Unfortunately, AMD doesn't have the marketing muscle or horde of fanatic, retarded consumer's that Apple does. Apple somehow convinced it's drones to start chanting "Megahertz doesn't matter" and later to double the G4 megahertz when comparing to x86 chips.
It really sucks that there is no standard benchmarking technique for processors with practically the same instruction set. While benchmarks can be misleading, I'm sure that they are (on average) better then pure megahertz numbers. Of course then there's the problem of not being able to benchmark the chip by itself...
Re::Free2OneAndOneHalfGrand
on
Penguin2Apple
·
· Score: 1
Re:Read Eric Hoffer's "True Believer"
on
Penguin2Apple
·
· Score: 1
He said that he there was some good
natured ribbing from the Linux users. He said
it was totally differnet from the mean spirted karma of the Apple shows
This behavior may have to do with how much of a threat Microsoft is perceived to be by different groups. People don't like being threatened and respond adversely.
Re:OS X is Linux what a protein bar is to a full m
on
Penguin2Apple
·
· Score: 1
That's because extended sustenance on "protien bars" can cause brain damage.
Wow!! I can get AOL for free with a Mac? Now I don't have to wait for days on end for a CD in the mail. Ooh...Acrobat Reader and IE too! OMG...did you say Mail comes bundled??!!!
Even so, you can emulate interfaces given the ability to do multiple inheritance but not vice verse.
I'm not saying that multiple inheritance is better, but it is definitely more powerful.
And how does garbage collection lead to more work looking for memory leaks? It may make a programmer more lazy because most of the time he doesn't have to worry about them, but I don't see how garbage collection is a bad thing (excluding performance factors)
I agree with most of what you said, but while more people run Windows, there is probably more intelligent scrutiny on an Linux because people have access to the source code.
Of course, it may be more valuable to developers to have more mindless bug reporters than fewer knowledgable ones.
...if you were serious enough to waste all of Slashdot's time with this question...
Is that even possible? Only those who deem the question useless and then proceed to respond to it are wasting their time. It is a voluntary thing. I don't think people who have read this far down this thread are constrained by a tight schedule.
That's why there are no mission critical servers running on cisc.
What does mission criticality (word?) have to do with CISC latency? It's not like all CISC chips may suddenly fail, thus making them unusable on a mission critical server. It probably has more to do with the operating system. It's not like the PowerPC is famous for running mission critical apps (since MacOS
This meta information could be summed up in the first byte. If there was no encryption, the byte could be set to zero. So you're right, the worst case is slightly greater than zero but in a 1MB stream, the ratio becomes 1.001:1.000 so it's really not that bad.
I just started using XML-FO and while it isn't really a full-featured document creator, it works for what I need.
For example, you can't exclude headers/footers from a specific page. If they're on one page, they're on them all.
It's also difficult to write XSL-FO directly (extremely verbose). If you know XSL, however, this shouldn't be a problem. Once you have your template, you can transform and then convert to PDF.
I use Apache's FOP but it seems to be really buggy. It crashes a lot. You will probably want to do the XSL transformations in one step with Xerces-Java (also from Apache) because it'll give you more useful error messages and then do the XSL-FO to PDF conversion with FOP.
...no algorithm can compress arbitrary random data at better than 1:1 in the long run.
In the worst case, GZIP would leave the file the same size as the original. In other cases, however, GZIP will make the file smaller.
That means that the compression ratio is bounded with 1:1 as the worst case. So, on the average, the compression ratio will be better than 1:1.
In perfectly random 8 bit text, the chance of 3 of the same character appearing consecutively is... 1 in 16777216
That's the chance of a specific character appearing three times in 8-bit text that is 3 characters in length.
In 8-bit text of length N characters (where N is at least 3), expected number of runs of C repeating characters is something like:
n-2 / 255^(c-1)
For C = 3, a run of 3 consecutive digits is expected approximately every 65k characters (which would be around 1 in 65k for every character). However, even current compression methods don't rely on repetition of specific characters. When certain computable patterns are taken into account, this ration should go up.
That (STL sorting) seems like the equivalent of running a code generator to generate a quicksort function for every required data type. A C-code generation program along with some Makefile rules could do this, right? (though to some, this may be unforgivably lacking in elegance)
I was wondering if it is possible (and reasonable) for a C compiler to inline functions at runtime; could a compiler leave a blank space of maybe 50 no-ops and copy the code for the comparison function into that section at runtime? Then C programmers could program with the rough equivalent of Lisp macros (or the rough equivalent of what I think Lisp macros are:-)
If the comparison function is greater than 50 (or how many ever) instructions, the function call overhead will be relatively insignificant and a function call can be plugged in instead. I don't know anything about current caching, branch prediction and no-op handling schemes and was wondering if runtime inlining would work well (if hardware was optimized to deal with it efficiently).
The iPod can hold around 5GB. Thats plenty. That's enough to hold every song I actually listen to.
Though the iPod is one of the more expensive players, with storage capacity shooting through the roof as it is, 5GB may become standard in a few years.
Size-wise, MP3 files are probably small enough. But there are other issues here as well.
Extra delimiters help convey the nature of code. Most of the time, you don't need the flexibility of every single entity being equivalent/interchangeable with every other one. The extra delimiters act as implicit, concise documentation.
Um... that's why Scheme and common lisp are "properly tail recursive" - you can write recursive routines without ever-increasing stack.
Um... the fibbonacci algorithm the parent post provided wasn't tail-recursive. Only tail-recursive functions are optimized by "properly tail recursive" compilers.
Slashdot Mirror
Unfortunately, AMD doesn't have the marketing muscle or horde of fanatic, retarded consumer's that Apple does. Apple somehow convinced it's drones to start chanting "Megahertz doesn't matter" and later to double the G4 megahertz when comparing to x86 chips.
It really sucks that there is no standard benchmarking technique for processors with practically the same instruction set. While benchmarks can be misleading, I'm sure that they are (on average) better then pure megahertz numbers. Of course then there's the problem of not being able to benchmark the chip by itself...
Um....that's probably not a good idea.
This behavior may have to do with how much of a threat Microsoft is perceived to be by different groups. People don't like being threatened and respond adversely.
That's because extended sustenance on "protien bars" can cause brain damage.
They're using their 'piracy' e-mail address to handle this...probably to scare the host.
Wow!! I can get AOL for free with a Mac? Now I don't have to wait for days on end for a CD in the mail. Ooh...Acrobat Reader and IE too! OMG...did you say Mail comes bundled??!!!
cool...now I can write 50 helloworld-type applications and become the company that makes the majority of MacOS apps. That's 5 times more than Apple!
Even so, you can emulate interfaces given the ability to do multiple inheritance but not vice verse.
I'm not saying that multiple inheritance is better, but it is definitely more powerful.
And how does garbage collection lead to more work looking for memory leaks? It may make a programmer more lazy because most of the time he doesn't have to worry about them, but I don't see how garbage collection is a bad thing (excluding performance factors)
Sourceforge has a shortened domain name too. So that would make it just net.sf.stalk.
I agree with most of what you said, but while more people run Windows, there is probably more intelligent scrutiny on an Linux because people have access to the source code.
Of course, it may be more valuable to developers to have more mindless bug reporters than fewer knowledgable ones.
Is that even possible? Only those who deem the question useless and then proceed to respond to it are wasting their time. It is a voluntary thing. I don't think people who have read this far down this thread are constrained by a tight schedule.
What does mission criticality (word?) have to do with CISC latency? It's not like all CISC chips may suddenly fail, thus making them unusable on a mission critical server. It probably has more to do with the operating system. It's not like the PowerPC is famous for running mission critical apps (since MacOS
I can't see how the original poster could have meant that, unless you consider KAI's/Intel's compiler a subsidiary of GCC.
This meta information could be summed up in the first byte. If there was no encryption, the byte could be set to zero. So you're right, the worst case is slightly greater than zero but in a 1MB stream, the ratio becomes 1.001:1.000 so it's really not that bad.
I just started using XML-FO and while it isn't really a full-featured document creator, it works for what I need.
For example, you can't exclude headers/footers from a specific page. If they're on one page, they're on them all.
It's also difficult to write XSL-FO directly (extremely verbose). If you know XSL, however, this shouldn't be a problem. Once you have your template, you can transform and then convert to PDF.
I use Apache's FOP but it seems to be really buggy. It crashes a lot. You will probably want to do the XSL transformations in one step with Xerces-Java (also from Apache) because it'll give you more useful error messages and then do the XSL-FO to PDF conversion with FOP.
There's a commercial XSL-FO package too and I've read that it is better than FOP.
Some people seem to have taken advantage of this idea in their revolutionary new program! Open source prevails over commercial software again!!
In the worst case, GZIP would leave the file the same size as the original. In other cases, however, GZIP will make the file smaller. That means that the compression ratio is bounded with 1:1 as the worst case. So, on the average, the compression ratio will be better than 1:1.
That's the chance of a specific character appearing three times in 8-bit text that is 3 characters in length.
In 8-bit text of length N characters (where N is at least 3), expected number of runs of C repeating characters is something like:
For C = 3, a run of 3 consecutive digits is expected approximately every 65k characters (which would be around 1 in 65k for every character). However, even current compression methods don't rely on repetition of specific characters. When certain computable patterns are taken into account, this ration should go up.
That (STL sorting) seems like the equivalent of running a code generator to generate a quicksort function for every required data type. A C-code generation program along with some Makefile rules could do this, right? (though to some, this may be unforgivably lacking in elegance)
:-)
I was wondering if it is possible (and reasonable) for a C compiler to inline functions at runtime; could a compiler leave a blank space of maybe 50 no-ops and copy the code for the comparison function into that section at runtime? Then C programmers could program with the rough equivalent of Lisp macros (or the rough equivalent of what I think Lisp macros are
If the comparison function is greater than 50 (or how many ever) instructions, the function call overhead will be relatively insignificant and a function call can be plugged in instead. I don't know anything about current caching, branch prediction and no-op handling schemes and was wondering if runtime inlining would work well (if hardware was optimized to deal with it efficiently).
The Generic Java variant supports generic types (though the types have to be object types).
The iPod can hold around 5GB. Thats plenty. That's enough to hold every song I actually listen to.
Though the iPod is one of the more expensive players, with storage capacity shooting through the roof as it is, 5GB may become standard in a few years.
Size-wise, MP3 files are probably small enough. But there are other issues here as well.
Except now it costs $750. The $80 version doesn't have Athlon and Pentium 4 support.
Extra delimiters help convey the nature of code. Most of the time, you don't need the flexibility of every single entity being equivalent/interchangeable with every other one. The extra delimiters act as implicit, concise documentation.
Um... that's why Scheme and common lisp are "properly tail recursive" - you can write recursive routines without ever-increasing stack.
Um... the fibbonacci algorithm the parent post provided wasn't tail-recursive. Only tail-recursive functions are optimized by "properly tail recursive" compilers.
If only physics was the major of choice among the mentally stable.