Nice troll on your part too! Let's see: Britain has had a parlimentary system for over four hundred years now, and I'd argue that the British Empire (remember that?) was just a wee bit more successful than the US.
Look at it this way. The CPU has a bunch of execution units on it. The P4, specifically, has two arithmatic units, two FPUs, and some other stuff. Since threads usually don't use all these units optimally, some are wasted. A second simultanious thread might be able to use the otherwise unused units, and thus the overall performance of the two threads combined increases.
Hah! The first poster on the OSNews thread about this story wasn't impressed either. Apparently, a lot of people don't have the attention span to read more than the first few tables in an article!
It is one binary, but there is most definately a top/below relationship. The whole "dual-kernel" is just Apple marketing. The BSD layer provides user level services using the core functionality present in the Mach kernel.
That's the fun part. For the low end 3D market, around $100 is the sweet spot. For that, you can get a GeForce4 Ti 4200, which is superior to most of the cheapo 3D cards you get. It's fast and has great driver (for Windows and Linux!) to boot. Compare this to a Xabre 400 card, which runs only $30 cheaper, but in many games offers half the performance. Those who are really going to feel that $30 hit can easily save another $10 and buy a GeForce4 MX 440 for $60. It's still faster than the Xabre 400, and even though it lacks pixel shaders, (1) at that price, you're not playing Doom III, (2) The Xabre's pixel shaders are so bad, for all intents and purposes, it doesn't have them.
True. For the last week or so, I've been using a keyboard with the keyboard drawer closed because I'm in too constrained of a space to sit back all the way. Those little nubs under your index fingers are there for a reason, you know.
Ha ha. Even the Java folks realized the power of generics. There are lots of things (take a look at Loki or Boost) that simply cannot be done in static languages like C/C++ and Java without something like the template mechanism. Remember, C++ isn't an OOP language, it is a multi-paradigm language. There are lots of cases (and data structures is a big one) where "proper OOP" sucks. Just look at the cast-fest that is the Java collection classes. One of the major powers of the template mechanism is that it allows high level design with low-level performance. And there are *lot's* of people (especially in graphics, science, systems-programming) that really do need low-level performance. Generic data structures without templates usually require either abandoning type safety (by using void pointers, the C way to do it) or abandoning performance (by using inheritence and virtual functions, the Java way to do it). With templates, you don't have to abandon either. In practice, the STL list class is about 50% the speed of a single-type C data structure. Very impressive considering the C structure has embedded link nodes and no copy on insert while the STL does. Vector and matrix classes using templates in scientific applications are often 100% the speed of an equivilent C structure, because of the fully optimizable nature of templates. Of course, like everything else in C++, templates are easy to abuse and use the wrong way. But they are hardly unnecessary and redundant.
In proper implementations, the STL doesn't add all that much overhead to a binary. A good linker will remove a lot of redundant code caused by the STL, and a good compiler will process STL code quite quickly. Intel C++, for example, handles the STL quite well. My first STL project was a physics simulation that was extremely data structure intensive. It took me at most a day to get the hang of things. The STL really is quite straightforward. Praytell, how would you simpify the STL string or list classes?
I find it different. I find it impossible to remember the totally bland, generic, and similar sounding names in Java, while I find the C++ names very concise and memorable.
The Bill of Rights was ignored by some of the contemporaries of the founding fathers. The tradition of being much more conservative about economic entities than individuals goes back to the time of the founding fathers and Hamilton's views on economics. The Bill of Rights are great for what it is applied to, individuals, but an economy is a different beast than society as a whole. And because you insist on turning this into a liberal vs conservative arguement, I will point out that one of the principal weaknesses of conservative ideology is the tendency to adhere so strongly to a set of logical principles, that the inherent complexity of society and government is simplified away and ideas start to be applied to situations where they aren't applicable.
Retarded naming scheme, small size, and wierd design of standard library. The damn thing's almost useless. >>>>>>>> I love the C++ STL naming scheme. Nice and concise, but still clear and memorable. Java and many other languages suffer far too much from name bloat. And the C++ STL is probably one of the last things a C++ programmer really understands, and it can be years before people truely appreciate the power of the STL.
It doesn't have anything to do with liberals or free speech. The US government long ago made the decision that the economy is to be closely watched, and corporations are distinct from individuals. Hence you have all sorts of laws (anti-trust, monopoly, SEC, trading, advertising) that go directly against many of the amendments. Look at it this way: corporations are not allowed to band together and engage in price fixing. Individuals are perfectly free to band together and come to a mutual agreement for their group's benefit.
I keep hearing "oh, these games aren't AAA titles, they must suck. Well, some of the most pure fun games I've played are obscure titles like these. They have nice graphics (not bleeding edge, but pretty) and fun gameplay. They not be the deep, sweeping experience of something like HalfLife, but they're still good entertainment. If the price is right, this games might be a good buy.
On Gentoo emerge nvidia-kernel emerge nvidia-glx vi/etc/X11/XF86Config; find 'nv' change to 'nvidia' Done. You can do the same for ATIs drivers, I think.
I still have the 210MB WD that came with my Tandy 486 SX/33 which I bought in 1990. I was actually using it as a swap drive (with a 1.2GB WD as the main drive) in the same Tandy (now running as a firewall) until about a year ago. The damn thing (HD, computer and all) is still kicking after 13 years, even though it's spent a significant amount of time sitting unprotected in my garage.!
Well, let's see. First, ripping DVD to VOB isn't the same as ripping CD to WAV, since the DVD is compressed to begin with and the CD isn't. From there, I know ever people who have 100 DVDs or more (doesn't have to be illegal, thanks to box sets and directors cuts and the fact that even buying individually, 100 DVDs is only a $1500-$2000 investment). As for games, 600MB is svelte. I don't know games that come on one disk anymore. UT2003 has a 2GB install, and even a 2D RPG like Icewind Dale is 700MB+ Then throw in Windows XP and it's 1GB+ minimum install, and even 1TB is starting to seem small.
Yes, they could do that. They could also hire their own mercenary army to take over the Justice Department (to back up the lobbyists;-). But will they? >>>>> Does it matter? Do you want the monkey holding the gun, whether or not it will shoot?
Because a lot of us are getting to the age where "going legit" seems like an attractive proposition. We would rather not run illegal software as our main environment. Besides, the Linux devs are respectable hackers, not a bunch of virus weenies.
The K6 never overtook the PII line in megahertz, which was what its real competitor was (chronologically). And while 3DNow! was significantly faster, it was no replacement for real FPU power (useful for scientific applications and 3D modeling) and was largely useless unless you were running one of the few applications directly coded for 3DNow!
http://www17.tomshardware.com/cpu/19980416/perform ance-02.html In everything from Quake II to 3D Studio, the 233 MHz K6 is about half as fast as the 233 MHz PII. I don't know why the Sandra benchmarks are different, but they are the exception, not the rule. Besides, Sandra is an artificial benchmark, not a real world one like 3D Studio or SPEC.
Slashdot Mirror
Nice troll on your part too! Let's see: Britain has had a parlimentary system for over four hundred years now, and I'd argue that the British Empire (remember that?) was just a wee bit more successful than the US.
Look at it this way. The CPU has a bunch of execution units on it. The P4, specifically, has two arithmatic units, two FPUs, and some other stuff. Since threads usually don't use all these units optimally, some are wasted. A second simultanious thread might be able to use the otherwise unused units, and thus the overall performance of the two threads combined increases.
Hah! The first poster on the OSNews thread about this story wasn't impressed either. Apparently, a lot of people don't have the attention span to read more than the first few tables in an article!
It is one binary, but there is most definately a top/below relationship. The whole "dual-kernel" is just Apple marketing. The BSD layer provides user level services using the core functionality present in the Mach kernel.
Um, isn't that what he just said? A Mach kernel with BSD layer on top?
That's the fun part. For the low end 3D market, around $100 is the sweet spot. For that, you can get a GeForce4 Ti 4200, which is superior to most of the cheapo 3D cards you get. It's fast and has great driver (for Windows and Linux!) to boot. Compare this to a Xabre 400 card, which runs only $30 cheaper, but in many games offers half the performance. Those who are really going to feel that $30 hit can easily save another $10 and buy a GeForce4 MX 440 for $60. It's still faster than the Xabre 400, and even though it lacks pixel shaders, (1) at that price, you're not playing Doom III, (2) The Xabre's pixel shaders are so bad, for all intents and purposes, it doesn't have them.
PCI is about 1 GBit per second. An external USB2 videocard would be impossible because of the overhead and latency of a USB connection.
[ ` &. All with my keyboard drawer closed. Frankly, I still don't know what keys they're on, but I know where to put my fingers.
True. For the last week or so, I've been using a keyboard with the keyboard drawer closed because I'm in too constrained of a space to sit back all the way. Those little nubs under your index fingers are there for a reason, you know.
Actually, my Inspiron has the exact same graphics card, and it came out months ago. And even that has already been surpassed by the Radeon 9000.
Ha ha. Even the Java folks realized the power of generics. There are lots of things (take a look at Loki or Boost) that simply cannot be done in static languages like C/C++ and Java without something like the template mechanism. Remember, C++ isn't an OOP language, it is a multi-paradigm language. There are lots of cases (and data structures is a big one) where "proper OOP" sucks. Just look at the cast-fest that is the Java collection classes. One of the major powers of the template mechanism is that it allows high level design with low-level performance. And there are *lot's* of people (especially in graphics, science, systems-programming) that really do need low-level performance. Generic data structures without templates usually require either abandoning type safety (by using void pointers, the C way to do it) or abandoning performance (by using inheritence and virtual functions, the Java way to do it). With templates, you don't have to abandon either. In practice, the STL list class is about 50% the speed of a single-type C data structure. Very impressive considering the C structure has embedded link nodes and no copy on insert while the STL does. Vector and matrix classes using templates in scientific applications are often 100% the speed of an equivilent C structure, because of the fully optimizable nature of templates. Of course, like everything else in C++, templates are easy to abuse and use the wrong way. But they are hardly unnecessary and redundant.
In proper implementations, the STL doesn't add all that much overhead to a binary. A good linker will remove a lot of redundant code caused by the STL, and a good compiler will process STL code quite quickly. Intel C++, for example, handles the STL quite well. My first STL project was a physics simulation that was extremely data structure intensive. It took me at most a day to get the hang of things. The STL really is quite straightforward. Praytell, how would you simpify the STL string or list classes?
I find it different. I find it impossible to remember the totally bland, generic, and similar sounding names in Java, while I find the C++ names very concise and memorable.
The Bill of Rights was ignored by some of the contemporaries of the founding fathers. The tradition of being much more conservative about economic entities than individuals goes back to the time of the founding fathers and Hamilton's views on economics. The Bill of Rights are great for what it is applied to, individuals, but an economy is a different beast than society as a whole. And because you insist on turning this into a liberal vs conservative arguement, I will point out that one of the principal weaknesses of conservative ideology is the tendency to adhere so strongly to a set of logical principles, that the inherent complexity of society and government is simplified away and ideas start to be applied to situations where they aren't applicable.
Retarded naming scheme, small size, and wierd design of standard library. The damn thing's almost useless.
>>>>>>>>
I love the C++ STL naming scheme. Nice and concise, but still clear and memorable. Java and many other languages suffer far too much from name bloat. And the C++ STL is probably one of the last things a C++ programmer really understands, and it can be years before people truely appreciate the power of the STL.
It doesn't have anything to do with liberals or free speech. The US government long ago made the decision that the economy is to be closely watched, and corporations are distinct from individuals. Hence you have all sorts of laws (anti-trust, monopoly, SEC, trading, advertising) that go directly against many of the amendments. Look at it this way: corporations are not allowed to band together and engage in price fixing. Individuals are perfectly free to band together and come to a mutual agreement for their group's benefit.
And a Porsche is just a car. Right...
I keep hearing "oh, these games aren't AAA titles, they must suck. Well, some of the most pure fun games I've played are obscure titles like these. They have nice graphics (not bleeding edge, but pretty) and fun gameplay. They not be the deep, sweeping experience of something like HalfLife, but they're still good entertainment. If the price is right, this games might be a good buy.
On Gentoo /etc/X11/XF86Config; find 'nv' change to 'nvidia'
emerge nvidia-kernel
emerge nvidia-glx
vi
Done.
You can do the same for ATIs drivers, I think.
I still have the 210MB WD that came with my Tandy 486 SX/33 which I bought in 1990. I was actually using it as a swap drive (with a 1.2GB WD as the main drive) in the same Tandy (now running as a firewall) until about a year ago. The damn thing (HD, computer and all) is still kicking after 13 years, even though it's spent a significant amount of time sitting unprotected in my garage.!
Well, let's see. First, ripping DVD to VOB isn't the same as ripping CD to WAV, since the DVD is compressed to begin with and the CD isn't. From there, I know ever people who have 100 DVDs or more (doesn't have to be illegal, thanks to box sets and directors cuts and the fact that even buying individually, 100 DVDs is only a $1500-$2000 investment). As for games, 600MB is svelte. I don't know games that come on one disk anymore. UT2003 has a 2GB install, and even a 2D RPG like Icewind Dale is 700MB+ Then throw in Windows XP and it's 1GB+ minimum install, and even 1TB is starting to seem small.
Yes, they could do that. They could also hire their own mercenary army to take over the Justice Department (to back up the lobbyists ;-). But will they?
>>>>>
Does it matter? Do you want the monkey holding the gun, whether or not it will shoot?
Because a lot of us are getting to the age where "going legit" seems like an attractive proposition. We would rather not run illegal software as our main environment. Besides, the Linux devs are respectable hackers, not a bunch of virus weenies.
The K6 never overtook the PII line in megahertz, which was what its real competitor was (chronologically). And while 3DNow! was significantly faster, it was no replacement for real FPU power (useful for scientific applications and 3D modeling) and was largely useless unless you were running one of the few applications directly coded for 3DNow!
http://www17.tomshardware.com/cpu/19980416/perform ance-02.html
In everything from Quake II to 3D Studio, the 233 MHz K6 is about half as fast as the 233 MHz PII. I don't know why the Sandra benchmarks are different, but they are the exception, not the rule. Besides, Sandra is an artificial benchmark, not a real world one like 3D Studio or SPEC.