But if we are to believe the media, not even Bin Laden expected the twin towers to collapse. How many more victims would you expect from hitting the WTC compared to, say, the Sears Tower in Chicago, if you think neither is going to crumble? I'll concede that the twin towers' height gave one advantage: they were easier targets to hit. Still the question remains why they didn't e.g. strike one tower and such a symbolic target as the Statue of Liberty. That would have been an even more effective way to get a twisted anti-freedoom message across.
Basically, he says that the 9/11 attacks were heinous crimes, that the islamic terrorists want to destroy free thought and liberty (no news flash there)
I have a news flash for you: the Statue of Liberty, arguably the most popular symbol of freedom and liberty worldwide, is right behind the corner from the WTC. Yet, the terrorists didn't bother with it. They rather flew TWO planes in the twin towers, which hardly were a symbol of liberty. Why?
If you were a terrorist that wanted "to destroy free thought and liberty", what would you target?
Eiffel remains inherently an object-oriented language, but in recent times it has borrowed some functionality typical of functional languages, through the new agent mechanism (think of iterators, map and for-all).
Kent Displays is already shipping LCD screens that are more reflective than other LCD technologies and keep their image for an indefinite amount of time when powered off. Unfortunately the refresh rate isn't that great.
The Linux Agenda VR3, to name just one example, is a PDA that uses a "64bit consumer processor", NEC's VR41xx. That's a MIPS64 CPU, although the firmware runs it in 32bit mode. The same processor is used in many MIPS-based WindowsCE PDAs, like the Casio Cassiopeia. Other 64bit consumer processors are Toshiba's TX-49 (another MIPS64) and SuperH's new SH-5, which has yet to ship.
Motorola warned people against a) assuming that all processors in the 68000 family would always address 24 bits at most and b) abusing the upper 8 bits of addresses to store extra data Yet, where did Mac developers learn the dirty trick above? From Apple's own system software (this is in the pre-MacII days).
Operations which benefit from knowing cache line sizes and using dcbz include 1) clearing memory regions (especially if you are going to access them again right after you are done clearing) 2) preventing unnecessary reads on the first write to a cache line, if you know you are going to replace anyway the entire contents of the line
Most developers won't have to worry about this problem, sure, but there are some that will have to tweak their compiler/library/optimised routine.
Code for desktop-class PowerPCs often assumes a 32-byte cache line size for operations like filling memory. That includes some of Apple's own code, too. Unfortunately such an assumption doesn't apply to some embedded PPCs (16-byte line size) or to the RS64/POWER3/POWER4 (64- and 128-byte line size). This is probably the one serious flaw in the PowerPC architecture.
Some code, especially the fine-tuned bits using a few specific PPC instructions to improve performance, WILL need to be tweaked, unless IBM reworks their Power4 derivative to use 32-byte lines - probably not a good idea.
If you want faster disk storage, then just wait a few more months for the Xserve RAID or whatever it was called. Rumour has it that Apple has already been restricted from exporting it to China, Iraq and Cuba. Saddam Hussein has the same problems as you: he got some heavy number crunching from the G4s he sneaked in his country, but found out that he can't store as much output and as fast as he needs! And wait for Apple's benchmarks comparing it to a Commodore datassette and a ST-506 hard disk!
The quantity of data being processed in genomics, although not irrelevant, is the lesser of evils. The worse part is that, as a rule of thumb, the size of the databases involved seems to grow faster than Moore's law.
Uh, why would that be a troll? Did I just offend a staunch pro-Jobs zealot?
Maybe somebody needs to go back to Newton OS, take a close look at it and then explain why OS X isn't technically a step back. Start a real debate, don't hide cowardly behind your moderation points. Things like Newton OS and the Dylan language are possibly Apple's greatest innovations ever, but it looks like Jobs and his devotees would rather promote the frivolous stuff.
Between OS X, Apple's core technologies and the iApps, they have the resources and technologies to truly extend the modern desktop computing experience to the mobile market.
Imagine an elegantly designed handheld computer running a stripped down version of OS X.
Now that's silly. Ditching NewtonOS, to replace it with a stripped down version of NeX... oops, OS X, is a step back. No matter how much eye candy you put in, no matter the fact that a microkernel is being used, OS X remains a glorified mutation of your average, backward Unix-like OS. It makes sense only inside Jobs' Reality Distortion Field.
If anything, Apple should have scaled NewtonOS up, but of course Jobs would never develop something he can't claim as his own creature, no matter how good or brilliant it is.
mpg123, which did the actual decoding, is no longer there. And if you check the logs of the xmms RPM, you'll see that on Aug 20 they removed the mpg123 plugin.
No hardware support for memory protection needed?!?
System/38-AS/400 have always relied on custom hardware to prevent pointers from being forged. This is done with a so-called tag bit, which is stored in memory along with error correction bits. OS/400 doesn't run on just any POWER or PowerPC chip, it requires one which supports the AS extensions to run in tags-active mode (you could say that it makes them 65bit CPUs). Such extensions are used to set/check the tag (a privileged operation) or to collect/restore the tags from/to a page all at once - to speed up things when swapping the page to or from disk.
To be fair, I am amazed as what passes for pizza in most places in Italy as well. There's been quite some activity in recent times to preserve the "true" pizza against bastardisation. For example, there's a standard and its certification mark. Efforts are going on to make it a national and European standard (if that hasn't happened yet). This way, a consumer knows what to expect when seeing a protected name. Same is happening with a lot of traditional food in Europe - at least the Union is giving some tangible benefits in return.
The other big offender is mozzarella. True mozzarella is made from water buffalo milk, has a porcelaine white colour and tastes nothing like the tennis shoe gum with the same name found in the US or the cheap imitation from cow milk sold even in most of Italy.
And don't get me started with so-called parmesan, which has little do with the real parmigiano reggiano...
OS/400 has always used 128bit pointers (since they started working on the AS/400's past incarnation, the System/38, in the late 70's). Only 64bits are actually used for addressing, though.
Re:IBM should have one-upped everybody
on
PowerPC Goes 64 bit
·
· Score: 2, Funny
If this chip is going to be used in iSeries systems, then it's actually going to be a 65-bit CPU for real, at least when run in tags-active ("OS/400") mode. Careful what you joke about!
IBM has alread licensed Altivec from Motorola as far back as 1998-2000, allegedly to help with the design of the PowerPC variant used in the Gamecube. The PowerPC 405 embedded processor in the Gamecube contains 38 additional instructions for vector FP math (vs. the 162 in Altivec).
IBM doesn't need an Altivec license for the Gekko extensions. As you point out, Gekko can treat a single 64bit FPU register as a pair of single-precision numbers. Altivec, instead, uses an additional set of 128bit registers. Gekko's paired-single extensions target one particular application, 3D graphics, and are pretty much like the MIPS3D extensions that SGI created some ten years ago. Altivec is much broader and is more like MIPS' MDMX (MaDMaX) extensions on steroids. Pretty much everything is done differently; even the programming model for condition codes is not the same.
The Dreamcast doesn't use the SH3, it's the Japanese Zaurus PDAs and a few WindowsCE PDAs that use it. The SH3 would be utterly useless for the Dreamcast, because it lacks a FPU. Instead, Sega used the SH4, whose FPU kicks ass.
By the way, SuperH (a separate company started by STMicro and Hitachi) introduced a few months ago the SH5, which introduces a new 64bit architecture (both for registers AND addressing).
Well, I think I might have heard somewhere that he actually said something nice about Algol, but I can't find a quote anywhere.
Erm, he co-designed Algol 60 and wrote its first implementation. He also appreciated Hoare's ideas for Algol W and rejected (like Hoare and Wirth) the monster called Algol 68.
No, I RYFT(roll) instead. You said that my post proved the bigger picture. Wrong. I even agree with you that the "you-idiot-don-t-get-it-why-don-t-you-go-back-to-w indoze" attitude stinks, but obviously in this case the original poster hasn't even attempted to look for the answer, which is one of the very first things the documentation clearly mentions - had it been otherwise, he would have had all the reasons in the world to complain.
Name another engine with realtime and cooperative editing. And if you think you can do better, go ahead and write that "GPL'd 3D engine that gets masses moving".
You are seriously missing the point, anyway. Cube doesn't want to be the ultimate, perfect engine with a ton of features. It doesn't want to please everyone, either. It's an attempt to do things in a certain way, out of intellectual curiosity. Don't you ever wonder "well, what happens if I followed THIS one approach thoroughly?". What matters the most is where you end up and what you learnt in the process.
Yes, just for the sake of featuritis you could do things differently if you really wanted to, but then it wouldn't be the same exercise. You would be missing the initial point altogether... pretty much like asking why that fuel-efficient car can't do 0-100Km/h as fast as a BMW does.
I'm far from being representative of the "Linux community".
Yet, one of the very first things Cube's Readme says is:
'note well: the engine is still in beta stages, and also VERY different from any engine you have seen before. Failure to read the documentation in its entirety may cause you to:
* miss out on the cool features.
* run it in an unoptimal way for your system.
* conclude it "sucks" prematurely.'
The -t option can be found simply by following the first link in readme.html, clearly labelled with "config: on running the game and configuring it for your machine". The link comes even before the blurb I quoted above. It's not like it is buried in an obscure file somewhere.
I usually go out of my way to help people and whine a lot myself when there's missing, cryptic or misleading information, but in this case it definitely would have taken less time for the original poster to just look at readme.html than to post on Slashdot (at the very least he had to wait the infamous 20 seconds).
Slashdot Mirror
But if we are to believe the media, not even Bin Laden expected the twin towers to collapse. How many more victims would you expect from hitting the WTC compared to, say, the Sears Tower in Chicago, if you think neither is going to crumble? I'll concede that the twin towers' height gave one advantage: they were easier targets to hit.
Still the question remains why they didn't e.g. strike one tower and such a symbolic target as the Statue of Liberty. That would have been an even more effective way to get a twisted anti-freedoom message across.
If you were a terrorist that wanted "to destroy free thought and liberty", what would you target?
Eiffel remains inherently an object-oriented language, but in recent times it has borrowed some functionality typical of functional languages, through the new agent mechanism (think of iterators, map and for-all).
Huh?!?!? For the record, the Eiffel language is almost 20 years old.
Or just use the -a switch with a ping from a recent version of iputils.
Kent Displays is already shipping LCD screens that are more reflective than other LCD technologies and keep their image for an indefinite amount of time when powered off. Unfortunately the refresh rate isn't that great.
The Linux Agenda VR3, to name just one example, is a PDA that uses a "64bit consumer processor", NEC's VR41xx. That's a MIPS64 CPU, although the firmware
runs it in 32bit mode. The same processor is used in many MIPS-based WindowsCE PDAs, like the Casio Cassiopeia. Other 64bit consumer processors are Toshiba's TX-49 (another MIPS64) and SuperH's new
SH-5, which has yet to ship.
Motorola warned people against
a) assuming that all processors in the 68000 family would always address 24 bits at most and
b) abusing the upper 8 bits of addresses to store extra data
Yet, where did Mac developers learn the dirty trick above? From Apple's own system software (this is in the pre-MacII days).
Operations which benefit from knowing cache line sizes and using dcbz include
1) clearing memory regions (especially if you are going to access them again right after you are done clearing)
2) preventing unnecessary reads on the first write to a cache line, if you know you are going to replace anyway the entire contents of the line
Most developers won't have to worry about this problem, sure, but there are some that will have to tweak their compiler/library/optimised routine.
There's one little problem you aren't mentioning.
Code for desktop-class PowerPCs often assumes a 32-byte cache line size for operations like filling memory. That includes some of Apple's own code, too. Unfortunately such an assumption doesn't apply to some embedded PPCs (16-byte line size) or to the RS64/POWER3/POWER4 (64- and 128-byte line size). This is probably the one serious flaw in the PowerPC architecture.
Some code, especially the fine-tuned bits using a few specific PPC instructions to improve performance, WILL need to be tweaked, unless IBM reworks their Power4 derivative to use 32-byte lines - probably not a good idea.
If you want faster disk storage, then just wait a few more months for the Xserve RAID or whatever it was called. Rumour has it that Apple has already been restricted from exporting it to China, Iraq and Cuba. Saddam Hussein has the same problems as you: he got some heavy number crunching from the G4s he sneaked in his country, but found out that he can't store as much output and as fast as he needs! And wait for Apple's benchmarks comparing it to a Commodore datassette and a ST-506 hard disk!
The quantity of data being processed in genomics, although not irrelevant, is the lesser of evils. The worse part is that, as a rule of thumb, the size of the databases involved seems to grow faster than Moore's law.
Uh, why would that be a troll? Did I just offend a staunch pro-Jobs zealot?
Maybe somebody needs to go back to Newton OS, take a close look at it and then explain why OS X isn't technically a step back. Start a real debate, don't hide cowardly behind your moderation points. Things like Newton OS and the Dylan language are possibly Apple's greatest innovations ever, but it looks like Jobs and his devotees would rather promote the frivolous stuff.
Blessed ignorance.
Now that's silly. Ditching NewtonOS, to replace it with a stripped down version of NeX... oops, OS X, is a step back. No matter how much eye candy you put in, no matter the fact that a microkernel is being used, OS X remains a glorified mutation of your average, backward Unix-like OS. It makes sense only inside Jobs' Reality Distortion Field.
If anything, Apple should have scaled NewtonOS up, but of course Jobs would never develop something he can't claim as his own creature, no matter how good or brilliant it is.
mpg123, which did the actual decoding, is no longer there. And if you check the logs of the xmms RPM, you'll see that on Aug 20 they removed the mpg123 plugin.
No hardware support for memory protection needed?!?
System/38-AS/400 have always relied on custom hardware to prevent pointers from being forged. This is done with a so-called tag bit, which is stored in memory along with error correction bits. OS/400 doesn't run on just any POWER or PowerPC chip, it requires one which supports the AS extensions to run in tags-active mode (you could say that it makes them 65bit CPUs). Such extensions are used to set/check the tag (a privileged operation) or to collect/restore the tags from/to a page all at once - to speed up things when swapping the page to or from disk.
To be fair, I am amazed as what passes for pizza in most places in Italy as well. There's been quite some activity in recent times to preserve the "true" pizza against bastardisation. For example, there's a standard and its certification mark. Efforts are going on to make it a national and European standard (if that hasn't happened yet). This way, a consumer knows what to expect when seeing a protected name. Same is happening with a lot of traditional food in Europe - at least the Union is giving some tangible benefits in return.
I don't know about the rest of the US, but if you ever end up in the Twin Cities, I have found a place in St. Paul that is certified (there are just two or three hundred in the world and most of them are of course in Naples, Italy). There is more background available online about pizza.
The other big offender is mozzarella. True mozzarella is made from water buffalo milk, has a porcelaine white colour and tastes nothing like the tennis shoe gum with the same name found in the US or the cheap imitation from cow milk sold even in most of Italy.
And don't get me started with so-called parmesan, which has little do with the real parmigiano reggiano...
OS/400 has always used 128bit pointers (since they started working on the AS/400's past incarnation, the System/38, in the late 70's). Only 64bits are actually used for addressing, though.
If this chip is going to be used in iSeries systems, then it's actually going to be a 65-bit CPU for real, at least when run in tags-active ("OS/400") mode. Careful what you joke about!
Windows NT for PowerPC? AFAIK it ran in little-endian mode.
IBM doesn't need an Altivec license for the Gekko extensions. As you point out, Gekko can treat a single 64bit FPU register as a pair of single-precision numbers. Altivec, instead, uses an additional set of 128bit registers. Gekko's paired-single extensions target one particular application, 3D graphics, and are pretty much like the MIPS3D extensions that SGI created some ten years ago. Altivec is much broader and is more like MIPS' MDMX (MaDMaX) extensions on steroids. Pretty much everything is done differently; even the programming model for condition codes is not the same.
The Dreamcast doesn't use the SH3, it's the Japanese Zaurus PDAs and a few WindowsCE PDAs that use it. The SH3 would be utterly useless for the Dreamcast, because it lacks a FPU. Instead, Sega used the SH4, whose FPU kicks ass.
By the way, SuperH (a separate company started by STMicro and Hitachi) introduced a few months ago the SH5, which introduces a new 64bit architecture (both for registers AND addressing).
No, I RYFT(roll) instead. You said that my post proved the bigger picture. Wrong. I even agree with you that the "you-idiot-don-t-get-it-why-don-t-you-go-back-to-w indoze" attitude stinks, but obviously in this case the original poster hasn't even attempted to look for the answer, which is one of the very first things the documentation clearly mentions - had it been otherwise, he would have had all the reasons in the world to complain.
You are seriously missing the point, anyway. Cube doesn't want to be the ultimate, perfect engine with a ton of features. It doesn't want to please everyone, either. It's an attempt to do things in a certain way, out of intellectual curiosity. Don't you ever wonder "well, what happens if I followed THIS one approach thoroughly?". What matters the most is where you end up and what you learnt in the process.
Yes, just for the sake of featuritis you could do things differently if you really wanted to, but then it wouldn't be the same exercise. You would be missing the initial point altogether... pretty much like asking why that fuel-efficient car can't do 0-100Km/h as fast as a BMW does.
I'm far from being representative of the "Linux community".
Yet, one of the very first things Cube's Readme says is:
'note well: the engine is still in beta stages, and also VERY different from any engine you have seen before. Failure to read the documentation in its entirety may cause you to:
* miss out on the cool features.
* run it in an unoptimal way for your system.
* conclude it "sucks" prematurely.'
The -t option can be found simply by following the first link in readme.html, clearly labelled with "config: on running the game and configuring it for your machine". The link comes even before the blurb I quoted above. It's not like it is buried in an obscure file somewhere.
I usually go out of my way to help people and whine a lot myself when there's missing, cryptic or misleading information, but in this case it definitely would have taken less time for the original poster to just look at readme.html than to post on Slashdot (at the very least he had to wait the infamous 20 seconds).