with faster bandwidth than present (but probably not 6 months from now) firewire
USB 2.0 might be faster on paper but it's really pushing the hardware that the original USB was designed to operate on. I fully expect that it actually operates at a lower speed in real life conditions - much like 802.11. (11Mbit/s - ya right... more like 5.)
Also, for many operations (like copying from HD to HD) that 400MBit will give one much more bandwith then USB would give even if it does operate at spec. WIth USB the data would go from the HD to the PC then back to the other HD. Your 480Mbit/s just dropped to 240. Firewire cuts the PC out of the loop so you get the full 400Mbit/s speed (plus lots of saved CPU time.)
Need I even mention that you can use more then one connection to a device? Gang a couple of 1394 chanels together and there is no way that USB is faster. That and you can pick up a 6 port 1394 card for under $100. In a bit more time, when Linux support if fully tested, 1394 could easily replace external SCSI for even the most demanding tasks.
there will be no reason for motherboard manufacturers to not put the interface out there
Except the obvious fact that this is an Intel technology that will probably require Intel silicon. On the other hand, Firewire is an open standard that many are supporting. For example:
VIA Technologies: http://www.cyrix.com/products/1394.htm
Of course there are many others like Sony and TI but the fact of the matter is that there are more Firewire options then USB 2.0. Unless Intel lets these companies create their own USB 2.0 chipsets I doubt that you'll see USB 2.0 widely supported. There is simply no reason to support it when 1394 is a shipping, open standard that is already present on most digital camcorders. Add this to the fact that 1394 also has a great deal of room to grow (ie, 1600 Mbit/s) and it's not a hard decision to make.
Second, a system with good Firewire support is makes it just as easy to plugin a Firewire device and go as it is for a USB device.
Actually Firewire has much better plug-and-play support. With USB you need a host CPU / OS / driver combination in order to comunicate. Firewire does it all in the controller so it is far less driver dependent. Simpler, generic drivers work for most devices.
Want to see what Tom thinks? Check out the following web address. He also comes to the conclusion that Firewire has better pnp support.
Most TV shows / movies are saved. The really good ones even get transferred to new types of playback media. They also provide a much better way for us to really see how life was like.
Words are great but their meanings can change over the years. Being able to hear words and see the situations in which they are used can give a much better overall representation of how society really was. (Granted you need to be selective with what you watch.;)
Those extra transitions and context switches have a definite cost. A CPU designed with that in mind can help (a lot!), but on current hardware, it can be a nightmare. That's why a monokernel tends to perform better in practice.
Is this not less of an issue with newer CPUs? Just look at the rate at which the CPU/bus speed ratio is increasing. Much of the time a modern CPU spends is spent waiting for data from memory (granted this depends on the application.) If the CPU has to spend some extra time switching between rings then it shouldn't have much of an overall impact on speed - much of it's time is spent waiting anyway. This will only increase in the future.
The way I see it, the microkernel approach is much better in theory because it allows one to easily upgrade hardware architectures. Now in reality, we're stuck with 8x86 (mainly because of Microsoft.) In an environment where the vast majority of systems contain a single CPU of an architecture that doesn't change (much) then the monolithic kernels start to look real good. I would however prefer a world where we were not tied to legacy architectures and could jump from one to another with a simple microkernel rewrite (ya, I know, I'll keep dreaming.;)
Linux is definitely well designed for todays systems but I would still argue that the microkernel philosophy in superior in theory. Too bad the real world isn't so easy to theorize.
This is not a problem associated with genetic engineering but with farming in general. Where there used to be hundreds if not thousands of different grains grown in North America there are now but a fraction of those available. Those that produce are reproduced and the others go extinct taking with them their genetic traits. We've imposed evolution based on profitability - not the species' ability to survive.
If anything GE well help preserve genetic diversity because at least we know the seeds will be available in some lab if required in the future. Not a good solution but better then letting they die out completely like we've done for the last hundred years.
The only difference with Apple and a generic PC is that with Apple you can only get a motherboard from Apple. The CPUs are also soldered onto a little board but their are venders out there selling replacements. As far as memory, IDE devices, PCI, USB, 1394,.... (the list goes on) go they use standard parts. They even have some instructions on accelerateyoumac.com that describe how to get a standard, made for PC, GForce2 MX card running in your Mac.
Now it would be cool if other people were allowed to make Mac hardware but that would be like suicide - they have too much R&D to support (unlike that vast majority of PC manufacturers.)
Willy
Re:OSX might support Java better than Linux...
on
Another Look At OS X
·
· Score: 1
You're right about the threading stuff. This is from the Apple website:
Blazingly Fast Threading
In Mac OS X, Java threads are implemented directly on top of native Mach threads, allowing efficient scheduling using true preemptive multitasking. This also allows threaded applications to automatically use multiple CPUs, for blazingly fast performance.
I also read that their Java implementation makes sure that no one object gets loaded into memory more then once - even across different applications. I don't know if the current Windows/Linux/Sun implementations do this or not but it sounds like a good idea - speed isn't much of an issue for many things (today's CPUs are fast and typically just wait for user input anyway) but memory usage sure is.
What happens when you want to plug in a new model, or something NOT from Apple?
It brings up a dialog box - "New hardware found.... Would you like to search for drivers?" (or something very similar.) Click yes and it connects to the internet, finds the driver, downloads the driver, then installs the driver.
What? Can't find that new driver disk?
Mac don't use floppy drives anymore, remember? It'll go to a central spot on the web (ie, Apple) to download the driver. Third parties will just supply their drivers to Apple.
What? That spiffy new printer isn't even supported?
"Sorry, can't find the driver used to enable xxxxxxx. Please contact the xxxxxxx manufacturer."
Even MacOS classic doesn't insulate you from this. Neither did NeXT really.
Actually the "classic" MacOS does basically do this for USB devies. Support isn't that great but you can bet they've improved things considerably for MacOSX.
Apple isn't really trying to make their MacOS do what you have described. Instead, they trashed it all and bought up a company that did what they needed (that company being Next.) If you were to look at MacOSX now you'll find it has more in common with NextStep then it does MacOS9.
This is a good thing because if they were to try to modify OS9 to suit their needs I'm sure they would fail - just as you predicted. Fortunately for Apple, they bought an OS that has already proven its self to be anything but a failure (at least in OS design.)
There's something good to be said about starting fresh. It's a lot of work but it sure pays off in the long run...
You're right when you say that this just moves the polution. The difference is that those big power generation plants are _much_ more efficient then your typical car engine. The typical oil power plant is ~40% efficient - compared to ~25% for a car engine.
The other point is that it is much easier to impose enviromental regulations on a few power plants then a few million cars. Most of the people where I live just bore out their catalytic converters. And how many people do you think are running around with catalytic converters that barely work? Just putting some high quality/well maintained converters onto the power plants would provide much better protection for the enviroment.
And by the way, why has this taken so long to arrive? It's still not something I can purchase yet, and I first heard of it back in 1992. There's something fishy
Blame Microsoft - their continuing use of DOS (95, 98, ME) prevent these technologies from being introduced. Once your basic Microsoft OS is based on NT, these technologies will start to be devoloped.
It currently makes more sense for INTEL/AMD to invest in technologies that will make ME/9x faster as well as NT/Unix. Once SMT capable OSes are the norm they will look at other ways to speed things up.
How about mandating a device inside cars that transmit a standard bluetooth like signal to cell phones in the car when the car is placed into drive. Make it a very week signal so it only extends a couple of feet from the antenna that could be located by the driver. This would solve the problem without interfearing with everyone else (including the person in the passenger seat if it's designed right.) There would however have to be a consensus on a "v-chip" like device for cell phones.
Cell phone induced accidents are probably only going to get worse. If this were made cheap it might be a good idea.
They could also easily attach a speaker to such a device so people can tell where the "heat" is coming from. Just start blaring some "Britney Spears" when you turn it on and people will start running - for one reason if not the other...
Once one side of their body starts burning. I don't know about everyone else but if I were to feel a "wave of heat" hitting my face I would instinctively turn around. Then if my back started burning, I would start running.
I don't mean to support this weapon but the arguement that people run in to the weapon is bogus. If half of your body is burning you'll run in the other direction.
Please... Microsoft invested $150 million way back when Apple's stock price was almost rock bottom (before the anti-trust case.) Not only did Microsoft greatly benefit from the investment, but it got them out of some legal troubles. At the time Apple had several laysuits files against Microsoft. Microsoft invests, lawsuits go away. All of this has nothing to do with the anti-trust case.
Now with 1.5 GHz CPU's, there really isn't reasonably priced cache memory available that runs at that speed. So to keep costs low on consumer models, they put the L2 cache into the CPU die
While I agree that it's not reasonable to make external cache at 1.5GHz, it's not due to manufacturing issues. It's because of the RF produced and signal degrigation when connecting it to the CPU. The original Athlon proved that one can't get go that far beyond 200MHz. This can also be shown by how the P4 uses a 200MHz motherboard (400MHz effective.) Oh I'm sure there are other tricks to bring up the speed but moving the cache onto the die was the only reasonable solution.
I'm 99% sure that the L2 cache on the Sparc chips have been on-die or on-module for years now. That's one of the reasons that the chips are so darned expensive.
Interesting. I would assume they use large external caches. The CPUs didn't go that fast (MHz not effective speed) so having an internal cache shouldn't have been that important. But hey, I'll take your word on it.;)
They didn't do well in the consumer market, so they came out with the Pentium II's that had on-module (but 1/2, then 1/3, then 1/4 speed) L2 cache. When they needed to really get back and compete in the workstation/server market, they came up with the Xeon, which was a Pentium II or III with a full-speed on-module cache ranging from 512K to 2MB.
The way I remember it the thing the Xeon really had going for it over the PII/PIII was beter SMP support. This is why you would almost never see a single CPU Xeon system. Even todays' PIIIs have crappy SMP support compared to the Xeon, SPARC, G4, Power series... did I miss any? Probably. It also looks like AMD might include some kick ass SMP support into their new chips. Oh ya, add the Alpha to that list.
The 1MB L2 cache on that SPARC was, I assume, an external cache. The new TBirds and PIIIs have an internal cache that while smaller (256k), opperate at much higher speeds.
You're quite right about not enlarging the internal L2 cache. It serves no purpose to make it much bigger. Just look at what Apple has done with their latest G4e computers - 733MHz G4e with 256k internal cache plus 1MB external cache. The cool thing is you can access that 1MB cache directly if you have a program that requires fast access to a small chunk of memory. Just think of how fast that 1.6GHz Athalon would be with an extra 1MB cache to keep those pipelines full..
Making a standard 256k internal cache chip then letting different hardware vendors include an external cache makes for a flexable solution while still minimizing the number of different parts needed to manufacture/design.
Ok, lets face facts. This technology is not going to be used to destroy our privacy. If they bother you, remove them.
What this will do is greatly help reduce theft - and this helps everyone out there who isn't a theif. When you go shopping a significant portion of the price of goods is to make up for shoplifting. Those stupid scanners we currently have are basically useless - it's too often that they're wrong. Now if a store owner could track all items in a database there wouldn't be any mistakes. When the alarm goes off you _know_ that someone is shoplifting.
Another thing this could do is reduce the cost of doing business - thus reducing prices for us. Just think of how easy inventory would be. No more need to close shop for a few days, just click the "tally" button on the cash register.
This technology will reduce the cost of doing business. In the process, we'll all save money and I think that is a good thing..
The US is the best place in the world to live... if you're a rich, caucasun male..
Kind of goes with this old joke: (Person1) Hey man, I just met GOD!!! (Person2) Really? What did he look like?? (Person1) She was black..
I don't mean to flame the US, but even those Americans out there have to admit there is some truth to that first one. It appears there's no perfect balance when it comes to different cultures. For example, here in Canada we're not as wealthy. But then again, we have less crime. We pay more taxes but then we also have universal medical coverage.
Is one "culture" superior? I don't think so. They all have their pros and cons.
As for the previous message, while I agree it really sucks that medicine is often limited to those that can afford it, it is unfortunately the price we pay to have these medicines developed. Those researchers can't be expected to work for free. As far as the companies investing more money into ads then R&D, well this is just business. If more ads can make them more money to pay for the R&D then so be it. And one can't forget, for each product they bring to market they invest in many others that go nowhere. That one product must pay for all of the R&D.
I think the only way to promote private business to develop this stuff while still making it available to the "general" public is to only offer limited time patents. After 5-10 years is up the clone companies get their shot at producing it for the masses. Not perfect - but then no solution is. I believe this is the sort of protection the R&D groups in Quebec get.
pure inductance is lossless so long as nothing is "leaching" the power. If what my instructors have told me is correct, there is a greater loss of energy from this "leaching" then from resistance at very high voltages. I've designed these systems on paper and to do so requires one to look at tabulated data for each size of conductor. It's very complex because there are so many different things to consider. For example, distance from the ground, distance from the other phases,.... basically lots of enviromental factors come into play. The end result, there are more losses from the "leaching" then from the resistance.
This works great on paper but no so in the real world. In reality, you have a rather large magnetic field changing direction at 60Hz. Now some of the energy from this magnetic field is absorbed into the surrounding enviroment - hence, is lost. Much of this depends on the enviroment but it's safe to say that anything metal will have induced currents. The end result - on HV systems most of the power loss is not due to resistance, but due to the inductance (indirectly, but still due to the inductance.)
Adding capacitance adjusts for phase shift. I remember doing all of these calculations is school - but that was 5 years ago so I've forgotten exactly how it works. As far as the actual losses are concerned, losses from inductance are greater then those from resistance on HV lines - even if caps are added to the transmittion lines.
Increasing the voltage isn't that easy. Quebec hydro uses 1MegaVolt trasmittion lines and they get some weird stuff happening. I guess you can get the occasional spontanious nuclear reaction along those lines (I'm talking about a few atoms - not a nuke..) I agree it's a good idea just not that easy to implement.
Slashdot Mirror
USB 2.0 might be faster on paper but it's really pushing the hardware that the original USB was designed to operate on. I fully expect that it actually operates at a lower speed in real life conditions - much like 802.11. (11Mbit/s - ya right... more like 5.)
Also, for many operations (like copying from HD to HD) that 400MBit will give one much more bandwith then USB would give even if it does operate at spec. WIth USB the data would go from the HD to the PC then back to the other HD. Your 480Mbit/s just dropped to 240. Firewire cuts the PC out of the loop so you get the full 400Mbit/s speed (plus lots of saved CPU time.)
Need I even mention that you can use more then one connection to a device? Gang a couple of 1394 chanels together and there is no way that USB is faster. That and you can pick up a 6 port 1394 card for under $100. In a bit more time, when Linux support if fully tested, 1394 could easily replace external SCSI for even the most demanding tasks.
Willy
Except the obvious fact that this is an Intel technology that will probably require Intel silicon. On the other hand, Firewire is an open standard that many are supporting. For example:
VIA Technologies: http://www.cyrix.com/products/1394.htm
ALi: http://www.ali.com.tw/eng/product/peri/m5251.htm
Of course there are many others like Sony and TI but the fact of the matter is that there are more Firewire options then USB 2.0. Unless Intel lets these companies create their own USB 2.0 chipsets I doubt that you'll see USB 2.0 widely supported. There is simply no reason to support it when 1394 is a shipping, open standard that is already present on most digital camcorders. Add this to the fact that 1394 also has a great deal of room to grow (ie, 1600 Mbit/s) and it's not a hard decision to make.
Willy
Actually Firewire has much better plug-and-play support. With USB you need a host CPU / OS / driver combination in order to comunicate. Firewire does it all in the controller so it is far less driver dependent. Simpler, generic drivers work for most devices.
Want to see what Tom thinks? Check out the following web address. He also comes to the conclusion that Firewire has better pnp support.
http://www6.tomshardware.com/consumer/00q3/000929/ index.html
Willy
Words are great but their meanings can change over the years. Being able to hear words and see the situations in which they are used can give a much better overall representation of how society really was. (Granted you need to be selective with what you watch. ;)
Willy
Is this not less of an issue with newer CPUs? Just look at the rate at which the CPU/bus speed ratio is increasing. Much of the time a modern CPU spends is spent waiting for data from memory (granted this depends on the application.) If the CPU has to spend some extra time switching between rings then it shouldn't have much of an overall impact on speed - much of it's time is spent waiting anyway. This will only increase in the future.
The way I see it, the microkernel approach is much better in theory because it allows one to easily upgrade hardware architectures. Now in reality, we're stuck with 8x86 (mainly because of Microsoft.) In an environment where the vast majority of systems contain a single CPU of an architecture that doesn't change (much) then the monolithic kernels start to look real good. I would however prefer a world where we were not tied to legacy architectures and could jump from one to another with a simple microkernel rewrite (ya, I know, I'll keep dreaming. ;)
Linux is definitely well designed for todays systems but I would still argue that the microkernel philosophy in superior in theory. Too bad the real world isn't so easy to theorize.
Willy
This is not a problem associated with genetic engineering but with farming in general. Where there used to be hundreds if not thousands of different grains grown in North America there are now but a fraction of those available. Those that produce are reproduced and the others go extinct taking with them their genetic traits. We've imposed evolution based on profitability - not the species' ability to survive.
If anything GE well help preserve genetic diversity because at least we know the seeds will be available in some lab if required in the future. Not a good solution but better then letting they die out completely like we've done for the last hundred years.
Willy
Now it would be cool if other people were allowed to make Mac hardware but that would be like suicide - they have too much R&D to support (unlike that vast majority of PC manufacturers.)
Willy
Blazingly Fast Threading In Mac OS X, Java threads are implemented directly on top of native Mach threads, allowing efficient scheduling using true preemptive multitasking. This also allows threaded applications to automatically use multiple CPUs, for blazingly fast performance.
I also read that their Java implementation makes sure that no one object gets loaded into memory more then once - even across different applications. I don't know if the current Windows/Linux/Sun implementations do this or not but it sounds like a good idea - speed isn't much of an issue for many things (today's CPUs are fast and typically just wait for user input anyway) but memory usage sure is.
Willy
NextStep is the OS, not the hardware.
It brings up a dialog box - "New hardware found .... Would you like to search for drivers?" (or something very similar.) Click yes and it connects to the internet, finds the driver, downloads the driver, then installs the driver.
What? Can't find that new driver disk?
Mac don't use floppy drives anymore, remember? It'll go to a central spot on the web (ie, Apple) to download the driver. Third parties will just supply their drivers to Apple.
What? That spiffy new printer isn't even supported?
"Sorry, can't find the driver used to enable xxxxxxx. Please contact the xxxxxxx manufacturer."
Even MacOS classic doesn't insulate you from this. Neither did NeXT really.
Actually the "classic" MacOS does basically do this for USB devies. Support isn't that great but you can bet they've improved things considerably for MacOSX.
Willy
This is a good thing because if they were to try to modify OS9 to suit their needs I'm sure they would fail - just as you predicted. Fortunately for Apple, they bought an OS that has already proven its self to be anything but a failure (at least in OS design.)
There's something good to be said about starting fresh. It's a lot of work but it sure pays off in the long run...
Willy
The other point is that it is much easier to impose enviromental regulations on a few power plants then a few million cars. Most of the people where I live just bore out their catalytic converters. And how many people do you think are running around with catalytic converters that barely work? Just putting some high quality/well maintained converters onto the power plants would provide much better protection for the enviroment.
Willy
Blame Microsoft - their continuing use of DOS (95, 98, ME) prevent these technologies from being introduced. Once your basic Microsoft OS is based on NT, these technologies will start to be devoloped.
It currently makes more sense for INTEL/AMD to invest in technologies that will make ME/9x faster as well as NT/Unix. Once SMT capable OSes are the norm they will look at other ways to speed things up.
die ME die!!!
Cell phone induced accidents are probably only going to get worse. If this were made cheap it might be a good idea.
Willy
Willy
I don't mean to support this weapon but the arguement that people run in to the weapon is bogus. If half of your body is burning you'll run in the other direction.
Willy
Willy
While I agree that it's not reasonable to make external cache at 1.5GHz, it's not due to manufacturing issues. It's because of the RF produced and signal degrigation when connecting it to the CPU. The original Athlon proved that one can't get go that far beyond 200MHz. This can also be shown by how the P4 uses a 200MHz motherboard (400MHz effective.) Oh I'm sure there are other tricks to bring up the speed but moving the cache onto the die was the only reasonable solution.
I'm 99% sure that the L2 cache on the Sparc chips have been on-die or on-module for years now. That's one of the reasons that the chips are so darned expensive.
Interesting. I would assume they use large external caches. The CPUs didn't go that fast (MHz not effective speed) so having an internal cache shouldn't have been that important. But hey, I'll take your word on it. ;)
They didn't do well in the consumer market, so they came out with the Pentium II's that had on-module (but 1/2, then 1/3, then 1/4 speed) L2 cache. When they needed to really get back and compete in the workstation/server market, they came up with the Xeon, which was a Pentium II or III with a full-speed on-module cache ranging from 512K to 2MB.
The way I remember it the thing the Xeon really had going for it over the PII/PIII was beter SMP support. This is why you would almost never see a single CPU Xeon system. Even todays' PIIIs have crappy SMP support compared to the Xeon, SPARC, G4, Power series... did I miss any? Probably. It also looks like AMD might include some kick ass SMP support into their new chips. Oh ya, add the Alpha to that list.
Willy
You're quite right about not enlarging the internal L2 cache. It serves no purpose to make it much bigger. Just look at what Apple has done with their latest G4e computers - 733MHz G4e with 256k internal cache plus 1MB external cache. The cool thing is you can access that 1MB cache directly if you have a program that requires fast access to a small chunk of memory. Just think of how fast that 1.6GHz Athalon would be with an extra 1MB cache to keep those pipelines full..
Making a standard 256k internal cache chip then letting different hardware vendors include an external cache makes for a flexable solution while still minimizing the number of different parts needed to manufacture/design.
Willy
What this will do is greatly help reduce theft - and this helps everyone out there who isn't a theif. When you go shopping a significant portion of the price of goods is to make up for shoplifting. Those stupid scanners we currently have are basically useless - it's too often that they're wrong. Now if a store owner could track all items in a database there wouldn't be any mistakes. When the alarm goes off you _know_ that someone is shoplifting.
Another thing this could do is reduce the cost of doing business - thus reducing prices for us. Just think of how easy inventory would be. No more need to close shop for a few days, just click the "tally" button on the cash register.
This technology will reduce the cost of doing business. In the process, we'll all save money and I think that is a good thing..
Willy
The US is the best place in the world to live... if you're a rich, caucasun male..
Kind of goes with this old joke: (Person1) Hey man, I just met GOD!!! (Person2) Really? What did he look like?? (Person1) She was black..
I don't mean to flame the US, but even those Americans out there have to admit there is some truth to that first one. It appears there's no perfect balance when it comes to different cultures. For example, here in Canada we're not as wealthy. But then again, we have less crime. We pay more taxes but then we also have universal medical coverage.
Is one "culture" superior? I don't think so. They all have their pros and cons.
As for the previous message, while I agree it really sucks that medicine is often limited to those that can afford it, it is unfortunately the price we pay to have these medicines developed. Those researchers can't be expected to work for free. As far as the companies investing more money into ads then R&D, well this is just business. If more ads can make them more money to pay for the R&D then so be it. And one can't forget, for each product they bring to market they invest in many others that go nowhere. That one product must pay for all of the R&D.
I think the only way to promote private business to develop this stuff while still making it available to the "general" public is to only offer limited time patents. After 5-10 years is up the clone companies get their shot at producing it for the masses. Not perfect - but then no solution is. I believe this is the sort of protection the R&D groups in Quebec get.
Willy
Willy
Willy
Willy
Willy