them for their music if nothing more (and they tend to appreciate the lyrics once they read the page I linked above).
Speaking of lyrics, if you read them, you will find that they are actually very clever political satire. The 3.5 song, and some of the earliest ones, are somewhat disappointing IMO, but the others are excellent (and 3.5 is as well if you can appreciate a parody a parody).
I highly recommend that all Slashdot readers take a look at the page and download the OpenBSD songs. You may even want to pick up an ISO of OpenBSD while you're there.
You are right in general, however, specifically on the Athlon64 in long mode, "long" is 64-bit (as is "long long"), just as int is 32-bit on Pentiums even though only 16 bits are guaranteed.
Hopefully they come up with something better than "long long long" before the advent of widely available 128-bit processors, eh?.
Er, oh yeah. Thanks. Clearly, I follow sports very closely. ;-)
Of course, Baseball is the definitive method for predicting the outcome of political elections.
(?)
Someone needs a gentle tap with a cluestick.
1) Being 64-bit does not necessarily improve performance and, in fact, can degrade performance when used on the VAST majority of applications that primarily use integer numbers of less than 4.3 billion (2^32 unsigned). Take a look at Solaris/SPARC64 for an example.
2) Even in applications that can make use of 64-bit integers, the AMD64 specification defines an "integer" as 32-bits. Software has to expressly use a "long" (or similar) to make use of the other half of the register size, and because on 95% of computers out there (read: vanilla x86 systems) a "long" is the same thing as an "int", this is done rarely at best.
3) Even if all software in the universe could get a staggering performance boost from 64-bit registers AND were instantly tuned to use them, it wouldn't matter because all of the software used to compare the Athlon64 to the Pentium IV is 32-bit software running on a 32-bit operating system, except in the occasional tests that are designed specifically to test the benefit of the Athlon64's 64-bit mode.
4) Even if every one of the professional review sites were manned by biased or clueless authors (generally true of Tom's Hardware and GamePC (and any review website run by your average l33t w4r3z d00d or non-technical game enthusiast), though the former appears to be improving), the 10% average gain when compiling software to use the 64-bit extentions of the Athlon64 is nowhere near the actual performance gain, in 32-bit software, that the Athlon64 has over the Pentium IV in most games and a number of other applications.
5) Even if the performance gain of 64-bit mode was greater by far than it is now, the bulk of the performance improvement in most software is from a: the integrated memory controller (which is also used in 32-bit mode), and b: the fact that the number of general-purpose registers has doubled from 8 to 16, greatly reducing the amount of register variable swapping needed. Again, most apps simply do not care if they can fit huge numbers in a register, because they do not need them.
So as you can see, your assertion is flawed.
If only the search tool would support OpenOffice documents, it would be more useful to many people. Surely the zip'd XML dormat is easier to figure out than the intentionally-difficult-to-parse Offiec format. [Hint for google employees]
I had opened the link in a new tab and was about to read it, and then I caught the "I use Mandrake, so that makes me suave and sophisticated." part, which instantly removed any faith I may have had in the test. ;-)
Surprisingly, that review seemed to be unbiased and made few if any rediculous claims based on miniscule performance differences. If they keep this up, I might start reading them again.