All the OO features are nice, but what's really missing in PHP are some critical "enteprise" features, like true connection pooling (and no, pconnect doesn't count).
An xbox + XBMP provides most of the functionality, and it's only 180$ (in all fairness, you do need a dolby decoder with your speakers, but it has HDTV in it)
Furthermore, by purchasing an XBOX without actually buying games you make MS lose money:) (they're losing money anyway with xbox, but this way they're losing even more)
These days it doesn't even take a screwdriver to hack the XBOX... The (albeit kludgy) software solution works well.
OS/2 was considerably more stable than Windoze - that cannot be said about AMD processors with respect to Intel processors. For one, the safety margins that Intel procs use are higher than AMD's (overclockers can attest that... )
Sure, you can complain about their processors being beaten by AMD's when running at the same clock, but the manufacturing quality of Intel processors is very high.
Seriously, Linux is eating their marketshare. It is considerably easier to migrate from Solaris/SPARC to Linux/x86 than from Windows to Linux.
The only reason they're in bed with Linux right now is because they don't really have a choice. They in fact boycotted Java/Linux for quite a while (version 1.3 was the first, and it was also largely developped by Blackdown)
For them to be like SCO, they should have threatened to sue customers which purchased Athlons with DDR, and ridiculously claim that "The only reason your system is so fast is because your memory has our IP in it; those underpaid AMD designers couldn't possibly have done a good system design just by themselves" (and not because you have a state-of-the-art processor...)
You forget the fact that C++ as we know it today was designed mostly by Bjarne Stroustroup. The initial requirements (be high-level and low-level at the same time)
Don't blame the community for the absurd C++ feature called "references" (dumb references - you can't actually use them instead of pointers, because they can't be NULL; at the same time, if you have code with references, pointers and stack-values, everything degenerates into an incomprehensible mess - java references are the way to go, or just pointers), or for the fact that the template sub-language is Turing-complete (yes, you can actually write partial recursive functions in template language... ugh).
... but it is simply fscked up... I would have no problem if you started with a clean, RISC, ISA, but x86 ? So ugly that even Intel tried to get away from it.
Not to mention that the time when humans wrote assembly that was better than a compiler's output is long gone. Processors these days are just too damn complex. Direct assembly might be the best choice for simple processors on embeded systems, but certainly not for superscalar processors.
How many assembly programmers unroll loops for instance ? Loop unrolling is crucial for good performance,
as branches are the 2nd biggest performance killer. How many actually think of instruction latencies and schedule instructions accordingly?
How many people realize that the processor actually has caches and missing to main memory can take as much as 400 cycles ? How many people use predication to get rid of branches ?
Yes, a good compiler has knowledge of these things...
It's much easier to vectorize Fortran loops than C loops, as Fortran does not have pointers and it's almost always safe to vectorize. Vectorizing is the easy part, figuring out if it's safe is extremely difficult in C/java.
So as strange as this may sound fortran can be much faster!
For the regular Xbox to get under a 100$. Combined
with xbox media player, it's basically the ultimate
media box. You'd have to shell out ~200$ to get a DVD player which does just divx, with an Xbox you can pretty much play anything, don't have to worry about DVD zoning, etc (and it does support HDTV and Dolby Digital, although you need speakers capable of decoding it)
I agree with you that CISC vs. RISC is not an issue any more. The decoded tracecache that P4 has, for instance, takes the conversion off the critical path.
x86 however has a ridiculously small number of registers. This means that you have to go to memory A LOT. It's easy to make register operations fast, extremely hard to make memory fast. The performance gap between memory and processors is constantly increasing.
That's why x86-64 has 16 general purpose registers, Alpha - 64 and Itanium... 128.
Bottomline: we do need a replacement for x86. Not because it's CISC, but because it has too few registers.
And, btw, it's not a bad idea to demand more from the compiler. Big table structures on the chip require space and affect clock cycles. Compilers just don't.
I'll start by saying that I am not a physicist (by far).
They create heavy elements, which are so unstable that they decay as quickly as they were created.
So I'm wondering - what's the point ? Just getting your name associated with an element in the periodic table ? It seems to me that the money would be better spent in doing stuff with real applications (like producing cheaper anti-matter or getting closer to controlled fusion)
For servers, addressing with more than 32 bits is crucial these days. The question is - how do you get performance improvement on a desktop ?
The real performance gain is in the change of ISA (instruction set architecture). True, calling it 64-bit vs. 32-bit is pretty much a marketing paint. The real issue with x86 is not even the fact that it's CISC - it's the number of registers. Few general-purpose registers means that you have to go to memory A LOT. x86 has 8 GPRs - the compiler can barely allocate 2 or (at most) 3 of them to variables.
It's much easier these to make register operations fast than memory ones. The x86-64 has 16 GPRs - you can actually do some useful register allocation with them and reduce the memory traffic. Itanium... well, it has 128 registers.
That's why you get performance improvement just by recompiling an app that doesn't even use "long longs" to "64 bits".
All raw images are black and white - how do they transform them into RGB colour ?
They clearly do a lot of artistic frequency remapping when they present photos of stars/galaxies/nebulae (i.e. convert invisible spectrum into visible one), but I'd really hope they do something more scientific with data from neighbouring Mars.
The current queen knighted one of the harshest dictators of Eastern Europe, Ceausescu, in 1969, for his stance on the invasion of the Chehoslovakia. All his record of abuses and violation of human rights were put to the side.
When the Romanian revolution happened in 1989, the queen promptly unknighted him....
The Xeon server processors do support an external L3 cache, but the memory still isn't any faster:). It does take no less than 400 cycles to go to memory to service an L2/L3 miss on such a processor.
The Opterons do slightly better, as the memory controller is on-chip.
Matlab is mostly loops. Loops generate branches with high predictability, and as a consequence deep pipelineing won't incur much performance loss. Furthermore there's a lot of parallelism in those loops, and the out-of-order execution engine is quite good at exploiting it (i.e. hide the long latency of FP ops by overlapping them)
It's much more likely the size of the L2 cache is affecting you (i.e. your working set does not fit into P4's L2 cache but it does in Barton's).
If you don't believe me, try the demo version of Intel Vtune performance analizer on matlab running one of your programs.
How well your caches perform is probably the most important thing for a processor today, as the speed of the main memory is a couple of orders of magnitude under the speed of the processor. It takes a couple of hundred cycles to service an L2 miss, while a long FP operation takes at most 20 cycles.
Not being a trade secret doesn't mean it's not prosecutable under DMCA.
The reason distros don't include libdvdcss is really DMCA.
DVD css was cracked through reverse engineering, which does not equate to stealing a trade secret. I do think that the outcome is important, but it won't really make that much of a difference IMHO (and of course, IANAL)
They're actually making the moon go away from Earth. They're effectively a friction force. To keep our moon we're really supposed to build huge dams in our oceans to stop tidal movement.
The real question is not what lunar mass removal will do to tidal waves, but to Earth's orbit. Our moon is Earth's orbit regulator. Mars doesn't have a big moon like Earth (big in relative terms, of course), and as a consequence it's orbit is much more irregular.
A change of a single tenth of a degree in Earth's tilt can have drastic consequences on everything we know on our planet.
I had a NEX II (not IIe) player that simply deadlocked after a while - out of the blue. There are quite a few people who reported similar behaviour.
Not to mention the fact that navigation is poor.
Furthermore, a 1GB flash card can be quite expensive (> 200$). A 1GB card + NEX player is just as much as a 20G HDD based player. Yes, you do get awesome battery time, but you're also very likely to listen to the same tune twice within that battery time.
Slashdot Mirror
All the OO features are nice, but what's really missing in PHP are some critical "enteprise" features, like true connection pooling (and no, pconnect doesn't count).
Furthermore, by purchasing an XBOX without actually buying games you make MS lose money :) (they're losing money anyway with xbox, but this way they're losing even more)
These days it doesn't even take a screwdriver to hack the XBOX ... The (albeit kludgy) software solution works well.
Yes, it's fair use, but the SNR of normal FM transmissions is less than 50 ...
Sure, you can complain about their processors being beaten by AMD's when running at the same clock, but the manufacturing quality of Intel processors is very high.
Seriously, Linux is eating their marketshare. It is considerably easier to migrate from Solaris/SPARC to Linux/x86 than from Windows to Linux.
The only reason they're in bed with Linux right now is because they don't really have a choice. They in fact boycotted Java/Linux for quite a while (version 1.3 was the first, and it was also largely developped by Blackdown)
For them to be like SCO, they should have threatened to sue customers which purchased Athlons with DDR, and ridiculously claim that "The only reason your system is so fast is because your memory has our IP in it; those underpaid AMD designers couldn't possibly have done a good system design just by themselves" (and not because you have a state-of-the-art processor ...)
Don't blame the community for the absurd C++ feature called "references" (dumb references - you can't actually use them instead of pointers, because they can't be NULL; at the same time, if you have code with references, pointers and stack-values, everything degenerates into an incomprehensible mess - java references are the way to go, or just pointers), or for the fact that the template sub-language is Turing-complete (yes, you can actually write partial recursive functions in template language ... ugh).
303 "See Other"
Not to mention that the time when humans wrote assembly that was better than a compiler's output is long gone. Processors these days are just too damn complex. Direct assembly might be the best choice for simple processors on embeded systems, but certainly not for superscalar processors.
How many assembly programmers unroll loops for instance ? Loop unrolling is crucial for good performance, as branches are the 2nd biggest performance killer. How many actually think of instruction latencies and schedule instructions accordingly? How many people realize that the processor actually has caches and missing to main memory can take as much as 400 cycles ? How many people use predication to get rid of branches ?
Yes, a good compiler has knowledge of these things ...
So as strange as this may sound fortran can be much faster!
For the regular Xbox to get under a 100$. Combined with xbox media player, it's basically the ultimate media box. You'd have to shell out ~200$ to get a DVD player which does just divx, with an Xbox you can pretty much play anything, don't have to worry about DVD zoning, etc (and it does support HDTV and Dolby Digital, although you need speakers capable of decoding it)
1. That camera has, most likely, a CMOS sensor (much, much slower than CCD, you can only take reasonable pictures in daylight)
2. Its cheap lens system makes you believe that you're in a different reality (i.e. all squares look round because of the radial distortion)
Integration of features is not bad, as long as you don't sacrifice quality.
x86 however has a ridiculously small number of registers. This means that you have to go to memory A LOT. It's easy to make register operations fast, extremely hard to make memory fast. The performance gap between memory and processors is constantly increasing.
That's why x86-64 has 16 general purpose registers, Alpha - 64 and Itanium ... 128.
Bottomline: we do need a replacement for x86. Not because it's CISC, but because it has too few registers.
And, btw, it's not a bad idea to demand more from the compiler. Big table structures on the chip require space and affect clock cycles. Compilers just don't.
"Yes, a hundred million"
"Whew ... for a second I thought you said a hundred billion ..."
They create heavy elements, which are so unstable that they decay as quickly as they were created.
So I'm wondering - what's the point ? Just getting your name associated with an element in the periodic table ? It seems to me that the money would be better spent in doing stuff with real applications (like producing cheaper anti-matter or getting closer to controlled fusion)
Well, I guess they could outlaw freenet as a protocol ...
For servers, addressing with more than 32 bits is crucial these days. The question is - how do you get performance improvement on a desktop ?
... well, it has 128 registers.
The real performance gain is in the change of ISA (instruction set architecture). True, calling it 64-bit vs. 32-bit is pretty much a marketing paint. The real issue with x86 is not even the fact that it's CISC - it's the number of registers. Few general-purpose registers means that you have to go to memory A LOT. x86 has 8 GPRs - the compiler can barely allocate 2 or (at most) 3 of them to variables.
It's much easier these to make register operations fast than memory ones. The x86-64 has 16 GPRs - you can actually do some useful register allocation with them and reduce the memory traffic. Itanium
That's why you get performance improvement just by recompiling an app that doesn't even use "long longs" to "64 bits".
They clearly do a lot of artistic frequency remapping when they present photos of stars/galaxies/nebulae (i.e. convert invisible spectrum into visible one), but I'd really hope they do something more scientific with data from neighbouring Mars.
When the Romanian revolution happened in 1989, the queen promptly unknighted him ....
The Opterons do slightly better, as the memory controller is on-chip.
It's much more likely the size of the L2 cache is affecting you (i.e. your working set does not fit into P4's L2 cache but it does in Barton's).
If you don't believe me, try the demo version of Intel Vtune performance analizer on matlab running one of your programs.
How well your caches perform is probably the most important thing for a processor today, as the speed of the main memory is a couple of orders of magnitude under the speed of the processor. It takes a couple of hundred cycles to service an L2 miss, while a long FP operation takes at most 20 cycles.
DVD css was cracked through reverse engineering, which does not equate to stealing a trade secret. I do think that the outcome is important, but it won't really make that much of a difference IMHO (and of course, IANAL)
The real question is not what lunar mass removal will do to tidal waves, but to Earth's orbit. Our moon is Earth's orbit regulator. Mars doesn't have a big moon like Earth (big in relative terms, of course), and as a consequence it's orbit is much more irregular.
A change of a single tenth of a degree in Earth's tilt can have drastic consequences on everything we know on our planet.
Not to mention the fact that navigation is poor.
Furthermore, a 1GB flash card can be quite expensive (> 200$). A 1GB card + NEX player is just as much as a 20G HDD based player. Yes, you do get awesome battery time, but you're also very likely to listen to the same tune twice within that battery time.
Sure, you might lose a couple of frames (at worst), but who cares ?