...Live and Neutral can not be inserted into the socket with out the longer Earth prong going in...
All well and good until the plastic earth pin snaps off a cheaply made doubly-insulated-transformer-built-onto-plug power supply, you end up jamming a biro lid into the earth socket in order to defeat the lock. How safe is that? Or you could replace the power supply I suppose...
Close - it was the plug, not the flex. Until recently it didn't have to be fitted by law. You *sometimes* had to wire it yourself but most of the time you didn't. This was presumably because some devices were sold to many different countries, with many and varied plug types.
Now, I expect the cost of the plug has been passed on to the consumer somehow, but we always have a good old bombproof 3 pin plug attached, normally with a moulded connector which is impossible to recycle onto anything else without a junction box.
I used to enjoy wiring my own plugs. I especially used to enjoy helping damsels in distress who had a new hairdryer but it had a flex with 3 bare copper wires...
You'd be better off getting an A5200 tray (or D1000 tray) and using the RAID-5 functions of Veritas Volume Manager instead. It actually has a shot at working:)
I hope your kidding.
Software RAID5 on arrays with no cache? Heavens no, it sucks. Read performance sucks pretty bad considering the number of drives involved in the stripes and write performance is worse than dreadful even on high end machines. Write performance gets *even* worse the more drives you add unless you go across arrays - even then it just sucks. It's better on Veritas than Disksuite, but not much. Mirror, don't use RAID 5 on anything other that A3x00, A1000 or T3. It's especially good on the T3 where the XORs are done on the controller and it's almost as fast as striping.
I agree though, RM6 is pretty bad but if managed properly it's deployable. I know of one of Sun's customers who threw out terabytes of A5x00 storage after the GBIC debacle - as in deposited on the pavement outside of Sun's City of London office - only to replace them with A1000's and lots of them.
New SPARC kit? Move along there's nothing to see.
on
Linux 2.4.18 Released
·
· Score: 2, Interesting
Linux was fantastic when Sun released Solaris 7 - their first 64 bit OS. It meant that you could give a new lease of life to those 'cherished' old SPARCstation 1/2/Classic/LX/10 etc. The reason? Solaris bloat. Solaris has grown up with the Sun hardware range. Versions 7 and 8 have a great deal of code which supports later generation hardware. It's getting pretty difficult to fit it on a 1 gig disk - even a custom install, let alone the 420MB drive which came with my old sparcy2.
The non-pagable kernel memory used to fit, just about, in 32MB with some to spare for buffer cache (well, 2.5.1 did) . Nowadays it just swaps horribly. Why you ask? the old SPARC workstations don't have much of the hardware which new versions of Solaris provide support for (much of it installed even if you don't have the hardware grr.). Solaris has a mature multithreaded kernel, it has amazingly well tuned, truly scalable, kernel synchronisation primitives (check out the book "Solaris Internels" - Mauro, Mc Dougall) it has in-kernel support for Sun's hardware enterprise features; dynamic reconfiguration (the ability to tell Solaris to stop using memory, CPU or IO devices on a certain system board, drain the memory to swap, re-dispatch the active processes to other CPUs, remap the IP addresses to other cards, detach the board, replace, reattach - start using the new hardware - no reboot), hotplug PCI, processor sets, dynamic system domains etc. etc.
Decent Sun boxes (by that I mean anything with more that 4MB L2 cache and SCSI disks - a curse on Ultra 5/10/X1/SunBlade 100s), will run Solaris 8 very well, plus you get a tier one Oracle/Sybase/Java platform, with all of your favourite window managers/web browsers/IRC clients etc. available for download.
Mark my words, once Linux starts making real inroads on the sort of Enterprise server kit (i.e. more than 8 SMP CPUs, and much more than 4gb RAM) that you need for serious financial/HR/government/pharms. type applications , it too will be bloated. You could argue it already is - my 486SX/8MB of RAM gave very good service as a firewall, using ipchains and kernel 2.2. Kernel 2.4 and iptables (and I suppose my new stateful filter) make it rather too slow to survive my next hardware cull. Ah well, out with the old...
Fud, fud, fud.
I can't speak for the other companies but Sun can easily afford to fund R&D on the next generation SPARC chip, they've got 6 billion $ cash in hand. Let alone investments, and have done for over 2 years.
BTW the current generation is UltraSPARCIII, UltraSPARCIV is just a fabrication improvement. Work is already underway on UltraSPARCV's design.
Sun's crown jewels are SPARC/Solaris, when Sun stops working on their own OS/CPU/Server platform it's time to stop investing in them.
...when first we practice to deceive.
It wasn't just the yanks and ruskies who were engaged in the spying game. The Imperial War Museum in London (just near Waterloo station) has a permanent exhibition about the British secret war.
The exhibition has lots of miniture radios and cameras, invisible writing equipment, escape maps, weaponry and cipher machines from M15, M16, SOE and Signals Intelligence. Super stuff, but far too many School Kids milling around on a week day.
My girlfriend owns a holiday cottage on the (UK's) North Yorkshire coast in a splendidly rural fishing village:
http://www.freefoto.com/regional/europe/united_kin gdom/england/yorkshire/runswick_bay/index.asp
I've been up to Runswick bay with her several times now, at various times of the year (Valentines - freezing snow + wonderful log fires, Summer - inclement + pints of ale while the weather improves - you get the idea).
We've always driven there (a 5-6 hour journey by motorway) and by tradition, during the journey and return, her family have always listened to the BBC's production of the Lord of The Rings (recorded presumably during early 80s?) on 90 minute cassettes. We now have the BBC's 13CD box set, which I've encoded into MP3s so that I can play them in the car.
I haven't seem the film yet, but I know it's going to have a hard time living up to this BBC production for me, I have many pleasant memories of listening to this splendid story. The cast The cast was fantastic and the music unforgettable.
WEP might be usable again - once the vendors get their arses in gear.
I spent GBP30 extra on each 128bit WEP card over cheaper WEP cards. I was particulary annoyed to find out 10 weeks later that the encryption was worthless.
If FreeSWAN wasn't such a pain in the arse to compile and configure I'd be using that (I stopped relying on kernel patches after getting my fingers burnt over the international crypto patch - Just downloaded 2.4.16? - latest crypto patch is 2.4.3. Oh and it corrupts your data if you use non-relative block numbers), however now I've had to give up using my cards - I live in a flat, I can use a long piece of cat5.
What I'm waiting for, is for Intel to sort out the problem. I don't care if they don't interoperate with other Wifi cards, I just want a cryptographically secure implementation of IVs with RC4 damn you!
It's a P200 - it's got to be out of warranty - who cares about the temperature. My P200 (running at a mighty 166MHz, the motherboard couldn't handle 200MHz) has been playing DVDs every evening in the lounge for 3 years now, if it pops I'll buy another one (if I can), or upgrade a decent box and retrofit the old CPU/Motherboard...
Sorry for the typos, I'd had a bottle of wine before I wrote my post last night.
I'm actually not that convinced by SPEC figures. From a customers point of view they are normally most concerned with whether or not the application actually runs (or is qualified by the ISV) on any chip architecture other than the one it was developed for (this is obviously less true with Java apps). The only real test for the performance of a CPU is how well the real-world application itself performs on a particular piece of kit. There are simply too many factors to take into consideration - ranging from OS performance (i.e. latency of system calls etc.), to network and disk IO (the usual killer on most real-world apps), to memory size and memory latency, to cache size and behaviour, to pipeline design (and bugs:) to ALU and FPU performance (and bugs:).
I've worked for a particular UNIX systems vendor whose chip design group work very closely with their compiler group who work very closely with their benchmarking team. A factor of ten improvement in benchmark figures is not unusual almost always achieved through low-down cunning and an incredible knowledge of the benchmarks themselves. Does it improve a real-world application's performance? Only possibly...
UltraSPARC I was Sun first 64 CPU, back in 95 as I recall. Of course you needed to wait until Solaris 7 before you got a 64 OS and hence the ability to use a 64bit address space. USI chips are detected by most 64bit versions of Solaris and it reverts to 32bit mode (I have some pre-beta UltraSPARC I hardware). This can be overridden, but leaves you vulnerable to a user land hack, which can hang the box.
Running at 167MHz these chips were hotish for their time, but compared to USII (now at a maximum of 480MHz) or USIII (just recently 1050MHz) they are rather slow. Every three years or so Sun rework the SPARC design to have better pipelines, better prediction, more TLBs etc. and speed increases in-between odd number releases are just fabrication improvements. Sun is a chip design company not a chip fabrication company.
It's hard to compare Itanium with SPARC, PA-RISC, PowerPC and Alpha - as far as I know there are no benchmarks in which is performs very well against modern 64bit RISC chips, Integer and particulary FP performance is generally considered rather inferiour.
The true test of a server class CPU is how well it handles cache coherency and memory latency issues on machines designed to support 8 or more CPUs. Itanium has not been shown to scale to these numbers. This may of course be because it's not yet been used in a server platform which supports that number of CPUs.
What I find particularly intriging is how Intel's marketing department is going to handle the clock speed differences in their product range. They have always used MHz as a marketing tool, but now they're going to have to concede that their prestigious server CPU is almost half the clock speed than that of their desktop CPU.
Re:Tell that to university sys admins
on
Linux Kernel Bugs
·
· Score: 1
Anyway, point is it isn't a good thing nor is it something that sysadmins should have to live with.
It causes a hassle, and even if it takes 1 hour to redo a box, that's still time wasted. My point all along is that this is a major security hassle.
If it's a Sun box, you can just jumpstart it. 10 seconds of a sys-admins time;
stop-A on the console
boot net - install
The box will go to it's config and image server(s), install a fresh OS, apply patches and run custom post-install scripts if required. It takes a couple of hours on older versions, but with recent Solaris 8 updates you can install from a flash archive - a bit like a Norton Ghost image (actually it's a fancy name for a.cpio.Z) - of a known good filesystem image, which takes just minutes.
And yes, I had root access on one of the shell servers at my university (a real dog of a machine), it was a buffer overflow on getopt() with Solaris 2.5.1 as I recall. Of course not being a malicous soul, all I did was type 'id' once or twice and logged back out, but it was fun at the time and I learnt quite a bit about SPARC assembler.
I have an Alcatel speedtouch USB ADSL modem, which I spent monthes trying to get working reliably under Linux. The damn thing would lock up after about 200 packets went through it, either using the open-ish source Alcatel driver (utter, utter, c**p), or the real open source user-land driver.
In the end I tracked the problem down to the UHCI controller code in the 2.4.x Linux kernel and after some brief hacking about I gave up trying to fix it. I was just about to fire up windows/winroute when I thought I might try a *BSD.
3 days later I had a pretty well locked down NAT/IPFilter gateway machine, which has been connected to my ISP for well over 100 days at a stretch (I turn it off when I go away). It operates well under load and I get excellent ping times - even with the user-land ppp - better than windows.
My only gripe with FreeBSD is the amount of documentation available. You pretty much have to work out most things for yourself, there aren't the sheer number of different HOWTOs available like there are with Linux.
Now if only I could get my wireless card to work in it...
You really need the Forte 6 compilers. Trying to compile this code on gcc is incredibly tedious, trust me. You can generate 30 day try-n-buy licenses somewhere on sun.com.
BTW if you do download gcc for SPARC, make sure that it's version 3. There is a new backend code generator for SPARC which makes it "suck less" when generating 64bit code - performance on previous versions is pretty terrible.
Slashdot Mirror
...Live and Neutral can not be inserted into the socket with out the longer Earth prong going in...
All well and good until the plastic earth pin snaps off a cheaply made doubly-insulated-transformer-built-onto-plug power supply, you end up jamming a biro lid into the earth socket in order to defeat the lock. How safe is that? Or you could replace the power supply I suppose...
Yes, but Brits have to be some of the most risk-adverse people in the world. That's why they make such crappy businessmen.
And yet somehow we have the largest finance industry in Europe...
Close - it was the plug, not the flex. Until recently it didn't have to be fitted by law. You *sometimes* had to wire it yourself but most of the time you didn't. This was presumably because some devices were sold to many different countries, with many and varied plug types.
Now, I expect the cost of the plug has been passed on to the consumer somehow, but we always have a good old bombproof 3 pin plug attached, normally with a moulded connector which is impossible to recycle onto anything else without a junction box.
I used to enjoy wiring my own plugs. I especially used to enjoy helping damsels in distress who had a new hairdryer but it had a flex with 3 bare copper wires...
Yes, I think it's a word used in Nazi standard battlefield English. A dialect often found in hollywood war films, as in:
"Hande Hoch! For you Britisher, ze vor is over."
...vehiculates the Shuttles to the launch pads...
You means 'moves' or perhaps 'transports'?
You yanks...
You'd be better off getting an A5200 tray (or D1000 tray) and using the RAID-5 functions of Veritas Volume Manager instead. It actually has a shot at working :)
I hope your kidding.
Software RAID5 on arrays with no cache? Heavens no, it sucks. Read performance sucks pretty bad considering the number of drives involved in the stripes and write performance is worse than dreadful even on high end machines. Write performance gets *even* worse the more drives you add unless you go across arrays - even then it just sucks. It's better on Veritas than Disksuite, but not much. Mirror, don't use RAID 5 on anything other that A3x00, A1000 or T3. It's especially good on the T3 where the XORs are done on the controller and it's almost as fast as striping.
I agree though, RM6 is pretty bad but if managed properly it's deployable. I know of one of Sun's customers who threw out terabytes of A5x00 storage after the GBIC debacle - as in deposited on the pavement outside of Sun's City of London office - only to replace them with A1000's and lots of them.
...Diffie-Hellman (sp?) algorithm, which is incidentally also the first public-key alg ever invented...
*cough*GCHQ*cough*
Linux was fantastic when Sun released Solaris 7 - their first 64 bit OS. It meant that you could give a new lease of life to those 'cherished' old SPARCstation 1/2/Classic/LX/10 etc. The reason? Solaris bloat. Solaris has grown up with the Sun hardware range. Versions 7 and 8 have a great deal of code which supports later generation hardware. It's getting pretty difficult to fit it on a 1 gig disk - even a custom install, let alone the 420MB drive which came with my old sparcy2.
The non-pagable kernel memory used to fit, just about, in 32MB with some to spare for buffer cache (well, 2.5.1 did) . Nowadays it just swaps horribly. Why you ask? the old SPARC workstations don't have much of the hardware which new versions of Solaris provide support for (much of it installed even if you don't have the hardware grr.). Solaris has a mature multithreaded kernel, it has amazingly well tuned, truly scalable, kernel synchronisation primitives (check out the book "Solaris Internels" - Mauro, Mc Dougall) it has in-kernel support for Sun's hardware enterprise features; dynamic reconfiguration (the ability to tell Solaris to stop using memory, CPU or IO devices on a certain system board, drain the memory to swap, re-dispatch the active processes to other CPUs, remap the IP addresses to other cards, detach the board, replace, reattach - start using the new hardware - no reboot), hotplug PCI, processor sets, dynamic system domains etc. etc.
Decent Sun boxes (by that I mean anything with more that 4MB L2 cache and SCSI disks - a curse on Ultra 5/10/X1/SunBlade 100s), will run Solaris 8 very well, plus you get a tier one Oracle/Sybase/Java platform, with all of your favourite window managers/web browsers/IRC clients etc. available for download.
Mark my words, once Linux starts making real inroads on the sort of Enterprise server kit (i.e. more than 8 SMP CPUs, and much more than 4gb RAM) that you need for serious financial/HR/government/pharms. type applications , it too will be bloated. You could argue it already is - my 486SX/8MB of RAM gave very good service as a firewall, using ipchains and kernel 2.2. Kernel 2.4 and iptables (and I suppose my new stateful filter) make it rather too slow to survive my next hardware cull. Ah well, out with the old...
Fud, fud, fud. I can't speak for the other companies but Sun can easily afford to fund R&D on the next generation SPARC chip, they've got 6 billion $ cash in hand. Let alone investments, and have done for over 2 years. BTW the current generation is UltraSPARCIII, UltraSPARCIV is just a fabrication improvement. Work is already underway on UltraSPARCV's design. Sun's crown jewels are SPARC/Solaris, when Sun stops working on their own OS/CPU/Server platform it's time to stop investing in them.
...when first we practice to deceive. It wasn't just the yanks and ruskies who were engaged in the spying game. The Imperial War Museum in London (just near Waterloo station) has a permanent exhibition about the British secret war. The exhibition has lots of miniture radios and cameras, invisible writing equipment, escape maps, weaponry and cipher machines from M15, M16, SOE and Signals Intelligence. Super stuff, but far too many School Kids milling around on a week day.
...ahh, I've just blown away mozilla's cookies, hence I can't automatically log in to slashdot. That's why I'm seeing stories from this idiot.
One that I know of. JavaOS - it ran on JavaStations. It was crap.
My girlfriend owns a holiday cottage on the (UK's) North Yorkshire coast in a splendidly rural fishing village: http://www.freefoto.com/regional/europe/united_kin gdom/england/yorkshire/runswick_bay/index.asp
I've been up to Runswick bay with her several times now, at various times of the year (Valentines - freezing snow + wonderful log fires, Summer - inclement + pints of ale while the weather improves - you get the idea).
We've always driven there (a 5-6 hour journey by motorway) and by tradition, during the journey and return, her family have always listened to the BBC's production of the Lord of The Rings (recorded presumably during early 80s?) on 90 minute cassettes. We now have the BBC's 13CD box set, which I've encoded into MP3s so that I can play them in the car.
I haven't seem the film yet, but I know it's going to have a hard time living up to this BBC production for me, I have many pleasant memories of listening to this splendid story. The cast The cast was fantastic and the music unforgettable.
WEP might be usable again - once the vendors get their arses in gear.
I spent GBP30 extra on each 128bit WEP card over cheaper WEP cards. I was particulary annoyed to find out 10 weeks later that the encryption was worthless.
If FreeSWAN wasn't such a pain in the arse to compile and configure I'd be using that (I stopped relying on kernel patches after getting my fingers burnt over the international crypto patch - Just downloaded 2.4.16? - latest crypto patch is 2.4.3. Oh and it corrupts your data if you use non-relative block numbers), however now I've had to give up using my cards - I live in a flat, I can use a long piece of cat5.
What I'm waiting for, is for Intel to sort out the problem. I don't care if they don't interoperate with other Wifi cards, I just want a cryptographically secure implementation of IVs with RC4 damn you!
It's a P200 - it's got to be out of warranty - who cares about the temperature. My P200 (running at a mighty 166MHz, the motherboard couldn't handle 200MHz) has been playing DVDs every evening in the lounge for 3 years now, if it pops I'll buy another one (if I can), or upgrade a decent box and retrofit the old CPU/Motherboard...
Sorry for the typos, I'd had a bottle of wine before I wrote my post last night.
:) to ALU and FPU performance (and bugs :).
I'm actually not that convinced by SPEC figures. From a customers point of view they are normally most concerned with whether or not the application actually runs (or is qualified by the ISV) on any chip architecture other than the one it was developed for (this is obviously less true with Java apps). The only real test for the performance of a CPU is how well the real-world application itself performs on a particular piece of kit. There are simply too many factors to take into consideration - ranging from OS performance (i.e. latency of system calls etc.), to network and disk IO (the usual killer on most real-world apps), to memory size and memory latency, to cache size and behaviour, to pipeline design (and bugs
I've worked for a particular UNIX systems vendor whose chip design group work very closely with their compiler group who work very closely with their benchmarking team. A factor of ten improvement in benchmark figures is not unusual almost always achieved through low-down cunning and an incredible knowledge of the benchmarks themselves. Does it improve a real-world application's performance? Only possibly...
UltraSPARC I was Sun first 64 CPU, back in 95 as I recall. Of course you needed to wait until Solaris 7 before you got a 64 OS and hence the ability to use a 64bit address space. USI chips are detected by most 64bit versions of Solaris and it reverts to 32bit mode (I have some pre-beta UltraSPARC I hardware). This can be overridden, but leaves you vulnerable to a user land hack, which can hang the box.
Running at 167MHz these chips were hotish for their time, but compared to USII (now at a maximum of 480MHz) or USIII (just recently 1050MHz) they are rather slow. Every three years or so Sun rework the SPARC design to have better pipelines, better prediction, more TLBs etc. and speed increases in-between odd number releases are just fabrication improvements. Sun is a chip design company not a chip fabrication company.
It's hard to compare Itanium with SPARC, PA-RISC, PowerPC and Alpha - as far as I know there are no benchmarks in which is performs very well against modern 64bit RISC chips, Integer and particulary FP performance is generally considered rather inferiour.
The true test of a server class CPU is how well it handles cache coherency and memory latency issues on machines designed to support 8 or more CPUs. Itanium has not been shown to scale to these numbers. This may of course be because it's not yet been used in a server platform which supports that number of CPUs.
What I find particularly intriging is how Intel's marketing department is going to handle the clock speed differences in their product range. They have always used MHz as a marketing tool, but now they're going to have to concede that their prestigious server CPU is almost half the clock speed than that of their desktop CPU.
Anyway, point is it isn't a good thing nor is it something that sysadmins should have to live with.
.cpio.Z) - of a known good filesystem image, which takes just minutes.
It causes a hassle, and even if it takes 1 hour to redo a box, that's still time wasted. My point all along is that this is a major security hassle.
If it's a Sun box, you can just jumpstart it. 10 seconds of a sys-admins time;
stop-A on the console
boot net - install
The box will go to it's config and image server(s), install a fresh OS, apply patches and run custom post-install scripts if required. It takes a couple of hours on older versions, but with recent Solaris 8 updates you can install from a flash archive - a bit like a Norton Ghost image (actually it's a fancy name for a
And yes, I had root access on one of the shell servers at my university (a real dog of a machine), it was a buffer overflow on getopt() with Solaris 2.5.1 as I recall. Of course not being a malicous soul, all I did was type 'id' once or twice and logged back out, but it was fun at the time and I learnt quite a bit about SPARC assembler.
Solaris 9 (the beta is out) runs linux binaries.
And so does Solaris 8, but only on x86 and not very well.
I have an Alcatel speedtouch USB ADSL modem, which I spent monthes trying to get working reliably under Linux. The damn thing would lock up after about 200 packets went through it, either using the open-ish source Alcatel driver (utter, utter, c**p), or the real open source user-land driver.
In the end I tracked the problem down to the UHCI controller code in the 2.4.x Linux kernel and after some brief hacking about I gave up trying to fix it. I was just about to fire up windows/winroute when I thought I might try a *BSD.
3 days later I had a pretty well locked down NAT/IPFilter gateway machine, which has been connected to my ISP for well over 100 days at a stretch (I turn it off when I go away). It operates well under load and I get excellent ping times - even with the user-land ppp - better than windows.
My only gripe with FreeBSD is the amount of documentation available. You pretty much have to work out most things for yourself, there aren't the sheer number of different HOWTOs available like there are with Linux.
Now if only I could get my wireless card to work in it...
Er, no - bluetooth is a "personal area network" technology, not more than ~10 metres I'm afraid.
You really need the Forte 6 compilers. Trying to compile this code on gcc is incredibly tedious, trust me. You can generate 30 day try-n-buy licenses somewhere on sun.com.
BTW if you do download gcc for SPARC, make sure that it's version 3. There is a new backend code generator for SPARC which makes it "suck less" when generating 64bit code - performance on previous versions is pretty terrible.