Indeed. While there are challenges in 3d manufacturing, there's nothing physically prohibiting it. Other options for improving performance include quantum computing (for specialized tasks only, mind you), and massive parallelism. The latter example is really helped by the increasing availability of programming features to make parallelism easier, such as futures and promises. Aka, stuff like:
auto a = fork(function1); auto b = fork(function2); auto c = fork(function3);
auto d = a + b + c;
Aka, thread function1, thread function2, thread function3, and then when all of them return, sum their results. Trying to do that sort of stuff with ordinary pthreads is so much more code, effort, and potential for errors.
Perhaps the fact that basically everyone has been marching under the banner of and shouting for democracy? Perhaps that as soon as a single crack in Gadhaffi's leadership broke, nearly the entire country rose up against him?
First, off, re. history: the US never armed al-Qaeda. The US armed the Mujahadeen. A very small fraction of the Mujahadeen ended up forming al-Qaeda after the war, with the overwhelming majority of the al-Qaeda membership coming from other parts of the arab world who had no involvement with the war in Afghanistan. Their main gripes were the US support of Israel and the US's support of their brutal, repressive dictators. We're doing the opposite of the latter one right now.
Please explain how your "rebels breaking the ceasefire" scenario matches up with hospitals getting shelled in Misurata and tanks trying to break into the city in Benghazi.
And "tyrant" does not belong in quotes. It's pretty indisputable that he's a tyrant.
Wow, your Bible sure has some elaborate secret messages hiding between the seemingly obvious plain text saying "judge not". Is it in lemon juice and you have to hold it up to a source of heat to reveal it?
I fail to see the correlation, especially since my ring finger is shorter than my index finger.
1) Do we really have to explain the difference between correlation and certainty? By your logic, you're neither male nor female, since *nobody* statistically has a shorter ring finger than index finger (men average a shorter index than ring, while women have the same size). 2) I seriously doubt you're using the measuring protocol, which involved a flatbed scanner and many techniques to control for measuring bias and how the hand is held outstretched.
Yes, statistically, gay men average higher testosterone levels. So do lesbians. Trans men (FTM) average significantly higher testosterone levels than cis women -- I've read around five studies on this, only one found a weak linkage, the rest found strong linkages, and one study showed an amazing half of all transmen with either PCOS or other hyperandrogenicity conditions. Trans women (MTF) do not average unusual estrogen or testosterone levels.
Note that these are all studies on adults; this says nothing about prenatal conditions.
Hormone levels do not seem to be the driving factor, however. The driving factor seems to lie in particular hypothalamic nuclei. As you may or may not know, the hypothalamus is a primitive structure that acts as the linkage between the brain and the endocrine system by controlling the pituitary (master) gland. There have been quite a few studies showing that it strongly affects sexuality, sexual behavior, and gender identity. INAH3 and BSTc in particular are amazingly strongly linked to gender identity. In various animal studies, inducing certain types of hypothalamic lesions has been shown to make ferrets exhibit homosexual behavior, while others have destroyed the ability for normal male copulatory behavior to occur in other lab animals.
There are a variety of other linkages that have been shown, such as responses to aromatic sex steroids, white matter microstructure, and whatnot. At the same time, this should not be interpreted as "MTFs are born with a female brain; FTMs are born with a male brain". Most of the brain structure matches their anatomical sex. There are just a few small regions that do not. However, once hormone replacement begins, the majority brain is radically restructured (even whole-brain size) in the direction of the target sex. Most TS brain structure linkages cannot be studied by MRIs, due to their tiny sizes (they require dissection); however, the changes from hormone replacement are very measurable on MRI images.
FYI: St. Thomas Aquinas felt that that passage was about women having non-vaginal intercourse.
And any way, you might notice that even the homosexuality among men (and among women, if you disagree with Aquinas) is listed as an *effect* of God's wrath, not a cause. And lastly, this is the one prominent person in the NT who never personally knew Jesus who's talking anyway. Jesus, the guy who told the parable of the Good Samaritan, who hung out with prostitutes, tax collectors, and other social outcasts, who said judge not lest ye be judged, the guy that "whoever believes in him should not perish, but have eternal life" (notice the lack of conditions there).
"There is neither Jew nor Greek, there is neither slave nor free, there is no male and female, for you are all one in Christ Jesus. " -- Galatians 3:28.
Yeah, some ceasefire when you keep bombing and shelling cities.
You seriously take this guy at his word? After all the absurd stuff about al-Qaeda and people hopped up on hallucinatory drugs (with the videos of prescription pain killers taken from a hospital as "evidence"), in between the talks of a "cleansing"?
The UN isn't referring to the Libyan people as "cockroaches" and ordering their military to go on a door-to-door "cleansing".
We're not going to put anyone in charge. We're not going to have anyone on the ground. What we will be doing (ostensibly purely as a side effect, although almost certainly by intent) is clearing the path for the rebels to do what they will. Which will almost certainly be the creation of a democracy (however flawed of one).
"A minority attacking a standing government"? It was a mass uprising against an incredibly brutal regime who they knew would gun them down (but they did it anyway), an uprising that nearly took the entire country before Gadaffi got his largely mercenary army back together and started shelling entire cities to take them back from the people. "Some military personnel"? This was largely a civilian revolt (which is the main reason they're so poorly trained; they have lots of equipment but don't know how to use it effectively; it took them over a week to figure out how to get their first airplane off the ground, and it got shot down the next day).
Most of the power generation and distribution hardware on each side is just fine still. However, perhaps this will be the impetus for them to start a slow project of national standardization, migrating the dividing line a bit in one direction or the other every year.
Where do you get that? I just did a search for HVDC link construction times, and ran into this, which cites the time to build the whole Cross-Sound Cable (CSC) project, which involved two terminals and a 40km submarine cable to transmit 330MW HVDC, at nine months. Sure as heck beats building a new nuclear power plant or whatnot.
I imagine the limiting factor will be global high-power thyristor production and stocks.
This earthquake/tsunami/meltdown/etc could be a Catastrotunity in that regard -- finally providing the impetus to modernize their grid. Laying new power lines should be far faster than building new power plants, and since we're talking high power/long distance and they'll need to match frequencies, I would expect that they'll be HVDC.
Another thing that they should be able to do faster than building new thermal power plants is to build power storage facilities to buffer day/night demand (battery storage, mini pumped-hydro, etc). China already uses these for demand buffering quite extensively. But they have a nice side effect of also helping support more intermittent power generation as well, because there is little difference between buffering supply and buffering demand. Which is great, because installing new photovoltaic capacity is also much faster than building new centralized thermal power plants (at least if global solar production can keep up).
If speed of getting new power into the region is of the essence, they may well end up with a very modern, very green grid purely as a side effect.
How exactly do you think this failure mode is a desirable one? There are spent fuel ponds containing more fuel than in the core itself which are in various states of overheating, half of the buildings have been gutted of sensors, pumps, cranes, etc by hydrogen explosions, the core is cracked on the #2, god only knows what's happening in the common fuel pool, the radiation level has been high enough to drive away *helicopters flying high overhead* some days, etc. You call that "reasonably safe"? God forbid we get a recriticality in a spent fuel pool. The explosions left half of them sitting exposed to the air and full of debris.
This is an INES level 6 disaster, same as the Kyshtym disaster that caused the Soviets to quietly have to remove 30 towns from their maps. The US has ordered a 50 mile exclusion zone for Americans around Fukushima Daiichi -- and the disaster is still unfolding. For comparison, it's under 40 miles from the reactors of Indian Point to Times Square, and far less to the outer reaches of NYC.
And, FYI, you're applying an observation bias to your "wrong thing at every decision point" argument. if they had done the right thing, you wouldn't have heard about it. There are nuclear accidents all the time. Nearly every reactor has had accidents of varying severity. Only the bigger ones (generally INES-5 and above) make the news. Naturally those are going to be the ones where more than one thing went wrong. And, FYI, both in Chernobyl and TMI, there were many *right* decisions made also. They simply didn't outweigh all of the wrong decisions. And many "wrong decisions" occur during the engineering stage, not the operation stage, but aren't known about until they reveal themselves many years later.
1/1000 chance per year =.999 annualized chance of normal operation.999^(~436 reactors) = 1 in 3 annualized chance of meltdown somewhere in the world.
Clearly your numbers are off, but still, the point remains the same: when it comes to things with great potential for harm, you need a far higher standard than just 1:1000 chance of disaster. What's an acceptable rate of time for a 50% probability for an INES Cat. 6 event? 1:500 years? Each reactor would need to have an annualized INES Cat. 6 failure probability of 0.000317% (a 1 in 315,000 chance).
The good game engines are cross-platform; this is either done by being OpenGL only or, much more common, supporting both OpenGL and DirectX. The only good one that I can think of that isn't is the CryEngine series. I'm sure there are others, mind you.
No, what I'm doing is the equivalent of 'screaming' that the carpenter should be using a nail gun.
The "carpenter" (programmer) shouldn't be using OpenGL, and they shouldn't be using Direct3D. They should be using a rendering engine. CrystalSpace, TrueVision, OGRE, Irrlicht, CryEngine, Unreal Engine, whatever.
The fact that I'm having to bother to explain this is really sad. It means lots of people are out there reinventing the wheel.
Do you know what a rendering engine is? Look it up before you respond to a post. Developers shouldn't be developing to either Direct3D *OR* OpenGL. They should be developing to a rendering engine and only care what platforms the rendering engine supports. The rendering engine should take care of the implementation (good ones run on both OpenGL and Direct3D).
I actually dealt with a variety of fast square root and inverse square root functions and profiled them when creating a particle system for a MMORPG (Eternal Lands). That algorithm is indeed fast (at the cost of accuracy)... but it's still significantly slower than the hardware SSE inverse square root routine. Hence this function:
static float invsqrt(const float f)
{ #ifdef __SSE__// Hardware fast invsqrt.
typedef union match { __m128 m128; struct { float x, y, z, w; }; } match;
match f2;
f2.m128 = _mm_rsqrt_ss(_mm_set_ss(f));
return f2.x; #else// Quake fast invsqrt.
typedef union match { int i; float f; } match;
match tmp;
float half = 0.5f * f;
tmp.f = f;
tmp.i = 0x5f3759df - (tmp.i >> 1);
f = tmp.f;
f = f * (1.5f - half * f * f);
return f; #endif
};
I also found that caching values and interpolating between them for the pow() function can get a relevant speed increase (at notable cost to accuracy -- but for most 3d uses of pow, you don't care), but my efforts creating a sin cache didn't pay off, as it didn't improve performance any relevant amount.
But back to the original issue: don't forget about __SSE__ (and later) instructions! There's some great ones, but a lot of the benefit only comes if you implement your class's members as __SSE__ (or other) data structures, bundling together multiple variables which are frequently used together. The performance increase on base classes (say, 3d vectors or RGBA colors) can be huge. The key is that you still have to have a naive implementation for hardware that doesn't whatever extensions you use. A good library will have half a dozen or more implementations of such a class for different hardware.
One of the many reasons why it's best to not reinvent the wheel and just use existing library functions where available; widely used libraries already take care of all of this stuff for you.
DirectX, OpenGL.... why should the end user care? Unless you're developing a rendering engine, you should probably just be using a rendering engine not caring about the underlying mechanics. I mean, how many times should we have to implement stencil shadows from scratch, mesh loading from scratch, paged geometry from scratch, and on and on? Odds are, if more than just a handful of people need a piece of functionality, a good rendering engine probably already has it implemented -- and a better, faster, more capable, more bugfree version than you'd write. And if not? Write it on your own and then contribute your code.
It's like using STL versus not. Sure, you *could* go and implement STL container functionality on your own, but why would you want to?
Erm, looks like I grabbed the wrong spectrum. 420-440, not 400-420. Let's see: I found a different page that lists "satellites, Pave Paws radar systems, radio beacons, military and Amateur Radio operators." I double checked PAVE PAWS (the radar system designed to detect and track ICBMs and satellites), and indeed, it's 435mhz. The radar installations are bloody huge, so I hope that if this passes, they can be reconfigured.
Slashdot Mirror
Indeed. While there are challenges in 3d manufacturing, there's nothing physically prohibiting it. Other options for improving performance include quantum computing (for specialized tasks only, mind you), and massive parallelism. The latter example is really helped by the increasing availability of programming features to make parallelism easier, such as futures and promises. Aka, stuff like:
auto a = fork(function1);
auto b = fork(function2);
auto c = fork(function3);
auto d = a + b + c;
Aka, thread function1, thread function2, thread function3, and then when all of them return, sum their results. Trying to do that sort of stuff with ordinary pthreads is so much more code, effort, and potential for errors.
Not to mention that homosexuality is widespread in the animal kingdom anyway. So unless they have some other definition of "natural"...
Perhaps the fact that basically everyone has been marching under the banner of and shouting for democracy? Perhaps that as soon as a single crack in Gadhaffi's leadership broke, nearly the entire country rose up against him?
Oh, wait, that's right, ragheads don't want and/or can't handle democracy, right?
First, off, re. history: the US never armed al-Qaeda. The US armed the Mujahadeen. A very small fraction of the Mujahadeen ended up forming al-Qaeda after the war, with the overwhelming majority of the al-Qaeda membership coming from other parts of the arab world who had no involvement with the war in Afghanistan. Their main gripes were the US support of Israel and the US's support of their brutal, repressive dictators. We're doing the opposite of the latter one right now.
Please explain how your "rebels breaking the ceasefire" scenario matches up with hospitals getting shelled in Misurata and tanks trying to break into the city in Benghazi.
And "tyrant" does not belong in quotes. It's pretty indisputable that he's a tyrant.
Wow, your Bible sure has some elaborate secret messages hiding between the seemingly obvious plain text saying "judge not". Is it in lemon juice and you have to hold it up to a source of heat to reveal it?
Right. Because what the world needs more than anything else is more reproducing people, right?
Catholic priests don't reproduce, either. Should we wipe out the Catholic Church, too?
I fail to see the correlation, especially since my ring finger is shorter than my index finger.
1) Do we really have to explain the difference between correlation and certainty? By your logic, you're neither male nor female, since *nobody* statistically has a shorter ring finger than index finger (men average a shorter index than ring, while women have the same size).
2) I seriously doubt you're using the measuring protocol, which involved a flatbed scanner and many techniques to control for measuring bias and how the hand is held outstretched.
Yes, statistically, gay men average higher testosterone levels. So do lesbians. Trans men (FTM) average significantly higher testosterone levels than cis women -- I've read around five studies on this, only one found a weak linkage, the rest found strong linkages, and one study showed an amazing half of all transmen with either PCOS or other hyperandrogenicity conditions. Trans women (MTF) do not average unusual estrogen or testosterone levels.
Note that these are all studies on adults; this says nothing about prenatal conditions.
Hormone levels do not seem to be the driving factor, however. The driving factor seems to lie in particular hypothalamic nuclei. As you may or may not know, the hypothalamus is a primitive structure that acts as the linkage between the brain and the endocrine system by controlling the pituitary (master) gland. There have been quite a few studies showing that it strongly affects sexuality, sexual behavior, and gender identity. INAH3 and BSTc in particular are amazingly strongly linked to gender identity. In various animal studies, inducing certain types of hypothalamic lesions has been shown to make ferrets exhibit homosexual behavior, while others have destroyed the ability for normal male copulatory behavior to occur in other lab animals.
There are a variety of other linkages that have been shown, such as responses to aromatic sex steroids, white matter microstructure, and whatnot. At the same time, this should not be interpreted as "MTFs are born with a female brain; FTMs are born with a male brain". Most of the brain structure matches their anatomical sex. There are just a few small regions that do not. However, once hormone replacement begins, the majority brain is radically restructured (even whole-brain size) in the direction of the target sex. Most TS brain structure linkages cannot be studied by MRIs, due to their tiny sizes (they require dissection); however, the changes from hormone replacement are very measurable on MRI images.
FYI: St. Thomas Aquinas felt that that passage was about women having non-vaginal intercourse.
And any way, you might notice that even the homosexuality among men (and among women, if you disagree with Aquinas) is listed as an *effect* of God's wrath, not a cause. And lastly, this is the one prominent person in the NT who never personally knew Jesus who's talking anyway. Jesus, the guy who told the parable of the Good Samaritan, who hung out with prostitutes, tax collectors, and other social outcasts, who said judge not lest ye be judged, the guy that "whoever believes in him should not perish, but have eternal life" (notice the lack of conditions there).
"There is neither Jew nor Greek, there is neither slave nor free, there is no male and female, for you are all one in Christ Jesus. " -- Galatians 3:28.
Yeah, some ceasefire when you keep bombing and shelling cities.
You seriously take this guy at his word? After all the absurd stuff about al-Qaeda and people hopped up on hallucinatory drugs (with the videos of prescription pain killers taken from a hospital as "evidence"), in between the talks of a "cleansing"?
The UN isn't referring to the Libyan people as "cockroaches" and ordering their military to go on a door-to-door "cleansing".
We're not going to put anyone in charge. We're not going to have anyone on the ground. What we will be doing (ostensibly purely as a side effect, although almost certainly by intent) is clearing the path for the rebels to do what they will. Which will almost certainly be the creation of a democracy (however flawed of one).
"A minority attacking a standing government"? It was a mass uprising against an incredibly brutal regime who they knew would gun them down (but they did it anyway), an uprising that nearly took the entire country before Gadaffi got his largely mercenary army back together and started shelling entire cities to take them back from the people. "Some military personnel"? This was largely a civilian revolt (which is the main reason they're so poorly trained; they have lots of equipment but don't know how to use it effectively; it took them over a week to figure out how to get their first airplane off the ground, and it got shot down the next day).
Baghdad Bob, is that you?
Most of the power generation and distribution hardware on each side is just fine still. However, perhaps this will be the impetus for them to start a slow project of national standardization, migrating the dividing line a bit in one direction or the other every year.
Who is "Sighonara"?
Where do you get that? I just did a search for HVDC link construction times, and ran into this, which cites the time to build the whole Cross-Sound Cable (CSC) project, which involved two terminals and a 40km submarine cable to transmit 330MW HVDC, at nine months. Sure as heck beats building a new nuclear power plant or whatnot.
I imagine the limiting factor will be global high-power thyristor production and stocks.
This earthquake/tsunami/meltdown/etc could be a Catastrotunity in that regard -- finally providing the impetus to modernize their grid. Laying new power lines should be far faster than building new power plants, and since we're talking high power/long distance and they'll need to match frequencies, I would expect that they'll be HVDC.
Another thing that they should be able to do faster than building new thermal power plants is to build power storage facilities to buffer day/night demand (battery storage, mini pumped-hydro, etc). China already uses these for demand buffering quite extensively. But they have a nice side effect of also helping support more intermittent power generation as well, because there is little difference between buffering supply and buffering demand. Which is great, because installing new photovoltaic capacity is also much faster than building new centralized thermal power plants (at least if global solar production can keep up).
If speed of getting new power into the region is of the essence, they may well end up with a very modern, very green grid purely as a side effect.
How exactly do you think this failure mode is a desirable one? There are spent fuel ponds containing more fuel than in the core itself which are in various states of overheating, half of the buildings have been gutted of sensors, pumps, cranes, etc by hydrogen explosions, the core is cracked on the #2, god only knows what's happening in the common fuel pool, the radiation level has been high enough to drive away *helicopters flying high overhead* some days, etc. You call that "reasonably safe"? God forbid we get a recriticality in a spent fuel pool. The explosions left half of them sitting exposed to the air and full of debris.
This is an INES level 6 disaster, same as the Kyshtym disaster that caused the Soviets to quietly have to remove 30 towns from their maps. The US has ordered a 50 mile exclusion zone for Americans around Fukushima Daiichi -- and the disaster is still unfolding. For comparison, it's under 40 miles from the reactors of Indian Point to Times Square, and far less to the outer reaches of NYC.
And, FYI, you're applying an observation bias to your "wrong thing at every decision point" argument. if they had done the right thing, you wouldn't have heard about it. There are nuclear accidents all the time. Nearly every reactor has had accidents of varying severity. Only the bigger ones (generally INES-5 and above) make the news. Naturally those are going to be the ones where more than one thing went wrong. And, FYI, both in Chernobyl and TMI, there were many *right* decisions made also. They simply didn't outweigh all of the wrong decisions. And many "wrong decisions" occur during the engineering stage, not the operation stage, but aren't known about until they reveal themselves many years later.
1/1000 chance per year = .999 annualized chance of normal operation .999^(~436 reactors) = 1 in 3 annualized chance of meltdown somewhere in the world.
Clearly your numbers are off, but still, the point remains the same: when it comes to things with great potential for harm, you need a far higher standard than just 1:1000 chance of disaster. What's an acceptable rate of time for a 50% probability for an INES Cat. 6 event? 1:500 years? Each reactor would need to have an annualized INES Cat. 6 failure probability of 0.000317% (a 1 in 315,000 chance).
Great risk requires great caution.
It all depends on the situation. In a number of situations, stencil shadows look a lot better.
Anyway, a good rendering engine lets you pick between shadow types.
The good game engines are cross-platform; this is either done by being OpenGL only or, much more common, supporting both OpenGL and DirectX. The only good one that I can think of that isn't is the CryEngine series. I'm sure there are others, mind you.
No, what I'm doing is the equivalent of 'screaming' that the carpenter should be using a nail gun.
The "carpenter" (programmer) shouldn't be using OpenGL, and they shouldn't be using Direct3D. They should be using a rendering engine. CrystalSpace, TrueVision, OGRE, Irrlicht, CryEngine, Unreal Engine, whatever.
The fact that I'm having to bother to explain this is really sad. It means lots of people are out there reinventing the wheel.
Do you know what a rendering engine is? Look it up before you respond to a post. Developers shouldn't be developing to either Direct3D *OR* OpenGL. They should be developing to a rendering engine and only care what platforms the rendering engine supports. The rendering engine should take care of the implementation (good ones run on both OpenGL and Direct3D).
You know, it helps if you don't stop reading a post after the first sentence.
I actually dealt with a variety of fast square root and inverse square root functions and profiled them when creating a particle system for a MMORPG (Eternal Lands). That algorithm is indeed fast (at the cost of accuracy)... but it's still significantly slower than the hardware SSE inverse square root routine. Hence this function:
static float invsqrt(const float f)
{
#ifdef __SSE__
typedef union match { __m128 m128; struct { float x, y, z, w; }; } match;
match f2;
f2.m128 = _mm_rsqrt_ss(_mm_set_ss(f));
return f2.x;
#else
typedef union match { int i; float f; } match;
match tmp;
float half = 0.5f * f;
tmp.f = f;
tmp.i = 0x5f3759df - (tmp.i >> 1);
f = tmp.f;
f = f * (1.5f - half * f * f);
return f;
#endif
};
I also found that caching values and interpolating between them for the pow() function can get a relevant speed increase (at notable cost to accuracy -- but for most 3d uses of pow, you don't care), but my efforts creating a sin cache didn't pay off, as it didn't improve performance any relevant amount.
But back to the original issue: don't forget about __SSE__ (and later) instructions! There's some great ones, but a lot of the benefit only comes if you implement your class's members as __SSE__ (or other) data structures, bundling together multiple variables which are frequently used together. The performance increase on base classes (say, 3d vectors or RGBA colors) can be huge. The key is that you still have to have a naive implementation for hardware that doesn't whatever extensions you use. A good library will have half a dozen or more implementations of such a class for different hardware.
One of the many reasons why it's best to not reinvent the wheel and just use existing library functions where available; widely used libraries already take care of all of this stuff for you.
DirectX, OpenGL.... why should the end user care? Unless you're developing a rendering engine, you should probably just be using a rendering engine not caring about the underlying mechanics. I mean, how many times should we have to implement stencil shadows from scratch, mesh loading from scratch, paged geometry from scratch, and on and on? Odds are, if more than just a handful of people need a piece of functionality, a good rendering engine probably already has it implemented -- and a better, faster, more capable, more bugfree version than you'd write. And if not? Write it on your own and then contribute your code.
It's like using STL versus not. Sure, you *could* go and implement STL container functionality on your own, but why would you want to?
Erm, looks like I grabbed the wrong spectrum. 420-440, not 400-420. Let's see: I found a different page that lists "satellites, Pave Paws radar systems, radio beacons, military and Amateur Radio operators." I double checked PAVE PAWS (the radar system designed to detect and track ICBMs and satellites), and indeed, it's 435mhz. The radar installations are bloody huge, so I hope that if this passes, they can be reconfigured.