I once made a microchip explode in lab when it came into contact with the contacts from a 560W DC power supply.
A corner of the chip 'chipped' off and left a mark in the desk when it he. Good thing I wasn't in the way.
Couple of points:
OS advocates say is fast release times, lower bugs, and quicker bug fixes then most CS projects, but, If a company wrote good software, fixed bugs quickly, and had a fast release time, why would we need OS?
If MS came out with a truly secure, well written, stable OS, wouldn't that kill most off the advantages of OS?
You have some good points, but you probably shouldn't use "OS" (Open Source) and OS (Operating System) in the same sentence. It's confusing.:-)
I thought one of Linux' charms was that it delayed obsolesence. I.e. find the lost power of that old 486/66.
Some people prefer not waiting 6 hours for the kernel to compile.:-)
FreeBSD's "make buildworld" command recompiles everything on the entire system. I wonder how adequate a 486 is then.
Yes, Linux and BSD are perfectly useable on ancient hardware...in console mode. Try KDE or Gnome,however, and feel the pain.
Voodoo 3? Yes, it's adequate if you don't mind running in 800x600 with (rather nastly looking) 16-bit color. Remember, when the Voodoo 3 was released, there were people saying "Why would you need a Voodoo 3? The most graphically intensive games are probably Quake2 and Unreal and a Voodoo1 up up works more than adequately for them." Note that gaming hardware advanced about two to three times faster than processors. Why buy a card like that when the games of tomorrow (and today) need more already?
Intel chips' FPUs (Floating-Point units for those not acronymnically inclined) have been inferior since the release of the first Athlon. Then the P4 core comes around and its SLOWER than that of the P3.
Of course, its clockspeed is higher so I guess that's all that matters to Joe Sixpack.
I update FreeBSD all the time directly from CVS. It gets from 10 to 75 updates/day (in the stable branch--I don't touch the current branch).
One command to update the source for everything on the system, one command to compile it, one command to install it.
Everything gets compiled using my favorite compiler settings which I put in/etc/make.conf
(-s -O2 -fschedule-insns2 -malign-functions=4 -fexpensive-optimizations -fomit-frame-pointer -pipe)
It is theoretically possible to get 100:1 or better compression. Assign a number to every file that has ever existed. It would surely take less than 64 bits to represent the number of files out there, and CERTAINLY less than 128.
Now, put every one of those files into the compressor (compressed if you like) and index them with numbers.
The compressed file would simply have a number or numbers of the files within. Even a full debian installation wouldn't exceed a few MB if even that.
The decompressor would take even more space than WindowsXP, and this would not work for newly created files, but it gets theoretical possibility out of the way. Now for practical possibility...
In practice, you would need a way to tell the decompressor what data to create from the compressed file. I notice that longest pattern in there is a string of nine "1's". How do you tell the compressor to create nine ones in less space than the nine ones would take uncompressed?
Furthermore, how would the decompressor know that the data telling it to make nine 1's wasn't compressed data itself? You would need to have some sort of agreement between the decompressor and compressor, which would limit your possibilities further or create a larger file. Remember, all data is ultimately binary regardless.
>Please put down the tin foil hat and step away from the wacko!
That was uncalled for. I did say "conceivably."
PC sales may or may not hurt Microsoft with regards to the X-box, but Microsoft is selling at below their final cost, as has been done with most console systems in the past. The consoles themselves are merely vessels for the profitable software; software being Microsoft's core business.
Sell the razor handles cheap, make your money on the blades themselves.
Microsoft is very likely indeed positioning themselves in home entertainment, pushing the fabled tale of convergence between it at the PC market. It is not possible to know what their exact "evil plans" are, but they are well thought out I am sure.
All I'm waiting for is an X-box Linux distro. Nothinbg would make me happier than running a Linux website on a Microsoft X-Box, with a little "Powered by Debian" and "Powered by X-Box" logo at the bottom, placed right next to each-other.:-)
HP, IBM, and depending on who you get, DirecTV are all good.
HP shipped a $400 scanner to replace my $200 scanner, overnight delivery, and paid for shipping both ways to me when my motherboard's USB controller was flaky. Not even their fault.
IBM offered to fly a technician to my house when OS/2 was refusing to use my soundcard. (it was a cheap SB clone at the time). Their suggestions got it working, so I was never able to test the offer.
DirecTV has at least reduced their wait time to just a few seconds, and now has around 5,000 customer service people in 6 centers. Quite a few of them are fairly clueless about DirecTC though, but at least they are willing to find an answer to a question they do not know.
Just wanted to point out some GOOD experiences, when there are so many bad ones to be heard about.
Consoles are sold at a loss, and the XBox is sold at more of a loss than any other console. Notice the business reviews of the XBox saying that Microsoft will lose several billion dollars and require over three years to turn a profit?
Console companies make their money on the software, and it will take quite a bit of software to make up the cost.
I'm sure that Microsoft knows what they are doing, though. They may not know how to design or impliment decent software. but nobody can argue that they know business.
The XBox, if you follow news on Microsoft, is to get their foot in the door of home electronics. The X-Box combined with Ultimate TV (Microsofts innovative clone of Tivo) can potentially allow Microsoft to eventually control advertising on your TV, track everything you watch, and conceivably eventually allow Microsoft to influence what is actually shown on television.
Compaq has a new business service called "Zero latency enterprise" which allows a company to look up your purchase history in less than one second. With an Xbox/UTV combo and a capability like this, Microsoft could display the absolute most effective ads for you and collect royalties from advertisers and networks.
I find it surprising that your system is stable with those Zalman coolers. They look neat, but that model is not efficient at cooling hot CPUs--look at any heatsink guide, such as Anandtech's.
The best heatsink for your motherboard would probably be the Thermalright SK-6. The S2460 doesn't have the 80mm mounting holes, so the very best coolers such as the overpriced Swiftech 80mm models and the Alpha PAL8035 are not compatible.
You can get SK'6s from 1coolpc.com for $30 and from Heatsinkfactory.com for a little more (I prefer 1coolpc, check their resellerratings.com rating)
You need to get fans separately. I use 80mm SunOn fans with an 80mm->60mm adapter, but the SK6 doesn't allow for enough airflow that my setup is any better than just using 60mm fans.
By the way, as a general rule of thumb right now, most coolers from Zalman and Thermalright are crap. The non-circular Thermalright heatsinks are good, but there are better.
1) Use external SCSI hard drives, put them in a closet or something. SCSI cables can be quite long, so this is actually pretty easy. You can even have very loud drives like the old Quantum Atlas 10K1 and not notice
2) The quietest hard drives are Seagate Barracuda ATA IV's. Check the noise comparisons on storagereview.com.
3) If you pump heat directly from heat sources out of the case, you can drastically reduce the number of fans. The only 2 items in a PC (other than the power supply) that typically need cooling are the CPU and video card. The vid card doesn't need noisy cooling, so you can deal with the CPU only. You can use a case like the Fong Kai FK-603 which funnels the output from the CPU fan directly out of the case, rather than lettign the hot air pass by the motherboard. You then need a grand total of 2 or three fans. 1 for the power supply, one for the CPU, and possible a small fan for the video card. Problem solved.
By the way, Dynamat doesn't do a thing for you unless your noise problems are caused by vibration rather than by spinning fans, hard disk platters, etc. It's also very expensive.
No.
Well, yes, the G4 does indeed run cooler, but you are incorrect as to why it runs cooler.
1) RISC/CISC has absolutely nothing to temperature. Alpha processors, for example, run hotter than most PC processors.
2) Temperature has nothing to do with instruction sets. It has alot to do with clock speed, but that is measurable only when comparing 2 chips of the exact same model.
3) RISC chips, which in a way don't technically exist, may execute most instructions in one clock cycle, but that is because the instructions are far simpler. It takes multiple instructions to do the work of an average single instruction on a CISC chip.
4) RISC = "Reduces Instruction Set Chip". The Motorola PowerPC 601 chip actually had/more/ instructions than the Pentium. RISC has become a marketing term and no longer really signifies any real-world difference between chips. That said, it is now rarely used as a marketing term because the RISC vs CISC debate has been dead since about mid-1999 or so. Neither method as a substantial advantage over the other. One executes more complete instructions per cycle but needs to execute more instructions to do the same thing.
What really effects processor speed:
*Transistors switching state. Every time a transistor switches its state, it takes power and generates heat. Better designed processors try to reduce unneccessary switching. This is one method AMD used to reduce heat between the Athlon T-bird and AthlonXP.
*Voltage. Even at the same clock speed, increasing voltage linearly increases the power dissipation of the chip.
*Transistor size..18 Micron transistors produce more heat than.25 micron transistors. Smaller transistors also allow lower voltage for stable operation.
*Number of transistors. Obviously, more transistors switching means more heat.
*Several other factors.
The G4 processor runs cooler because it is a simpler, more efficient design with less transistors. Additionally, Intel compatible chips have to have quite a bit of extra logic (=more transistors) to deal with the ancient, aging x86 architecture quickly. Actually, modern processors like the Athlon break down instructions into smaller operations so could technically be called "RISC core" processors. Hmm.
That doesn't sound very likely. The only significant architectural advantages that the P3 has over the T-bird are SSE and a much wider L2 cache (256 bit VS the Athlon's 64-bit)
In the majority of real-world applications, a T-bird will be within 5% of the speed of the P3. Sometimes 5% faster, sometimes 5% slower, or somewhere in-between (depends onthe application).
If you have an Athlon system (which has a clockspeed advantage, no less) that "seems" slower, it is likely either a hardware problem, as in some other component of the system is holding it back, or you suffer from the 'feel good' feeling that some people have with an Intel chip VS. anyone else.
Not that the P3 isn't a good chip--far from it. It is amazing that it can compete with the Athlon (other than in clockspeed) with a core that was designed almost 7 years ago. It is a tribute to Intel's engineering prowess.
If only the P4 was a tribute to anything but consumer ignorance.
"P4 is 2GHz. Ath-a-lon is 1.6GHz. 2 is bigger than 1.6, therefore P4 is faster."
By that logic, every CPU on earth has the same performance, so it doesn't matter other than how high you can get he clockspeed. I guess a 1GHz IBM Power4 is slower than a 1.3GHz Pentium-4. (or not...)
That said, if your Athlon system has a 100MHz FSB, something is wrong. The only chipset that defaults to such as speed is the prehistoric AMD750, which doesn't work with T-birds anyway. It is also not a DDR chipset. Come to think of it, the i815 (which is a great chipset) supports a 133MHz FSB as well. Was that a typo?
Are you really so misinformed that you believe a 1.4GHz Pentium 4 can outperform a 1.1GHz Athlon Thunderbird in more than a handful of applications? (all of which are special purpose and highly P4 optimized)
If you compare the design decisions made, all but the least geeky of us can see that the P4 is designed for clockspeed (for marketing--the average joe equates clockspeed linearly with performance) whereas the Athlon was designed for real-world performance.
If the only material you have is "what if the heatsink falls off.", that's fairly pathetic. Most Athlon motherboards have mounting holes around the CPU. If you mount the heatsink to those, you'd have to rip a chunk of the motherboard off in order for the heatsink to come loose--and even then it would still be attached to the CPU! (Just that the CPU portion of the motherboard would have been liberated)
Never tried it with no heatsink, but when building a system with an AthlonXP I installed Windows 2000 without the fan ever operating. The heatsink got HOT, but the system never crashed nor did the chip die.
Slashdot Mirror
Cell phones = communicators
Babelfish = universal translator
Taser(tm) = Phaser on stun
2-way videophone = Screen on the bridge
PC = Enterprise computer terminal
Now antimatter propulsion.
Was this guy good or what?
Problem solved.
I once made a microchip explode in lab when it came into contact with the contacts from a 560W DC power supply.
A corner of the chip 'chipped' off and left a mark in the desk when it he. Good thing I wasn't in the way.
Couple of points: OS advocates say is fast release times, lower bugs, and quicker bug fixes then most CS projects, but, If a company wrote good software, fixed bugs quickly, and had a fast release time, why would we need OS? If MS came out with a truly secure, well written, stable OS, wouldn't that kill most off the advantages of OS?
:-)
You have some good points, but you probably shouldn't use "OS" (Open Source) and OS (Operating System) in the same sentence. It's confusing.
I thought one of Linux' charms was that it delayed obsolesence. I.e. find the lost power of that old 486/66. Some people prefer not waiting 6 hours for the kernel to compile. :-)
FreeBSD's "make buildworld" command recompiles everything on the entire system. I wonder how adequate a 486 is then.
Yes, Linux and BSD are perfectly useable on ancient hardware...in console mode. Try KDE or Gnome,however, and feel the pain.
Voodoo 3? Yes, it's adequate if you don't mind running in 800x600 with (rather nastly looking) 16-bit color. Remember, when the Voodoo 3 was released, there were people saying "Why would you need a Voodoo 3? The most graphically intensive games are probably Quake2 and Unreal and a Voodoo1 up up works more than adequately for them." Note that gaming hardware advanced about two to three times faster than processors. Why buy a card like that when the games of tomorrow (and today) need more already?
Intel chips' FPUs (Floating-Point units for those not acronymnically inclined) have been inferior since the release of the first Athlon. Then the P4 core comes around and its SLOWER than that of the P3.
Of course, its clockspeed is higher so I guess that's all that matters to Joe Sixpack.
Remember, they are targeting businessmen that probably already use Windows on their servers. The target market is already gullible.
I update FreeBSD all the time directly from CVS. It gets from 10 to 75 updates/day (in the stable branch--I don't touch the current branch). /etc/make.conf
One command to update the source for everything on the system, one command to compile it, one command to install it.
Everything gets compiled using my favorite compiler settings which I put in
(-s -O2 -fschedule-insns2 -malign-functions=4 -fexpensive-optimizations -fomit-frame-pointer -pipe)
Piece of cake.
This article isn't about FreeBSD advocacy per se. Besides, FreeBSD is my secret weapon. Not very secret if a I flaunting it everywhere... Oops.
Do you believe everything you read on the internet? Probably not, or you'd be posting from a Windows XP system. (You aren't, are you?)
You can cut the cost in half by subscribing to one or more operating systems. They also ship subscriptions sooner than ragular orders.
It is theoretically possible to get 100:1 or better compression. Assign a number to every file that has ever existed. It would surely take less than 64 bits to represent the number of files out there, and CERTAINLY less than 128.
Now, put every one of those files into the compressor (compressed if you like) and index them with numbers.
The compressed file would simply have a number or numbers of the files within. Even a full debian installation wouldn't exceed a few MB if even that.
The decompressor would take even more space than WindowsXP, and this would not work for newly created files, but it gets theoretical possibility out of the way. Now for practical possibility...
In practice, you would need a way to tell the decompressor what data to create from the compressed file. I notice that longest pattern in there is a string of nine "1's". How do you tell the compressor to create nine ones in less space than the nine ones would take uncompressed?
Furthermore, how would the decompressor know that the data telling it to make nine 1's wasn't compressed data itself? You would need to have some sort of agreement between the decompressor and compressor, which would limit your possibilities further or create a larger file. Remember, all data is ultimately binary regardless.
>Please put down the tin foil hat and step away from the wacko!
:-)
That was uncalled for. I did say "conceivably."
PC sales may or may not hurt Microsoft with regards to the X-box, but Microsoft is selling at below their final cost, as has been done with most console systems in the past. The consoles themselves are merely vessels for the profitable software; software being Microsoft's core business.
Sell the razor handles cheap, make your money on the blades themselves.
Microsoft is very likely indeed positioning themselves in home entertainment, pushing the fabled tale of convergence between it at the PC market. It is not possible to know what their exact "evil plans" are, but they are well thought out I am sure.
All I'm waiting for is an X-box Linux distro. Nothinbg would make me happier than running a Linux website on a Microsoft X-Box, with a little "Powered by Debian" and "Powered by X-Box" logo at the bottom, placed right next to each-other.
HP, IBM, and depending on who you get, DirecTV are all good.
HP shipped a $400 scanner to replace my $200 scanner, overnight delivery, and paid for shipping both ways to me when my motherboard's USB controller was flaky. Not even their fault.
IBM offered to fly a technician to my house when OS/2 was refusing to use my soundcard. (it was a cheap SB clone at the time). Their suggestions got it working, so I was never able to test the offer.
DirecTV has at least reduced their wait time to just a few seconds, and now has around 5,000 customer service people in 6 centers. Quite a few of them are fairly clueless about DirecTC though, but at least they are willing to find an answer to a question they do not know.
Just wanted to point out some GOOD experiences, when there are so many bad ones to be heard about.
Consoles are sold at a loss, and the XBox is sold at more of a loss than any other console. Notice the business reviews of the XBox saying that Microsoft will lose several billion dollars and require over three years to turn a profit?
Console companies make their money on the software, and it will take quite a bit of software to make up the cost.
I'm sure that Microsoft knows what they are doing, though. They may not know how to design or impliment decent software. but nobody can argue that they know business.
The XBox, if you follow news on Microsoft, is to get their foot in the door of home electronics. The X-Box combined with Ultimate TV (Microsofts innovative clone of Tivo) can potentially allow Microsoft to eventually control advertising on your TV, track everything you watch, and conceivably eventually allow Microsoft to influence what is actually shown on television.
Compaq has a new business service called "Zero latency enterprise" which allows a company to look up your purchase history in less than one second. With an Xbox/UTV combo and a capability like this, Microsoft could display the absolute most effective ads for you and collect royalties from advertisers and networks.
Okay, enough conspiracy theories.
When Microsoft rules the world, I'm moving.
A microsoft product defective?
Poor technical support?
Hell has now frozen over.
I find it surprising that your system is stable with those Zalman coolers. They look neat, but that model is not efficient at cooling hot CPUs--look at any heatsink guide, such as Anandtech's.
The best heatsink for your motherboard would probably be the Thermalright SK-6. The S2460 doesn't have the 80mm mounting holes, so the very best coolers such as the overpriced Swiftech 80mm models and the Alpha PAL8035 are not compatible.
You can get SK'6s from 1coolpc.com for $30 and from Heatsinkfactory.com for a little more (I prefer 1coolpc, check their resellerratings.com rating)
You need to get fans separately. I use 80mm SunOn fans with an 80mm->60mm adapter, but the SK6 doesn't allow for enough airflow that my setup is any better than just using 60mm fans.
By the way, as a general rule of thumb right now, most coolers from Zalman and Thermalright are crap. The non-circular Thermalright heatsinks are good, but there are better.
1) Use external SCSI hard drives, put them in a closet or something. SCSI cables can be quite long, so this is actually pretty easy. You can even have very loud drives like the old Quantum Atlas 10K1 and not notice
2) The quietest hard drives are Seagate Barracuda ATA IV's. Check the noise comparisons on storagereview.com.
3) If you pump heat directly from heat sources out of the case, you can drastically reduce the number of fans. The only 2 items in a PC (other than the power supply) that typically need cooling are the CPU and video card. The vid card doesn't need noisy cooling, so you can deal with the CPU only. You can use a case like the Fong Kai FK-603 which funnels the output from the CPU fan directly out of the case, rather than lettign the hot air pass by the motherboard. You then need a grand total of 2 or three fans. 1 for the power supply, one for the CPU, and possible a small fan for the video card. Problem solved.
By the way, Dynamat doesn't do a thing for you unless your noise problems are caused by vibration rather than by spinning fans, hard disk platters, etc. It's also very expensive.
Hope this helps.
No.
/more/ instructions than the Pentium. RISC has become a marketing term and no longer really signifies any real-world difference between chips. That said, it is now rarely used as a marketing term because the RISC vs CISC debate has been dead since about mid-1999 or so. Neither method as a substantial advantage over the other. One executes more complete instructions per cycle but needs to execute more instructions to do the same thing.
.18 Micron transistors produce more heat than .25 micron transistors. Smaller transistors also allow lower voltage for stable operation.
Well, yes, the G4 does indeed run cooler, but you are incorrect as to why it runs cooler.
1) RISC/CISC has absolutely nothing to temperature. Alpha processors, for example, run hotter than most PC processors.
2) Temperature has nothing to do with instruction sets. It has alot to do with clock speed, but that is measurable only when comparing 2 chips of the exact same model.
3) RISC chips, which in a way don't technically exist, may execute most instructions in one clock cycle, but that is because the instructions are far simpler. It takes multiple instructions to do the work of an average single instruction on a CISC chip.
4) RISC = "Reduces Instruction Set Chip". The Motorola PowerPC 601 chip actually had
What really effects processor speed:
*Transistors switching state. Every time a transistor switches its state, it takes power and generates heat. Better designed processors try to reduce unneccessary switching. This is one method AMD used to reduce heat between the Athlon T-bird and AthlonXP.
*Voltage. Even at the same clock speed, increasing voltage linearly increases the power dissipation of the chip.
*Transistor size.
*Number of transistors. Obviously, more transistors switching means more heat.
*Several other factors.
The G4 processor runs cooler because it is a simpler, more efficient design with less transistors. Additionally, Intel compatible chips have to have quite a bit of extra logic (=more transistors) to deal with the ancient, aging x86 architecture quickly. Actually, modern processors like the Athlon break down instructions into smaller operations so could technically be called "RISC core" processors. Hmm.
That doesn't sound very likely. The only significant architectural advantages that the P3 has over the T-bird are SSE and a much wider L2 cache (256 bit VS the Athlon's 64-bit)
In the majority of real-world applications, a T-bird will be within 5% of the speed of the P3. Sometimes 5% faster, sometimes 5% slower, or somewhere in-between (depends onthe application).
If you have an Athlon system (which has a clockspeed advantage, no less) that "seems" slower, it is likely either a hardware problem, as in some other component of the system is holding it back, or you suffer from the 'feel good' feeling that some people have with an Intel chip VS. anyone else.
Not that the P3 isn't a good chip--far from it. It is amazing that it can compete with the Athlon (other than in clockspeed) with a core that was designed almost 7 years ago. It is a tribute to Intel's engineering prowess.
If only the P4 was a tribute to anything but consumer ignorance.
"P4 is 2GHz. Ath-a-lon is 1.6GHz. 2 is bigger than 1.6, therefore P4 is faster."
By that logic, every CPU on earth has the same performance, so it doesn't matter other than how high you can get he clockspeed. I guess a 1GHz IBM Power4 is slower than a 1.3GHz Pentium-4. (or not...)
That said, if your Athlon system has a 100MHz FSB, something is wrong. The only chipset that defaults to such as speed is the prehistoric AMD750, which doesn't work with T-birds anyway. It is also not a DDR chipset. Come to think of it, the i815 (which is a great chipset) supports a 133MHz FSB as well. Was that a typo?
Charles Burns
Are you really so misinformed that you believe a 1.4GHz Pentium 4 can outperform a 1.1GHz Athlon Thunderbird in more than a handful of applications? (all of which are special purpose and highly P4 optimized)
If you compare the design decisions made, all but the least geeky of us can see that the P4 is designed for clockspeed (for marketing--the average joe equates clockspeed linearly with performance) whereas the Athlon was designed for real-world performance.
If the only material you have is "what if the heatsink falls off.", that's fairly pathetic. Most Athlon motherboards have mounting holes around the CPU. If you mount the heatsink to those, you'd have to rip a chunk of the motherboard off in order for the heatsink to come loose--and even then it would still be attached to the CPU! (Just that the CPU portion of the motherboard would have been liberated)
Charles Burns
http://www.octools.com/index.cgi?caller=articles/s ubmersion/submersion.html
Never tried it with no heatsink, but when building a system with an AthlonXP I installed Windows 2000 without the fan ever operating. The heatsink got HOT, but the system never crashed nor did the chip die.