Looking at the specs, I noticed that they said there was a crusoe chip inside it. Aha, I though, that'll appeal to the geek community.
However, if you only need the horsepower to firewall a cable connection even a 486 would spend most of its time idle, so I'd doubt if they'd spend so much money (and 16Mb of translation RAM) on the processor.
If I made one of these, I'd use a strongarm.
In the UK, i've been worried about this special deal that teachers get - They get subsidised training, computers and software but of course it HAS to be Microsoft, They get Windows2000 and Office 2000.
In a lot of ways its good, because the UK could do with more computer-literate teachers, but to push office tools onto them provides them with something that they don't need, causes incompatibilities with school systems which are usually running what was state of the art maybe 4 years ago and I believe a certain amount of pro-microsoft evangelism goes on.
When the heads of Blackthorn and Thumbswood schools here asked me about it (I've provided good advice in the past) I suggested throwing away Office2000 and running Office97 for compatibility with the school.
Personally though, I don't think the school children should have access to high versions of word for the same reason that you should learn to do maths without a calculator - these word versions will spell-check and prompt all sorts of other things the children should know how to do themselves. Microsoft "write" should be enough.
I looked and saw it was made up of 18 or 32 Gb disks, I figured that some sort of tardis effect was required to get 600Gb into a 1u space.
(thats 20 disks + RAID overhead).
Neither I nor my workplace actually paid for RedHat releases on Sparc, but I have helped with support, bugfixing and development so I don't feel bad, and work's evaluation is still at a very early stage.
The sort of company that would run linux on sparc is likely to be very slow in taking up this new upstart operating system, "linux".
Ah well, I suspected there was more mileage in linux on the os390 because of the efficiency gains in reaching legacy data.
I run a few machines with sparclinux, and have been involved with it for quite a few years (as a lurker). Its painfully obvious that SUN don't provide the same levels of support for their hardware that you can find from any other vendor (eg. Digital{now Compaq} providing Linus with an alpha, coding efforts within IBM and SGI...)
Yet, SUN are happy to gain news inches on the back of linux.
Looks like I'll have to move to debian or suse though...
and of course, if the sparc based machines have to move then my intel ones will have to follow.
When you add a driver to the solaris kernel you don't actually link it then....
drivers are included dynamically at boot time the same as linux modules, and therefore should be treated the same.
I can't see that SUN has done anything wrong, maybe slightly rude in using someone's code without asking, but when you GPL something you explicitly give up your ability to restrict how others use your code.
Performance wise, NFS and samba have come out almost identical for me. I guess you've got it configured wrongly if you've got a big difference between them.
Administration wise, samba uses user names, which can be made to match between UNIX and NT, its easy. Trying to introduce UID's to NT has previously (e.g. with Maestro) been a real pain.
There's a parallel happening here on the internet, ISP's are using border caches, most of which have the capability to rewrite URL's or change content.
These caches are transparent and unavoidable.
Does anyone know (for the paranoid) of any trusted proxy servers, and how do we know they're to be trusted.
Most of your reasons stand, but heat and size don't - have you seen a 270MHz Arm chip - I have, its the size of my little finger nail, and needs no heat sink.
Though I would say that if you were looking for a processor for this purpose I would think it would have to be sparc or i386 for there to be a client available from seti@home.
Hmmm, I seem to remember that SCO is the owner of the UNIX trademark.
Linux has up to now been called unix-ish because designing something to be POSIX compatible makes a system behave very unixy anyway. However the real UNIX trademark is based on money, not just compatibility. Maybe the Caldera move will actually enable us to call it UNIX, in the same way that Solaris, AIX and HPUX are all UNIX.
also...
SCO also owns a lot of intellectual property which whilst a lot of it has been re-engineered as open source there are a few things that would give Caldera an edge.
When comparing ARM chips with Intel chips its extremely difficult to cater for the differences in operating systems. Even a 30MHz ARM feels significantly faster than a 300MHz Pentium when comparing RiscOS to Windows. When running a similar operating system I suspect that the ARM chip may be double the speed of the intel at the same clock speed. The Intel processor will carry out more complex instructions in more clock cycles, but with parallelism in the newer processors the optimisations from a good compiler may make a huge difference in favour of the intel chip.
I've just bought one of these faster processors for an Acorn PowerPC. Had this system been RiscOS compatible I would have begrudged paying top dollar for the upgrade, but when dealing with users your primary driver for the computer is the applications it will run.
RiscOS is an operating system which gives MacOS a run for its money in the usability stakes. These creatures are fast and silent (no fans on the CPU, hell, not even a heat sink:-) )
Unfortunately I believe this particular machine's battle for survival will be lost over applications and device drivers, no matter how good the motherboard is its not much use when you have nothing to run and can't plug devices in.
Firstly, the heap sort. When I was learning about computers I studied sorting methods. The idea that inputting your random data into a linked list allowed you to just read it out in the right order was a revelation. The next important algorithm that I came across but unfortunately can't describe properly here is used to produce molecular shapes from X-ray diffraction patterns. This helped mankind to understand DNA. My third most important algorithm is the state engine that drives the TCP/IP stack. This is whats bringing my burblings to you now. I'm happy with quite a few of the algorithms in the top 10 list - fourier transforms in particular but some of them seem rather esoteric.
Maintenance of a large number of machines comes down to managing differences between them. If they're the same then handling 200 is no worse than handling 2000. E10000's are far more tricky beasts than a simple linux box, especially if you're wanting to do domaining (the only reason you would choose an e10000 over an e6500).
There are some big ISV's that will port to linux far faster than other OS's. This is for a variety of reasons including: 1. There are a lot of eyeballs on linux at the moment. 2. If you port to linux, you can work on something that scales between a huge range of machines. 3. Linux will be the first commercial OS available on the ia64 platform. I know that most other OS's are better in different ways - NT's GUI, VMS's clustering, OS390's reliability, Dynix's NUMA but look at take-up per OS per year and you'll see a clear leader.
But on a 32-bit normal PCI, at 33MHz normal speed this card soaks not 20% but 80%. Also, if it hogs the bus with large data transfers like scsi can then the latency of other bus requests will rise. 1500 bytes from a max-size network packet will grab the bus for a far shorter time than a multi-Mb collected DMA from disk.
From memory, the bandwidth they're advertising is a significant percentage of the PCI bandwidth - which infers that putting two of these in will not increase your throughput - and even one of these running flat out will mean other devices such as PCI video and networking may struggle.
I had hoped the quantum leap we benefitted from when moving from ISA to PCI would last a bit longer than 7 years, but I suppose thats 3 x moores law doubling so its not that bad.
Compaq 8500's and Sparc e420's have multiple busses. If you can afford one of these puppies you can afford a good platform to run it on.
Well I've played on it at work - lynx, apache and samba all compiled fine.
It looks mostly like linux on intel - configure can barf when it sees *-s390-* as the host to configure against but thats an application configuration problem, easily worked round.
The Hawking radiation effect seems to assume that more "negative energy" halves get absorbed than positive. Statistically I would assume it would be 50% either way, so there would be no net radiation.
Thus this test may be invalid for reasons other than testing for a singularity.
Since I'm under NDA not to discuss my Demon.net -fronted ADSL knowledge I'll just stick to saying that I look forward to finding out about NTL cable modems.
I run a linux server at home, and want to host a MUD.
My mother-in-law's eyes are failing, so I've been sensitive to the needs of the partially-sighted. Most "blind" people do have partial sight, and can often use a computer screen if the picture is modified for them.
This is one area that the microsoft windows crowd have an advantage in - from my research the partially sighted find it easier to use a split screen to use the computer, the top half shows a normal view, which is good enough for seeing if something is flashing or a new window has popped up and the bottom half shows a magnified view area taken from the top.
My opinion is that since so much content is produced for the sighted by the sighted it is often impossible to make a meaningful audio version of what is going on on screen.
With respect to this article, audio prompts for the partially sighted on OS installation is probably not as useful as buying a pre-installed machine from VA or Dell which can also include support.
The thing that worries me most is if a well-meaning ISP (or not) uses a transparent proxy cache with some of this censoring software. You'd not easily be able to tell that pieces of the internet were unavailable.
Its only a short step between denying content and rewriting URL's, but lets not go to that scary place this time...
If you buy software that is designed to limit your access to the internet, and if it extends its censoring unethically at least you can disagree and stop using it. This assumes that you can actually find out.
Slashdot Mirror
Looking at the specs, I noticed that they said there was a crusoe chip inside it. Aha, I though, that'll appeal to the geek community.
However, if you only need the horsepower to firewall a cable connection even a 486 would spend most of its time idle, so I'd doubt if they'd spend so much money (and 16Mb of translation RAM) on the processor.
If I made one of these, I'd use a strongarm.
In the UK, i've been worried about this special deal that teachers get - They get subsidised training, computers and software but of course it HAS to be Microsoft, They get Windows2000 and Office 2000.
In a lot of ways its good, because the UK could do with more computer-literate teachers, but to push office tools onto them provides them with something that they don't need, causes incompatibilities with school systems which are usually running what was state of the art maybe 4 years ago and I believe a certain amount of pro-microsoft evangelism goes on.
When the heads of Blackthorn and Thumbswood schools here asked me about it (I've provided good advice in the past) I suggested throwing away Office2000 and running Office97 for compatibility with the school.
Personally though, I don't think the school children should have access to high versions of word for the same reason that you should learn to do maths without a calculator - these word versions will spell-check and prompt all sorts of other things the children should know how to do themselves. Microsoft "write" should be enough.
I looked and saw it was made up of 18 or 32 Gb disks, I figured that some sort of tardis effect was required to get 600Gb into a 1u space.
(thats 20 disks + RAID overhead).
Nope, just shoddy reporting thats all.
To follow up my own message,
Neither I nor my workplace actually paid for RedHat releases on Sparc, but I have helped with support, bugfixing and development so I don't feel bad, and work's evaluation is still at a very early stage.
The sort of company that would run linux on sparc is likely to be very slow in taking up this new upstart operating system, "linux".
Ah well, I suspected there was more mileage in linux on the os390 because of the efficiency gains in reaching legacy data.
I run a few machines with sparclinux, and have been involved with it for quite a few years (as a lurker). Its painfully obvious that SUN don't provide the same levels of support for their hardware that you can find from any other vendor (eg. Digital{now Compaq} providing Linus with an alpha, coding efforts within IBM and SGI...)
Yet, SUN are happy to gain news inches on the back of linux.
Looks like I'll have to move to debian or suse though...
and of course, if the sparc based machines have to move then my intel ones will have to follow.
When you add a driver to the solaris kernel you don't actually link it then....
drivers are included dynamically at boot time the same as linux modules, and therefore should be treated the same.
I can't see that SUN has done anything wrong, maybe slightly rude in using someone's code without asking, but when you GPL something you explicitly give up your ability to restrict how others use your code.
Performance wise, NFS and samba have come out almost identical for me. I guess you've got it configured wrongly if you've got a big difference between them.
Administration wise, samba uses user names, which can be made to match between UNIX and NT, its easy. Trying to introduce UID's to NT has previously (e.g. with Maestro) been a real pain.
No quirkiness, it just works, very well.
There's a parallel happening here on the internet, ISP's are using border caches, most of which have the capability to rewrite URL's or change content.
These caches are transparent and unavoidable.
Does anyone know (for the paranoid) of any trusted proxy servers, and how do we know they're to be trusted.
It doesnt seem to work with this particular ASF stream...
? StreamID=528441&r
Resolving host: 'playlist.broadcast.com'
connecting to: playlist.broadcast.com
sending request [296 bytes]
waiting for reply...
reply: 200 - OK
Parsing URL: 'mms://wmcontent02.broadcast.com/ibm/6/528441.asf
u=UNKNOWN'
Resolving host: 'wmcontent02.broadcast.com'
connecting to: wmcontent02.broadcast.com
sending request [316 bytes]
waiting for reply...
reply: 200 - OK
receiving ASF header...
Opening file '528441.asf?StreamID=528441&ru=UNKNOWN' for output
connecting to: wmcontent02.broadcast.com
sending request [430 bytes]
waiting for reply...
reply: 200 - OK
receiving ASF header...
receiving stream...
0 kB ( 0%), HDR: $4824, 211 bytes, seq $00000000, tc: ???
receiving ASF header...
Unable to parse this ASF header!
Most of your reasons stand, but heat and size don't - have you seen a 270MHz Arm chip - I have, its the size of my little finger nail, and needs no heat sink.
Though I would say that if you were looking for a processor for this purpose I would think it would have to be sparc or i386 for there to be a client available from seti@home.
Hmmm, I seem to remember that SCO is the owner of the UNIX trademark.
Linux has up to now been called unix-ish because designing something to be POSIX compatible makes a system behave very unixy anyway. However the real UNIX trademark is based on money, not just compatibility. Maybe the Caldera move will actually enable us to call it UNIX, in the same way that Solaris, AIX and HPUX are all UNIX.
also...
SCO also owns a lot of intellectual property which whilst a lot of it has been re-engineered as open source there are a few things that would give Caldera an edge.
When comparing ARM chips with Intel chips its
extremely difficult to cater for the differences
in operating systems.
Even a 30MHz ARM feels significantly faster than
a 300MHz Pentium when comparing RiscOS to Windows.
When running a similar operating system I suspect
that the ARM chip may be double the speed of the
intel at the same clock speed.
The Intel processor will carry out more complex
instructions in more clock cycles, but with
parallelism in the newer processors the
optimisations from a good compiler may make a huge
difference in favour of the intel chip.
I've just bought one of these faster processors
:-) )
for an Acorn PowerPC. Had this system been RiscOS
compatible I would have begrudged paying top
dollar for the upgrade, but when dealing with
users your primary driver for the computer is the
applications it will run.
RiscOS is an operating system which gives MacOS a
run for its money in the usability stakes.
These creatures are fast and silent (no fans
on the CPU, hell, not even a heat sink
Unfortunately I believe this particular
machine's battle for survival will be
lost over applications and device drivers, no
matter how good the motherboard is its not much
use when you have nothing to run and can't plug
devices in.
Firstly, the heap sort. When I was learning about computers I studied sorting methods. The idea that inputting your random data into a linked list allowed you to just read it out in the right order was a revelation.
The next important algorithm that I came across but unfortunately can't describe properly here is used to produce molecular shapes from X-ray diffraction patterns. This helped mankind to understand DNA.
My third most important algorithm is the state engine that drives the TCP/IP stack. This is whats bringing my burblings to you now.
I'm happy with quite a few of the algorithms in the top 10 list - fourier transforms in particular but some of them seem rather esoteric.
Maintenance of a large number of machines
comes down to managing differences between
them. If they're the same then handling
200 is no worse than handling 2000.
E10000's are far more tricky beasts than a
simple linux box, especially if you're wanting
to do domaining (the only reason you would choose
an e10000 over an e6500).
There are some big ISV's that will port to linux far faster than other OS's. This is for a variety of reasons including:
1. There are a lot of eyeballs on linux at the moment.
2. If you port to linux, you can work on something that scales between a huge range of machines.
3. Linux will be the first commercial OS available on the ia64 platform.
I know that most other OS's are better in different ways - NT's GUI, VMS's clustering, OS390's reliability, Dynix's NUMA but look at take-up per OS per year and you'll see a clear leader.
But on a 32-bit normal PCI, at 33MHz normal speed this card soaks not 20% but 80%. Also, if it hogs the bus with large data transfers like scsi can then the latency of other bus requests will rise. 1500 bytes from a max-size network packet will grab the bus for a far shorter time than a multi-Mb collected DMA from disk.
From memory, the bandwidth they're advertising
is a significant percentage of the PCI bandwidth -
which infers that putting two of these in will not
increase your throughput - and even one of these
running flat out will mean other devices such as PCI video and networking may struggle.
I had hoped the quantum leap we benefitted from
when moving from ISA to PCI would last a bit
longer than 7 years, but I suppose thats 3 x
moores law doubling so its not that bad.
Compaq 8500's and Sparc e420's have multiple
busses. If you can afford one of these
puppies you can afford a good platform to run it
on.
Well I've played on it at work - lynx, apache
and samba all compiled fine.
It looks mostly like linux on intel - configure
can barf when it sees *-s390-* as the host
to configure against but thats an application
configuration problem, easily worked round.
The Hawking radiation effect seems to assume
that more "negative energy" halves get absorbed
than positive. Statistically I would assume it
would be 50% either way, so there would be no
net radiation.
Thus this test may be invalid for reasons
other than testing for a singularity.
Since I'm under NDA not to discuss my Demon.net -fronted ADSL knowledge I'll just stick to saying that I look forward to finding out about NTL cable modems.
I run a linux server at home, and want to host a MUD.
Non-sucky
- Barney
- cream-coloured optical wheelmouse
- Age of Empires
- most other microsoft hardware
- Word 2
Sucky
- "Nike" mouse
- anything that stops working unexpectedly
I think that covers it all.
My mother-in-law's eyes are failing, so I've been sensitive to the needs of the partially-sighted. Most "blind" people do have partial sight, and can often use a computer screen if the picture is modified for them.
This is one area that the microsoft windows crowd have an advantage in - from my research the partially sighted find it easier to use a split screen to use the computer, the top half shows a normal view, which is good enough for seeing if something is flashing or a new window has popped up and the bottom half shows a magnified view area taken from the top.
My opinion is that since so much content is produced for the sighted by the sighted it is often impossible to make a meaningful audio version of what is going on on screen.
With respect to this article, audio prompts for the partially sighted on OS installation is probably not as useful as buying a pre-installed machine from VA or Dell which can also include support.
You may find the following more useful... http://linuxmall.com/news/feature s/000322zipspeak
The thing that worries me most is if a well-meaning ISP (or not) uses a transparent proxy cache with some of this censoring software.
You'd not easily be able to tell that pieces of the internet were unavailable.
Its only a short step between denying content and rewriting URL's, but lets not go to that scary place this time...
If you buy software that is designed to limit your access to the internet, and if it extends its censoring unethically at least you can disagree and stop using it. This assumes that you can actually find out.
--