I've still yet to see an alpha build of OpenOffice, or any other 64-bit arch.
Sun bought StarOffice, so it's not entirely their fault, however, the fact that it still doesn't properly build is a problem.
However, the fact that it still doesn't work is a blemish on StarDivision, Sun and those working on OpenOffice.
It's the Netscape situation all over, the code is simply not up to par, and it's no suprise that things break. (Anyone remember how long Mozilla took to actually become usable again, due to needing to rewrite, or have it become worse? OOo is better, but it's still not great.)
According to what I've heard, it should be fixed in OO 2. Which would be nice, if it happens.
OOo has not been "64-bit" clean since it was open sourced.
There are NO 64-bit builds of it that I am aware of since it has come out. (alpha, sparc64, ppc64, etc.)
It is not entirely Sun's fault, nor credit, as they bought StarOffice from I believe StarDivision. (Might have gotten the name wrong.)
And unlike almost all other 32->64-bit conversions, there are definite improvements from x86->x86_64 (amd64, e64, or whatever one wants to call it), and OpenOffice is *still* slow enough I'd take even a 10% performance gain from that. (I can run the latest Abiword on a Pentium 75 just fine, for comparison. OpenOffice is horribly slow in comparison to almost any other Office suite, only MS Office *may* be as slow (on a bad day).) It makes sense to build OOo as a 64-bit app, because unlike other types, *sun sparcs* it doesn't slow down in 64-bit mode, and actually runs faster.
I've got a 3000+ Winchester (1.8GHz, nominal) which goes up to 2.8GHz (and a.1V increase 1.4V->1.5, I'll use 2.7GHz as reference, 300FSB, in this post.). However, the vast majority of the time, it runs at 1000MHz@1.1V, because contrary to the gentoo jokes, that doesn't all that much CPU time, and it more or less just idles.
This is achieved simply by messing with the FSB, and having a motherboard that allows other modifications (A8N SLI)
The biggest problem with overclocking, is that I have to disable AMD's cool and quiet. (The chip won't run at 1.5GHz 300x5 at 1.1V)
Now, if you take into consideration that Pretty much any chip above a P3-733 uses more power (by spec) than an Athlon 64 at idle (1GHz), it translates into savings of energy, if it's left on for any significant period of time. (And even if you locked the clock on slow, it'd still perform better.)
Unlike AMD, Intel is in a pretty bad situation heat/power wise.
"whatever legal entity owns the Linux kernel in the copyright sense"
There is no one, in fact, you've got hundreds, if not thousands of different people's code, all owned by them. Most simply like working on it. Some have rather 'Free or die' mentalities. Some (IBM) have cash, and no reason to want it to be closed.
So the Linux Kernel is a very very bad example.
Now, take for example a 1 person OS project. If it were bought, the person would likely be hired to work on it. The project would likely get more time devoted to it. As long as it remained open, there is no loss. (Of course, that's a big issue to many OS developers I know, they wouldn't mind being paid to do what they like to do for fun. However, they don't want it closed.)
Please, explain to me: a) Why you want a GUI on the server itself.
b) Why you cannot use remote access without a GUI. (ssh)
c) Why if you really need graphical applications running on the server you cannot use remote access. (ssh+X forwarding, vnc, rdesktop, etc)
Having a display on a server shows a lack of ability to administer said server remotely, which any competent admin should be perfectly capable of doing, barring hardware failure.
Dual Xeon boards that take 32GB RAM are common enough, but I think that's 16GB per processor and I have no idea how it's addressed or whether any given application knows how to address it.
It's addressed via PXE/"Highmem" extensions, allowing the p6 core to use the 36-bit real memory bus to remap that into the 32-bit virtual address. It requires any app using more than 4 gig to do some "weird shit" with memory requests, as well as in general only having 3-3.5Gig available at any one time (OS virtual overhead), with parts of the program duplicated a number of times. (Or at least that's how it was, Hopefully it's improved since then)
And we aren't even considering how horribly that would perform on anything more than a dual. Shared bus = easy 'multi-core' processors = Shit for scaling to more processors than 2. Any server chip is going to have better than that, hell, the original athlons did (because they used an alpha ev6 bus)
One of the few advantages Sparcs have over Xeon, has evaporated as Intel follows AMD's x86-64 spec, and replaces the memory with 48-bit virtual (Opterons/A64s only do 40-bit physical. Oh dear, by the time I've got that much ram, I'll have a better processor!)
The Opteron/Nforce4 boards are still the only 16x/16x sli solutions available.
I just double checked on Intel's website, and the best I could find was 8x/8x (3 x8 and 1 x4 PCI express slots (28 lanes total)) And with that it is not possible to have multiple x16 slots (Heck, it's impossible to have 1) (It's possible I missed a better one. I was looking in the server section.)
The main reason that Tyan can do that is because of AMD's superior Hypertransport-based bus design in Opterons, over the shared bus favored by Intel. It's also the reason why Opteron scales a lot better than Xeon.
The other reason Tyan can do that is that Nvidia realized how easy it would be to make very slightly different chipsets that facilitated that. Basically they are just Nforce 4 chipsets, that can operate in parallel, giving 40 Pci express lanes (2-way) or 80 PCI express lanes for a 4-way Opteron. (Note a maximum of 4 x16s, as the other 16 can only be a max of x4, due to the 20 lanes per nforce4)
You can't do x16/x16 with any Intel Processor, as of now. (Though having seen how little x16/x4 or x16/x2 hurts benchmarks (vs standard x8/x8) I'm not convinced it's a big deal at all.)
Actually no, it's only caught up to Mac OS X (and actually exceeded, due to the way 64-bitness is in OS X so far.).
This is simply the OS that's running in 64-bit mode now. The programs are all still 32-bit. (Admittedly, this means a lot more for Windows vs Mac OS, as the underlying command set changes just a bit ia32->x86_64.)
Now, if they could catch up to Redhat 6.0 on alpha, that'd be impressive.;)
(Note, that while NT was on Alpha, it was treated as a 32-bit arch, and the first fully 64-bit port of Windows was to ia64...the 2nd 64-bit arch it was on... All the sudden I get this feeling of impending doom for x86_64.)
That's interesting... because SB Live! cards do NOT do hardware mp3 decoding.
Any of them.
The Soundblaster Live! and the variants MP3+ & X-gamer are the exact same card... with different software bundles. At least the original ones. Later cards have some differences, but the only thing the MP3 & gamer ones had was different software bundles.
If you believe(d) otherwise, and paid money for it... you've been had.
(I'm unable to find a particular email describing the differences between the cards, from people actually writing drivers for them, at the moment.)
NASA's Columbia cluster
^ 512-way SGI machines running Linux (actually 20 of them...)
Not to mention "Columbia's record results were achieved running the LINPACK benchmark on 8,192 of the NASA supercomputer's 10,240 processors. Columbia also achieved an 88 percent efficiency rating on the LINPACK benchmark, the highest efficiency rating ever attained in a LINPACK test on large systems." from http://www.sgi.com/company_info/newsroom/press_rel eases/2004/october/worlds_fastest.html
... or do it the way konqueror's "Smart" mode does:
Only allow popups in response to mouse clicks.
Durring the time I have used this. (Quite a while, I'd have to check when it got introduced.)
I can remember two sites who got around it (one being a BTVS guide site), and I've only got 2 sites on what is essentially an override list. (ie: they wouldn't work, AND I did want them to pop up things.) One of those being dlink.com
Considering that this is AT LEAST a year, I'd say it's pretty effective.
(Sorry to shoot down you example.;) )
The only way that worked, was when SGI owned Cray for a while, Cray was making Alpha systems.
More commonly people confuse it, because DEC did use mips chips.
SGI's Altrix is the only system that actually gives the Itanium enough memory bandwidth to work in more than 2-way multiprocessor. Intel STILL has a shared bus processor design. SGI's chipset + other hardware, makes it a NUMA design with 2 processors accessing each chip. (Ie: every processor has a twin which shares northbridge access, but that northbridge allows other chips to access the memory.)
That might remind you of other 64-bit chips. One that's pretty cheap called an Opteron. (Heck once it gets dual core, it'll be even closer. Provided you just want 16 cores... or more, if you want to design a whole new chipset, like SGI did for Itanium;) )
I see no problems in a week old cvs build. Dispite trying both.
I'll check the refresh issue above.
...checked, and it doesn't work. However, I figured out both why it didn't work, and a temporary fix.
"Smart" Javascript (Config Konqueror->Java & Javascript->Javascript tab-> Open window policy: Smart) blocks it.
The reason (and why Smart really is smart) is it only allows new windows to be opened in response to a mouse click. (no OnLoad events or in this case I believe a refresh timer.) Thus the pop-up is NOT allowed to open. This also lets the vast majority of legitimate sites work properly. (I've only got 2 listed that it doesn't work with...)
Konqueror's SMART setting prevents it. Allow doesn't, Ask might (depends on user...). Deny also prevents it.
Re: One learns something useful everyday. Thanks.
on
Preview of KDE 3.4
·
· Score: 1
This is similar to my experences. Generally latency is 500ms, Generally ranging 400-600ms, the highest I've personally seen is 1100ms. (Compared to any other type of home connection, that's horrible.) Also, TRY to have 4 bars, otherwise the latency goes way up.
Sanyo 8100, Sanyo 4900, and Sanyo VM5040 (Whatever the one that can take video as well is) all work. Same instructions. (#777 etc) (And have the same cable interface)
From looking at usenet it appears, or at least the general belief seems to be: There's a limit at which they flag you for watching. I've heard 150, 200, 250MB/month, after which they send a letter telling you to cut it out or buy a plan, otherwise, they will cut off your "unlimited" vision service (Which means some relatively huge rate per kb).
Though, It's been said that sprint really doesn't care too much, provided you aren't using it constantly or downloading a lot. (Ie using maximium bandwidth, because that does effectively cut out multiple voice circuits I forget exactly what, but voice is significantly less than 144k claimed by sprint. I seem to recall 8, 16, and 24 but it's several voice lines. If it's done at night, they don't seem to care at all, durring the day on the other hand...)
All rumors, and such. I know I've used more than 100MB/month before, and had no trouble at all. (Due to that, due to other things... yes)
That would be nice, but 10k rpm drives have three disadvantages.
#1 They consume much more power. #2 They make more heat. #3 The majority are quite loud. (Doesn't mean all) But unless you buffer the sound a lot, which insulates the drive more, compounding #2)
On a friend's brand new dell (1.5GHz P-M, Radeon 9600, the thing that gets hot to the touch is not the chip or radeon (when playing something 3d), it's the hard drive.
Another thing is modern moble processors will clock down that far already, (assuming they have absolutely equal performance and power consumption at ths same MHz, and I think Pentium Ms and Athlon mobiles both have an advantage there.) thus there is little to no gain from that. Low end power consumption, it would be the same, but there would be no way to get more performance, as P-Ms and A-Ms do by clocking up. (Though if the chip were designed specifically for this, you could use one of their cores, and cut the pipeline back down.)
So, yeah kinda. What will work however, is having a huge amount of RAM, so that things wanted on the hard drive are already in memory. This would likely resolve the issue with my friend's laptop and allow the drive to spin down, consuming less power, making less heat AND actually making the programs run faster.
You don't WANT to run Itanium 8-way.
I'd be willing to bet an 8-way Operton against a straight SMP Itanium on a well threaded application.
I believe that the Itanium's SHARED bus to memory is 6.4GB/sec (It was at one time, I'm not sure if that's been upped. I'm not sure, but that may also include talking to everything else. (I thought it had a second bus for that. I hope they do.)) This means at maximum theoretical performance each Itanium will have a maximum of.8GB/sec memory bandwidth. Vs a Opteron 8xx, with something like the 6.4 GB/sec memory bandwidth per processor. Occasionally, something won't be in an Opteron's local memory, and it will be slower, but HT is still 4.8GB/sec. Itanium is fast per clock yes, and single processor wise would probably compare favorably against a Operon 8xx. However, the Opeteron's use of NUMA means that As it scales, it's going to begin, on any threaded application, to completely thrash the Itanium's pure SMP w/shared bus.
Which is why no one, other than *possibly* HP (I don't know exactly how their Integrity servers are layed out), does large pure-SMP Itaniums. SGI essentially attempts to emulate the Opteron's design with their Itanium design, which only does 2 way smp, with lots of interconnects. The interconnects are actually faster than the Opterons, but are also a lot more expensive.
Price/Performance unless on perhaps one or more specific apps that are highly tuned, will see the Opteron wallop the Itanium.
Languages evolve. Deal with it.
What are you some sort of grammer exceptionist? "Its" does't follow standard grammatical rules for contractions. It's evovling towards actually fitting general rules in this case. Isn't it better that way? There is however, another possibility: perhaps you just like being an ass with your superior knowledge of useless things?
IMNSHO It's better. Quit with the knee-jerk reactions. You understood what was meant, and is that not the general purpose of the english language?
Slashdot Mirror
Welcome to the trouble that is 64-bit OpenOffice.
I've still yet to see an alpha build of OpenOffice, or any other 64-bit arch.
Sun bought StarOffice, so it's not entirely their fault, however, the fact that it still doesn't properly build is a problem.
However, the fact that it still doesn't work is a blemish on StarDivision, Sun and those working on OpenOffice.
It's the Netscape situation all over, the code is simply not up to par, and it's no suprise that things break. (Anyone remember how long Mozilla took to actually become usable again, due to needing to rewrite, or have it become worse? OOo is better, but it's still not great.)
According to what I've heard, it should be fixed in OO 2. Which would be nice, if it happens.
OOo has not been "64-bit" clean since it was open sourced.
There are NO 64-bit builds of it that I am aware of since it has come out. (alpha, sparc64, ppc64, etc.)
It is not entirely Sun's fault, nor credit, as they bought StarOffice from I believe StarDivision. (Might have gotten the name wrong.)
And unlike almost all other 32->64-bit conversions, there are definite improvements from x86->x86_64 (amd64, e64, or whatever one wants to call it), and OpenOffice is *still* slow enough I'd take even a 10% performance gain from that. (I can run the latest Abiword on a Pentium 75 just fine, for comparison. OpenOffice is horribly slow in comparison to almost any other Office suite, only MS Office *may* be as slow (on a bad day).) It makes sense to build OOo as a 64-bit app, because unlike other types, *sun sparcs* it doesn't slow down in 64-bit mode, and actually runs faster.
Was OpenOffice itself 64-bit?
Check if you think it was.
(Not that that would be any faster on sparc64 vs sparc32)
They did have cache, it was only L1 though.
Yes and no.
.1V increase 1.4V->1.5, I'll use 2.7GHz as reference, 300FSB, in this post.). However, the vast majority of the time, it runs at 1000MHz@1.1V, because contrary to the gentoo jokes, that doesn't all that much CPU time, and it more or less just idles.
I've got a 3000+ Winchester (1.8GHz, nominal) which goes up to 2.8GHz (and a
This is achieved simply by messing with the FSB, and having a motherboard that allows other modifications (A8N SLI)
The biggest problem with overclocking, is that I have to disable AMD's cool and quiet. (The chip won't run at 1.5GHz 300x5 at 1.1V)
Now, if you take into consideration that Pretty much any chip above a P3-733 uses more power (by spec) than an Athlon 64 at idle (1GHz), it translates into savings of energy, if it's left on for any significant period of time. (And even if you locked the clock on slow, it'd still perform better.)
Unlike AMD, Intel is in a pretty bad situation heat/power wise.
And FDR didn't get his way.
And therein lies the problem for your whole idea.
"whatever legal entity owns the Linux kernel in the copyright sense"
There is no one, in fact, you've got hundreds, if not thousands of different people's code, all owned by them. Most simply like working on it. Some have rather 'Free or die' mentalities. Some (IBM) have cash, and no reason to want it to be closed.
So the Linux Kernel is a very very bad example.
Now, take for example a 1 person OS project. If it were bought, the person would likely be hired to work on it. The project would likely get more time devoted to it. As long as it remained open, there is no loss. (Of course, that's a big issue to many OS developers I know, they wouldn't mind being paid to do what they like to do for fun. However, they don't want it closed.)
Please, explain to me:
a) Why you want a GUI on the server itself.
b) Why you cannot use remote access without a GUI. (ssh)
c) Why if you really need graphical applications running on the server you cannot use remote access. (ssh+X forwarding, vnc, rdesktop, etc)
Having a display on a server shows a lack of ability to administer said server remotely, which any competent admin should be perfectly capable of doing, barring hardware failure.
It's addressed via PXE/"Highmem" extensions, allowing the p6 core to use the 36-bit real memory bus to remap that into the 32-bit virtual address. It requires any app using more than 4 gig to do some "weird shit" with memory requests, as well as in general only having 3-3.5Gig available at any one time (OS virtual overhead), with parts of the program duplicated a number of times. (Or at least that's how it was, Hopefully it's improved since then)
And we aren't even considering how horribly that would perform on anything more than a dual. Shared bus = easy 'multi-core' processors = Shit for scaling to more processors than 2. Any server chip is going to have better than that, hell, the original athlons did (because they used an alpha ev6 bus)
One of the few advantages Sparcs have over Xeon, has evaporated as Intel follows AMD's x86-64 spec, and replaces the memory with 48-bit virtual (Opterons/A64s only do 40-bit physical. Oh dear, by the time I've got that much ram, I'll have a better processor!)
I just double checked on Intel's website, and the best I could find was 8x/8x (3 x8 and 1 x4 PCI express slots (28 lanes total)) And with that it is not possible to have multiple x16 slots (Heck, it's impossible to have 1) (It's possible I missed a better one. I was looking in the server section.)
The main reason that Tyan can do that is because of AMD's superior Hypertransport-based bus design in Opterons, over the shared bus favored by Intel. It's also the reason why Opteron scales a lot better than Xeon.
The other reason Tyan can do that is that Nvidia realized how easy it would be to make very slightly different chipsets that facilitated that. Basically they are just Nforce 4 chipsets, that can operate in parallel, giving 40 Pci express lanes (2-way) or 80 PCI express lanes for a 4-way Opteron. (Note a maximum of 4 x16s, as the other 16 can only be a max of x4, due to the 20 lanes per nforce4)
You can't do x16/x16 with any Intel Processor, as of now. (Though having seen how little x16/x4 or x16/x2 hurts benchmarks (vs standard x8/x8) I'm not convinced it's a big deal at all.)
Actually no, it's only caught up to Mac OS X (and actually exceeded, due to the way 64-bitness is in OS X so far.).
;)
This is simply the OS that's running in 64-bit mode now. The programs are all still 32-bit. (Admittedly, this means a lot more for Windows vs Mac OS, as the underlying command set changes just a bit ia32->x86_64.)
Now, if they could catch up to Redhat 6.0 on alpha, that'd be impressive.
(Note, that while NT was on Alpha, it was treated as a 32-bit arch, and the first fully 64-bit port of Windows was to ia64...the 2nd 64-bit arch it was on... All the sudden I get this feeling of impending doom for x86_64.)
Kcontrol -> Desktop -> Window Behavior -> Translucency Tab. That controls transparency, and the Focus tab has Focus follows Mouse.
There is one, I can't remember the name for sure (Kugar?) The report or database app.
However, it comes up with a wizard to configure it as well.
I've used KDE since 98... and that's the only app I've noticed that does anything like that in that time, and that's Koffice, not KDE proper.
That's interesting... because SB Live! cards do NOT do hardware mp3 decoding. Any of them. The Soundblaster Live! and the variants MP3+ & X-gamer are the exact same card... with different software bundles. At least the original ones. Later cards have some differences, but the only thing the MP3 & gamer ones had was different software bundles. If you believe(d) otherwise, and paid money for it... you've been had. (I'm unable to find a particular email describing the differences between the cards, from people actually writing drivers for them, at the moment.)
NASA's Columbia cluster ^ 512-way SGI machines running Linux (actually 20 of them...) Not to mention "Columbia's record results were achieved running the LINPACK benchmark on 8,192 of the NASA supercomputer's 10,240 processors. Columbia also achieved an 88 percent efficiency rating on the LINPACK benchmark, the highest efficiency rating ever attained in a LINPACK test on large systems." from http://www.sgi.com/company_info/newsroom/press_rel eases/2004/october/worlds_fastest.html
... or do it the way konqueror's "Smart" mode does: Only allow popups in response to mouse clicks. Durring the time I have used this. (Quite a while, I'd have to check when it got introduced.) I can remember two sites who got around it (one being a BTVS guide site), and I've only got 2 sites on what is essentially an override list. (ie: they wouldn't work, AND I did want them to pop up things.) One of those being dlink.com Considering that this is AT LEAST a year, I'd say it's pretty effective. (Sorry to shoot down you example. ;) )
The only way that worked, was when SGI owned Cray for a while, Cray was making Alpha systems. More commonly people confuse it, because DEC did use mips chips. SGI's Altrix is the only system that actually gives the Itanium enough memory bandwidth to work in more than 2-way multiprocessor. Intel STILL has a shared bus processor design. SGI's chipset + other hardware, makes it a NUMA design with 2 processors accessing each chip. (Ie: every processor has a twin which shares northbridge access, but that northbridge allows other chips to access the memory.) That might remind you of other 64-bit chips. One that's pretty cheap called an Opteron. (Heck once it gets dual core, it'll be even closer. Provided you just want 16 cores... or more, if you want to design a whole new chipset, like SGI did for Itanium ;) )
It does. (See my sibling post.)
I see no problems in a week old cvs build. Dispite trying both.
...checked, and it doesn't work. However, I figured out both why it didn't work, and a temporary fix.
I'll check the refresh issue above.
"Smart" Javascript (Config Konqueror->Java & Javascript->Javascript tab-> Open window policy: Smart) blocks it.
The reason (and why Smart really is smart) is it only allows new windows to be opened in response to a mouse click. (no OnLoad events or in this case I believe a refresh timer.) Thus the pop-up is NOT allowed to open. This also lets the vast majority of legitimate sites work properly. (I've only got 2 listed that it doesn't work with...)
Konqueror's SMART setting prevents it. Allow doesn't, Ask might (depends on user...). Deny also prevents it.
View -> Icon Size
What advantages does this have over 'hardened' Linux?
Also: http://kerneltrap.org/node/view/3823
(I don't know which subset of BSD's jail it doesn't implement. Anyone care to share who knows both?)
I've got it as a backup/mobile connection.
This is similar to my experences. Generally latency is 500ms, Generally ranging 400-600ms, the highest I've personally seen is 1100ms.
(Compared to any other type of home connection, that's horrible.) Also, TRY to have 4 bars, otherwise the latency goes way up.
Sanyo 8100, Sanyo 4900, and Sanyo VM5040 (Whatever the one that can take video as well is) all work. Same instructions. (#777 etc) (And have the same cable interface)
From looking at usenet it appears, or at least the general belief seems to be: There's a limit at which they flag you for watching. I've heard 150, 200, 250MB/month, after which they send a letter telling you to cut it out or buy a plan, otherwise, they will cut off your "unlimited" vision service (Which means some relatively huge rate per kb).
Though, It's been said that sprint really doesn't care too much, provided you aren't using it constantly or downloading a lot. (Ie using maximium bandwidth, because that does effectively cut out multiple voice circuits I forget exactly what, but voice is significantly less than 144k claimed by sprint. I seem to recall 8, 16, and 24 but it's several voice lines. If it's done at night, they don't seem to care at all, durring the day on the other hand...)
All rumors, and such. I know I've used more than 100MB/month before, and had no trouble at all. (Due to that, due to other things... yes)
That would be nice, but 10k rpm drives have three disadvantages.
#1 They consume much more power.
#2 They make more heat.
#3 The majority are quite loud. (Doesn't mean all) But unless you buffer the sound a lot, which insulates the drive more, compounding #2)
On a friend's brand new dell (1.5GHz P-M, Radeon 9600, the thing that gets hot to the touch is not the chip or radeon (when playing something 3d), it's the hard drive.
Another thing is modern moble processors will clock down that far already, (assuming they have absolutely equal performance and power consumption at ths same MHz, and I think Pentium Ms and Athlon mobiles both have an advantage there.) thus there is little to no gain from that. Low end power consumption, it would be the same, but there would be no way to get more performance, as P-Ms and A-Ms do by clocking up. (Though if the chip were designed specifically for this, you could use one of their cores, and cut the pipeline back down.)
So, yeah kinda. What will work however, is having a huge amount of RAM, so that things wanted on the hard drive are already in memory. This would likely resolve the issue with my friend's laptop and allow the drive to spin down, consuming less power, making less heat AND actually making the programs run faster.
You don't WANT to run Itanium 8-way. I'd be willing to bet an 8-way Operton against a straight SMP Itanium on a well threaded application. I believe that the Itanium's SHARED bus to memory is 6.4GB/sec (It was at one time, I'm not sure if that's been upped. I'm not sure, but that may also include talking to everything else. (I thought it had a second bus for that. I hope they do.)) This means at maximum theoretical performance each Itanium will have a maximum of .8GB/sec memory bandwidth. Vs a Opteron 8xx, with something like the 6.4 GB/sec memory bandwidth per processor. Occasionally, something won't be in an Opteron's local memory, and it will be slower, but HT is still 4.8GB/sec. Itanium is fast per clock yes, and single processor wise would probably compare favorably against a Operon 8xx. However, the Opeteron's use of NUMA means that As it scales, it's going to begin, on any threaded application, to completely thrash the Itanium's pure SMP w/shared bus.
Which is why no one, other than *possibly* HP (I don't know exactly how their Integrity servers are layed out), does large pure-SMP Itaniums. SGI essentially attempts to emulate the Opteron's design with their Itanium design, which only does 2 way smp, with lots of interconnects. The interconnects are actually faster than the Opterons, but are also a lot more expensive.
Price/Performance unless on perhaps one or more specific apps that are highly tuned, will see the Opteron wallop the Itanium.
Languages evolve. Deal with it. What are you some sort of grammer exceptionist? "Its" does't follow standard grammatical rules for contractions. It's evovling towards actually fitting general rules in this case. Isn't it better that way? There is however, another possibility: perhaps you just like being an ass with your superior knowledge of useless things? IMNSHO It's better. Quit with the knee-jerk reactions. You understood what was meant, and is that not the general purpose of the english language?