I just need to correct your assertion that FreeBSD was the mother of the BSD families.
It's not.;)
In the beginning was Berkeley Unix. Around 1990, Berkeley released Net/2, based on 4.3BSD. This became the base for the Jolitz' 386BSD. 386BSD was the original open-source BSD. Unfortunately, a lot of 386BSD users got frustrated by the slow release cycle, so two projects started independently at about the same time (1993?): FreeBSD and NetBSD.
FreeBSD and NetBSD had fairly different goals. FreeBSD wanted to continue what the Jolitz' started vis creating a Unix optimized for the inexpensive 386. NetBSD wanted to create a Unix that could run on anything, and also continue what the Jolitz' started vis creating a clean Unix that dumped a lot of the silly cruft that was introduced during midnight hacks at Berkeley.
Around 1994, Berkeley released 4.4BSD-Lite, a completely unencumbered version of their Unix (ie. no code owned by AT&T). This code was integrated into FreeBSD 2.0 and NetBSD 1.0.
So you see, 386BSD and 4.4BSD-Lite together form the base upon which the current FreeBSD and NetBSD projects were built.
Re:NetBSD stopped being useful once I forked OpenB
on
NetBSD 1.5ZB
·
· Score: 1
Man, can't you even spell your own name right?! It's Theo de Raadt, not Theo DeRaadt.
Good job though. Plenty of folks can now consider themselves trolled.
If I may add something... there's a long thread on exactly this subject in comp.arch right now. Look for the thread called "SPEC2k results for G4". There's some very interesting comments from people that mostly seem know what they're talking about.
Sure, but that's not really relevant to the article. SPEC is a measure of the capability of a CPU+Memory Subsystem+compiler in performing Real Work, where Real Work is number/symbol crunching. You know - linear algebra, compiling, those kinds of things. SPEC couldn't care less about measuring how responsive a GUI is, and rightly so. Assuming the OS has decent process+priority management (as it should for Mach+BSD), the GUI bloat should not be relevant.
I was actually quite surprised by the poor performance of the G4. Although if the other posters are correct in stating that SPEC doesn't test very well for single-precision floats nor SIMD, it's probably not a very good test of the capabilities of the CPU. Of course, Apple has crippled their systems with PC133. That doesn't help either.
I believe the PS/2 uses a R5x00 cpu, no? If that is related in any way other than name to R5k cpus used in olden time SGIs, then it's highly optimized for single-precision multiply-add instructions. So if your renderer makes heavy use of MADD, it would probably outperform said PII and Celeron.
As mentioned before, since you can't use Maya (no port to Linux-Mips), you would have to find an open source renderer. You would then have the opportunity to hack it to optimize it for the R5x00. Wheeeeee....
Yes, extinction has happened without the involvement of humans and it will continue to happen without the involvement of humans. (that sentence has a double-entendre, btw.;) That is not in dispute.
What is in dispute is the value humans give to diversity. -There is economic value to diversity in the form useful genes and groups of genes. Also, zoos and nature documentaries are fairly lucrative.;) -There is ethical value to diversity. While you may not care much about the wonderful variety of organisms, there are a great many people who do. And that matters.
As for the value of searching out new species --- you can never predict the value of a scientific endeavour. In this case however, you can make an educated guess that discovering new species will provide new insight into evolutionary, ecological, anatomical, physiological, genetic, biochemical, and behavioural processes. That in itself is quite a return on the investment! Imagine what the world would be like if no one had gone out looking for archaebacteria. We wouldn't know about taq polymerase, an enzyme isolated from the archaebacterium Thermophilus aquaticus -- the world wouldn't have the polymerase chain reaction as we know it, and that means genetic research would be hampered to some degree. You just NEVER know what you're gonna find if you go looking. You're bound to be surprised.
Someone already mentioned the dodo... I would add to that by mentioning the Calvaria major tree. Without the dodo to ingest its seeds and prepare them for germination, the tree is doomed in its natural habitat. No C. major trees have sprouted since the dodo went the way of the dodo. I believe there currently is a group that's trying to preserve the species by using turkeys instead to digest the seed coat. Now consider all those other species suffering a similar fate because their ecologies aren't well understood.
I was just about to ask this myself. There doesn't seem to be a suitable topic for interviews pertaining to Plan9, Inferno, nor Vita Nuova.
Perhaps they should consider creating such a topic if only to showcase the Plan9 mascot, Glenda. By far the most [endearing|1337|funny] mascot among the open source OSes.
There must be some secret society that insists on mispelling "Athlon" so they can recognize each other. C'mon, the name's been official for a long time now --- it's not Athelon or Athalon. It's Athlon.
Btw, let me take this opportunity to thank you for the SGI page. Cheers!
Please reverify the current licensing. It has been changed since RMS wrote that. Vita Nuova paraphrases the current license as such:
The full text of the Plan 9 Open Source Licence can be found at the Bell Labs Plan 9 site. The licence is similar to many Open Source licences.
The main points are:
-You can modify, copy and distribute the source code as you wish.
-There are no royalty payments on the distribution.
I believe there is a clause, or combinations of clauses, that require you to provide Lucent with source if you distribute binaries only. Sort of looks like a semi-BSDified GPL. Please correct me if I'm wrong -- I'm not completely fluent in legalese.
Well, what's "too slow for web browsing" for one is "perfectly acceptable" for another.;) But I admit I wouldn't bother trying to get Flash and the Java plug-in going on 6.5.
I don't accept the notion that they're "not much use"... Granted, they're not much for general use, considering what "general use" entails nowadays for most people, and they're not much for professional use, considering what "professional use" means to an SGI nowadays (although they did just fine when released). So that leaves specialized use..... admittedly, it would have to be pretty specialized. Maybe it's just me.... but I don't see any good reason, other than code bloat, for a machine that animated dinosaurs (among other things) to be obsolete. It seems like such a waste. (Sure, its lack of hardware texturing would make it a poor choice for animating now, but the system as a whole is rock solid).
As for SGI releasing their dev tools or even IRIX 5.3, 6.2, 6.5.x to hobbyists... well, that's occasionally brought up on comp.sys.sgi.x. From winter 2000/2001 straight through summer 2001, there was endless ranting about SGI, including on this particular topic. I personally wish they would make it easier for hobbyists or even professionals who might depend on the second-hand market (which SGI has recently gotten into with reasonable prices for hardware). They stand to lose nothing by releasing older versions of IRIX for free, nor by making MipsPro available for these older systems. The whole licensing scheme is bogus. GCC works on Mips, but the quality of code produced by MipsPro is unbelievable. In fact, there's a thread on comp.os.plan9 now in which Rob Pike states he wouldn't be surprised if it was officially revealed that there's over 25,000 lines of code in MipsPro just for instruction reordering.
Look, if you really feel that bad about only being able to run 6.2 on it, I'll gladly take it off your hands. Not that's it's legal to run it at home though. And it would probably interfere with the equipment at work.
It's funny that older operating systems don't have as much appeal as the hardware they ran on. Irix 6.2 is solid and provides the hardware support and interface (and it requires less memory than 6.5... a bonus when you consider the cost of adding memory to most of these things). And GCC provides everything GNU. You're not really missing anything. Unix is unix is unix... if it supports posix and BSD extensions, you're all set -- SGI's website even has a slew of tips for compiling with gcc (I bet you already knew this since you knew about the caveats of using gcc). At least that's how I see it.;) I respect older unices for their efficiency -- of which 6.2 pales in comparison with 5.3.
Of course, if you actually DID want to run commercial 6.5 software on it.... you would probably also have the money to buy a new workstation or a second-hand Onyx.
Re:The Indigo was a nice machine...
on
Iris Indigo Case Mod
·
· Score: 2, Informative
It may be slow if you plan on using it as a workstation running the latest CAD apps and compilers, but it is perfectly usable and quite snappy as a desktop system. Depending on the graphics/video board(s) in it, you may be able to capture video, or do other fun things too.
It's not true that they've "mostly" got R3k cpus. R3k models do take weird memory though. And all Indigos have proprietary keyboard and mouse ports. While R4k Indigos with enough memory (at least 64MB) can run Irix 6.5 just fine, R3k models can only run up to Irix 5.3. Despite what you're implying, 5.3 is also a "good" version of Irix, if older than 6.5. But come on, with the free development files from SGI, you can compile anything you want with gcc. It IS unix after all.
I actually hate to see people doing things like this to such a remarkable system. I would rather they sell them to resellers so that afficionados can get their hands on them. They DID play a role in the making of Jurassic Park after all.;)
As to Indigo2s being better than Indys.... well, it depends. The indy does have a built-in port for an IndyCam and it is a LOT quieter and smaller and sucks up less power. The Indigo2 is newer though, and can support r10k cpus and hardware texturing.
No, you're not. I found it odd to see several jokes posted so quickly and modded up so quickly.
I heard about a russian man who had his face bitten off by a bear some years ago. He had one eye left, so he could function. The pictures were pretty gruesome - a big gaping hole in his face. The russian doctors couldn't do much, so he lived wearing a scarf over his face for a long time. Eventually, an american institution offered to create a face for him. I don't remember which institution it was. Might be the same one.
But this guy that lost not just his face, both his eyes also... I shudder to think about it. It's remarkable that he had the will to go on like he did... thanks in part to a dream about his granddaughter.
From SGI's website,
Fuel: 977 BTU/hour
Octane and Octane2: 2400 BTU/hour
Fuel thus runs considerably cooler than the Octanes. But it runs 77 BTU/hour hotter than an O2.
semantic confusion... my meaning was not that A/UX was mach-based but that A/UX was briefly an all original OS, according to the map, before incorporating BSD (again, according to the map)...
That being said, thanks for the info! I had never even heard of UniSoft or UniPlus. Heh, as if the history wasn't complicated enough...
Fascinating map, I could spend hours just researching from it to find out what the actual relationships are (i.e. what portions of OS A were integrated into OS B)...
One thing: IRIX is shown beginning around 1986 right out of the blue. According to SGI.com's section on IRIX, it was incepted in '82. I'm curious which it is, and whether it was derived or really conjured up from scratch.
The only others that appear out of thin air (other than UNICS) are Minix, Xinu, and Mach (A/UX briefly so before getting BSDed). Does anyone know if IRIX, like these, was an original design? It's certainly unique in its own right.
Also, did I miss UNICOS in there? As I understand it, it is a UNIX (looks like a duck, quacks like a duck) derived from SysV with some BSDisms thrown in... showed up in '85, I reckon.
Check out the internal bandwidth of those boxes, 5 year old O2's and Indigo2's still have greater memory bandwidth than the latest pentium motherboards
latest x86 = ~800MB/s memory bandwidth (or so the rumours go) Indigo2 = 400MB/s (half x86)... see www.futuretech.vuurwerk.nl/indigo2 O2 = 2.1GB/s (lots more than either...) see www.sgi.com/o2
Advanced Micro Devices Microstar K7T Pro motherboard 1.0GHz #cpus 1 Spec2000 fp Base 290 Spec2000 fp Peak 302 (just showed up recently on spec.org... still doesn't do any good without G4 results, that's true... why doesn't Motorola publish spec info on the spec website anymore? even if it is just for evaluation boards, it'd be fun to see...)
anyway, spec95 results are available for Athlon 1000 and G4 450... Athlon base FP = 29.4, G4 FP = 20.4 (not sure if it's base or peak... got the info from the CPU Info Center)
as usual, take the results for what they're worth...
tell me about it... got a cbc recording from way back but I no longer have the betamax to watch it... gonna have to consider getting the 'soon-to-be-released' dvd... and a dvd-player/drive to watch it...
The reason Intel's Pentium division and AMD's Athlon division don't make chips as cool as Crusoe and StrongARM (another Intel division now) is because there's always a market for the fastest cpus at a given time. How do they get so fast? High clockspeed and a risc-like core. This also results in high energy consumption. It's no different than Alpha, R10k, PA-RISC and USparc. This is still the easiest way to make a fast cpu. An interesting bit of cpu trivia is that one of the main designers of the Alpha is one of lead designers of the Athlon.
Take a look at the great big CPU comparison chart at bwrc.eecs.berkeley.edu/CIC and compare Spec results with power consumption, making sure to control for year of introduction. And yes, Spec scores are valid in this case because we're discussing solely the "speed" of the cpu.
The thinking has not stopped. AIM has produced some impressive PPC chips that require little power and perform quite well... just not as fast as the rest... same goes for the StrongARM -- not super fast, but its power consumption is practically nil when compared with the rest of a desktop computer (embedded systems are another matter -- not nil in this case, but pretty low still).
Slashdot Mirror
I should have added a link to the BSD family tree diagram.
I just need to correct your assertion that FreeBSD was the mother of the BSD families.
;)
It's not.
In the beginning was Berkeley Unix. Around 1990, Berkeley released Net/2, based on 4.3BSD. This became the base for the Jolitz' 386BSD. 386BSD was the original open-source BSD. Unfortunately, a lot of 386BSD users got frustrated by the slow release cycle, so two projects started independently at about the same time (1993?): FreeBSD and NetBSD.
FreeBSD and NetBSD had fairly different goals. FreeBSD wanted to continue what the Jolitz' started vis creating a Unix optimized for the inexpensive 386. NetBSD wanted to create a Unix that could run on anything, and also continue what the Jolitz' started vis creating a clean Unix that dumped a lot of the silly cruft that was introduced during midnight hacks at Berkeley.
Around 1994, Berkeley released 4.4BSD-Lite, a completely unencumbered version of their Unix (ie. no code owned by AT&T). This code was integrated into FreeBSD 2.0 and NetBSD 1.0.
So you see, 386BSD and 4.4BSD-Lite together form the base upon which the current FreeBSD and NetBSD projects were built.
Man, can't you even spell your own name right?! It's Theo de Raadt, not Theo DeRaadt.
Good job though. Plenty of folks can now consider themselves trolled.
If I may add something... there's a long thread on exactly this subject in comp.arch right now. Look for the thread called "SPEC2k results for G4". There's some very interesting comments from people that mostly seem know what they're talking about.
Sure, but that's not really relevant to the article. SPEC is a measure of the capability of a CPU+Memory Subsystem+compiler in performing Real Work, where Real Work is number/symbol crunching. You know - linear algebra, compiling, those kinds of things. SPEC couldn't care less about measuring how responsive a GUI is, and rightly so. Assuming the OS has decent process+priority management (as it should for Mach+BSD), the GUI bloat should not be relevant.
I was actually quite surprised by the poor performance of the G4. Although if the other posters are correct in stating that SPEC doesn't test very well for single-precision floats nor SIMD, it's probably not a very good test of the capabilities of the CPU. Of course, Apple has crippled their systems with PC133. That doesn't help either.
I believe the PS/2 uses a R5x00 cpu, no? If that is related in any way other than name to R5k cpus used in olden time SGIs, then it's highly optimized for single-precision multiply-add instructions. So if your renderer makes heavy use of MADD, it would probably outperform said PII and Celeron.
As mentioned before, since you can't use Maya (no port to Linux-Mips), you would have to find an open source renderer. You would then have the opportunity to hack it to optimize it for the R5x00. Wheeeeee....
Yes, extinction has happened without the involvement of humans and it will continue to happen without the involvement of humans. (that sentence has a double-entendre, btw. ;) That is not in dispute.
;)
What is in dispute is the value humans give to diversity.
-There is economic value to diversity in the form useful genes and groups of genes. Also, zoos and nature documentaries are fairly lucrative.
-There is ethical value to diversity. While you may not care much about the wonderful variety of organisms, there are a great many people who do. And that matters.
As for the value of searching out new species --- you can never predict the value of a scientific endeavour. In this case however, you can make an educated guess that discovering new species will provide new insight into evolutionary, ecological, anatomical, physiological, genetic, biochemical, and behavioural processes. That in itself is quite a return on the investment! Imagine what the world would be like if no one had gone out looking for archaebacteria. We wouldn't know about taq polymerase, an enzyme isolated from the archaebacterium Thermophilus aquaticus -- the world wouldn't have the polymerase chain reaction as we know it, and that means genetic research would be hampered to some degree. You just NEVER know what you're gonna find if you go looking. You're bound to be surprised.
Someone already mentioned the dodo... I would add to that by mentioning the Calvaria major tree. Without the dodo to ingest its seeds and prepare them for germination, the tree is doomed in its natural habitat. No C. major trees have sprouted since the dodo went the way of the dodo. I believe there currently is a group that's trying to preserve the species by using turkeys instead to digest the seed coat. Now consider all those other species suffering a similar fate because their ecologies aren't well understood.
I was just about to ask this myself. There doesn't seem to be a suitable topic for interviews pertaining to Plan9, Inferno, nor Vita Nuova.
Perhaps they should consider creating such a topic if only to showcase the Plan9 mascot, Glenda. By far the most [endearing|1337|funny] mascot among the open source OSes.
There must be some secret society that insists on mispelling "Athlon" so they can recognize each other. C'mon, the name's been official for a long time now --- it's not Athelon or Athalon. It's Athlon.
Btw, let me take this opportunity to thank you for the SGI page. Cheers!
I think he's laughing at the way you said what you said, not at your preferred interface.
Please reverify the current licensing. It has been changed since RMS wrote that. Vita Nuova paraphrases the current license as such:
The full text of the Plan 9 Open Source Licence can be found at the Bell Labs Plan 9 site. The licence is similar to many Open Source licences.
The main points are:
-You can modify, copy and distribute the source code as you wish.
-There are no royalty payments on the distribution.
I believe there is a clause, or combinations of clauses, that require you to provide Lucent with source if you distribute binaries only. Sort of looks like a semi-BSDified GPL. Please correct me if I'm wrong -- I'm not completely fluent in legalese.
Well, what's "too slow for web browsing" for one is "perfectly acceptable" for another. ;) But I admit I wouldn't bother trying to get Flash and the Java plug-in going on 6.5.
I don't accept the notion that they're "not much use"... Granted, they're not much for general use, considering what "general use" entails nowadays for most people, and they're not much for professional use, considering what "professional use" means to an SGI nowadays (although they did just fine when released). So that leaves specialized use..... admittedly, it would have to be pretty specialized. Maybe it's just me.... but I don't see any good reason, other than code bloat, for a machine that animated dinosaurs (among other things) to be obsolete. It seems like such a waste. (Sure, its lack of hardware texturing would make it a poor choice for animating now, but the system as a whole is rock solid).
As for SGI releasing their dev tools or even IRIX 5.3, 6.2, 6.5.x to hobbyists... well, that's occasionally brought up on comp.sys.sgi.x. From winter 2000/2001 straight through summer 2001, there was endless ranting about SGI, including on this particular topic. I personally wish they would make it easier for hobbyists or even professionals who might depend on the second-hand market (which SGI has recently gotten into with reasonable prices for hardware). They stand to lose nothing by releasing older versions of IRIX for free, nor by making MipsPro available for these older systems. The whole licensing scheme is bogus. GCC works on Mips, but the quality of code produced by MipsPro is unbelievable. In fact, there's a thread on comp.os.plan9 now in which Rob Pike states he wouldn't be surprised if it was officially revealed that there's over 25,000 lines of code in MipsPro just for instruction reordering.
"Dust off the ol crimson"??
;) I respect older unices for their efficiency -- of which 6.2 pales in comparison with 5.3.
*faint*
Look, if you really feel that bad about only being able to run 6.2 on it, I'll gladly take it off your hands. Not that's it's legal to run it at home though. And it would probably interfere with the equipment at work.
It's funny that older operating systems don't have as much appeal as the hardware they ran on. Irix 6.2 is solid and provides the hardware support and interface (and it requires less memory than 6.5... a bonus when you consider the cost of adding memory to most of these things). And GCC provides everything GNU. You're not really missing anything. Unix is unix is unix... if it supports posix and BSD extensions, you're all set -- SGI's website even has a slew of tips for compiling with gcc (I bet you already knew this since you knew about the caveats of using gcc). At least that's how I see it.
Of course, if you actually DID want to run commercial 6.5 software on it.... you would probably also have the money to buy a new workstation or a second-hand Onyx.
It may be slow if you plan on using it as a workstation running the latest CAD apps and compilers, but it is perfectly usable and quite snappy as a desktop system. Depending on the graphics/video board(s) in it, you may be able to capture video, or do other fun things too.
;)
It's not true that they've "mostly" got R3k cpus. R3k models do take weird memory though. And all Indigos have proprietary keyboard and mouse ports. While R4k Indigos with enough memory (at least 64MB) can run Irix 6.5 just fine, R3k models can only run up to Irix 5.3. Despite what you're implying, 5.3 is also a "good" version of Irix, if older than 6.5. But come on, with the free development files from SGI, you can compile anything you want with gcc. It IS unix after all.
I actually hate to see people doing things like this to such a remarkable system. I would rather they sell them to resellers so that afficionados can get their hands on them. They DID play a role in the making of Jurassic Park after all.
As to Indigo2s being better than Indys.... well, it depends. The indy does have a built-in port for an IndyCam and it is a LOT quieter and smaller and sucks up less power. The Indigo2 is newer though, and can support r10k cpus and hardware texturing.
Except after c.
Oh, and you forgot the t in your first nitpick.
No, you're not. I found it odd to see several jokes posted so quickly and modded up so quickly.
I heard about a russian man who had his face bitten off by a bear some years ago. He had one eye left, so he could function. The pictures were pretty gruesome - a big gaping hole in his face. The russian doctors couldn't do much, so he lived wearing a scarf over his face for a long time. Eventually, an american institution offered to create a face for him. I don't remember which institution it was. Might be the same one.
But this guy that lost not just his face, both his eyes also... I shudder to think about it. It's remarkable that he had the will to go on like he did... thanks in part to a dream about his granddaughter.
From SGI's website,
Fuel: 977 BTU/hour Octane and Octane2: 2400 BTU/hour
Fuel thus runs considerably cooler than the Octanes. But it runs 77 BTU/hour hotter than an O2.
semantic confusion... my meaning was not that A/UX was mach-based but that A/UX was briefly an all original OS, according to the map, before incorporating BSD (again, according to the map)...
That being said, thanks for the info! I had never even heard of UniSoft or UniPlus. Heh, as if the history wasn't complicated enough...
Fascinating map, I could spend hours just researching from it to find out what the actual relationships are (i.e. what portions of OS A were integrated into OS B)...
One thing: IRIX is shown beginning around 1986 right out of the blue. According to SGI.com's section on IRIX, it was incepted in '82. I'm curious which it is, and whether it was derived or really conjured up from scratch.
The only others that appear out of thin air (other than UNICS) are Minix, Xinu, and Mach (A/UX briefly so before getting BSDed). Does anyone know if IRIX, like these, was an original design? It's certainly unique in its own right.
Also, did I miss UNICOS in there? As I understand it, it is a UNIX (looks like a duck, quacks like a duck) derived from SysV with some BSDisms thrown in... showed up in '85, I reckon.
Check out the internal bandwidth of those boxes, 5 year old O2's and Indigo2's still have greater memory bandwidth than the latest pentium motherboards
latest x86 = ~800MB/s memory bandwidth (or so the rumours go)
Indigo2 = 400MB/s (half x86)... see www.futuretech.vuurwerk.nl/indigo2
O2 = 2.1GB/s (lots more than either...) see www.sgi.com/o2
Advanced Micro Devices
Microstar K7T Pro motherboard 1.0GHz
#cpus 1
Spec2000 fp Base 290
Spec2000 fp Peak 302
(just showed up recently on spec.org... still doesn't do any good without G4 results, that's true... why doesn't Motorola publish spec info on the spec website anymore? even if it is just for evaluation boards, it'd be fun to see...)
anyway, spec95 results are available for Athlon 1000 and G4 450... Athlon base FP = 29.4, G4 FP = 20.4 (not sure if it's base or peak... got the info from the CPU Info Center)
as usual, take the results for what they're worth...
acquiring OpenBSD
I think one word will eliminate any further entertainment of the notion... that word would be "Theo"...
tell me about it... got a cbc recording from way back but I no longer have the betamax to watch it...
gonna have to consider getting the 'soon-to-be-released' dvd... and a dvd-player/drive to watch it...
Rock N' Rule... dig it...
The reason Intel's Pentium division and AMD's Athlon division don't make chips as cool as Crusoe and StrongARM (another Intel division now) is because there's always a market for the fastest cpus at a given time. How do they get so fast? High clockspeed and a risc-like core. This also results in high energy consumption. It's no different than Alpha, R10k, PA-RISC and USparc. This is still the easiest way to make a fast cpu. An interesting bit of cpu trivia is that one of the main designers of the Alpha is one of lead designers of the Athlon.
Take a look at the great big CPU comparison chart at bwrc.eecs.berkeley.edu/CIC and compare Spec results with power consumption, making sure to control for year of introduction. And yes, Spec scores are valid in this case because we're discussing solely the "speed" of the cpu.
The thinking has not stopped. AIM has produced some impressive PPC chips that require little power and perform quite well... just not as fast as the rest... same goes for the StrongARM -- not super fast, but its power consumption is practically nil when compared with the rest of a desktop computer (embedded systems are another matter -- not nil in this case, but pretty low still).