One positive effect from this is that the writeable top CD disks will have to be free from manufacturer logos either etched or printed. I'd almost pay 10 cents more per disk right now for if I could find really "blank" disks with no marks on them. Verbatim used to have some wonderful pure white surfaces with NOTHING on them, but I can't seem to find these any more... I'm tired of advertiseing for a disc maker every time a give a CD away...
...any more than IBM would ditch Power4/5 architecture, just because they have a commodity market x86 chip with 64-bit address extensions (Opteron).
In the 'big iron' enterprise market against RISC where Itanium is beating everything handily (check out the latest TPC-C list Top 10 where Itanium holds spots 1,3,4,7,10 (5 out of the Top 10 are Itanium systems running a mix of Linux, HP-UX and Windows on HP and NEC systems), Itanium is gradually out-selling all of the big RISC opponents like Power4. Note that IBM is certainly not spending the money to put up an Opteron cluster for the TPC-C test(no 32-way or 64-way scaled solutions for it on the horizon) even if they got good enough results (which they wouldn't) if they can't beat Itanium 2 right now with the high-margin Power 4. No doubt they'll have a run at Itanium again this year with Power 5.
But there's no way that Opteron OR a 64-bit Xeon plays in the big high thoughput space, so people that assume Intel would get rid of Itanium simply don't know what they're talking about.
As for Itanium not selling, That's funny. Itanium sold over 100,000 cpus last year which is a big number for the enterprise server market (That's more than some other major RISC processors sold in 2003 (like Power 4)). If you don't believe me Google "Itanium" "100,000" and "Otellini" and you'll see lots of links to Intel pres Paul Otellini's announcement back in Nov that Intel would ship over 100,000 Itanium processors in 2003.
Barrett is quoted saying, 'There will be one operating system that will support all (64-bit) extended systems.' WRONG. Barrett did not say this, a Microsoftie did. (Makes sense since Intel doesn't make OSes)
The article says:
Intel's approach is compatible with AMD's, the Microsoft representative said. "There will be one operating system that will support all (64-bit) extended systems," the representative said.
Note that this refers to "(64-bit) extended systems" -the pure, native (non-extended) Itanium 64-bit version of Windows Server 2003 which is already shipping is, a will always be different from the x86-extended code base since it takes advantage of the EPIC instruction set.
Sorry I don't want to delete the post. Despite your alleged implications the situation, surely you're not denying that had the problem occured on a Windows platform many slashdotters would make reflex comments about the platform whether the problem lay with the coders or not?
I certainly don't concede your suggestion that RISC/UNIX systems are so stable that it would be clear that it was bad coding. There's lot's of examples of flaws in various *NIX OSs and physical issues on RISC servers (ask current Sun users about system quality and reliability on the low-end V series boxes)
BTW, I spend about 50% of my time with Windows environments and 50% with *UNIX/Linux environments. My single most consistant observation is that quality of end-user management disipline by managers and adminstrators is the most likely source of system failures and variance here is far more important than the particular platform.
In this vein, the closest I'll come to your position is that Microsoft made it easy (via quick GUIs and reams of pre-baked defaults) for poorly trained people to poorly deploy Windows systems - that do run. *nix/Linux systems are just hard enough to get installed that if you really don't know what you're doing you can't put them into production at all. Economics ensures that some fraction of the business community will go to the lowest cost option that seems to work (badly or not)-which is why there's so many poor Windows installations out there. Because there CAN be.
I can't believe how many people jump on anything to do with Intel or AMD and turn it into flame wars about cpus. This story is about optical comm gear breakthoughs and has NOTHING to do with processors.
Google "Itanium" "100,000" and "Otellini" And you'll see dozen's of links to Intel pres Paul Otellini's announcement back in Nov 2003 that Intel would ship over 100,000 Itanium processors in 2003.
It would be pretty hard to lie about that, Pan, esp if your assumption that they shipped less than they did in 2002 was true.
"What really signals that Itanium is doomed is the fact that no one is buying it."
That's funny. Itanium sold over 100,000 cpus last year which is a big number for the enterprise server market. That's more than some other major RISC processors sold in 2003 (like Power 4). Just because it's too expensive for you and your LAN-party gamer buddies to buy on your allowance, doesn't mean serious businesses doing big serious tasks wouldn't we willing to spend a lot of money (but less than they do on IBM and Sun and the other big RISC platforms...) for Itanium
...any more than IBM would ditch Power4/5 architecture, just because they have a commodity market x86 chip with 64-bit address extensions (Opteron).
In the 'big iron' enterprise market against RISC where Itanium is beating everything handily (check out the latest TPC-C list Top 10 where Itanium holds spots 1,2,3,4 (6 out of the Top 10 are Itanium systems running a mix of Linux, HP-UX and Windows on HP and NEC systems), Itanium is gradually out-selling all of the big RISC opponents like Power4. Note that IBM is certainly not spending the money to put up an Opteron cluster (no 32-way or 64-way scaled solutions for it on the horizon) even if they got good enough results (which they wouldn't) if they can't beat Itanium 2 right now with the high-margin Power 4. No doubt they'll have a run at Itanium again this year with Power 5. But there's no way that Opteron OR a 64-bit Xeon plays in the big high thoughput space, so people that assume Intel would get rid of Itanium simply don't know what they're talking about.
Ah yes, when will someone release a simple distribution that uses non-GNU components so we can tell RMS: SEE! There CAN be Linux without GNU, so calling it all GNU/Linux is wrong!
Howver, HT as Intel has implemented it, will reduce overall latency running more than one single threaded app, not just multi-threaded apps. Who has a system with ONLY ONE THREAD executing?
...is that the used-car salesman knows when he's lying.
There's a really interesting philosopical point here, BTW. If you are chartered to (or are pretending to know) something that you don't really understand, can you really claim that you didn't lie (because you didn't realize what you said was false) or do you have a responsibility to be correct if you offer yourself as an authority on a subject?
You SAY you're against compulsion, but you're not
on
What You Can't Say
·
· Score: 1
...since you implicitly want to do a search and seizure on any encypted packets crossing your network which may be hiding illicit VPN tunnels. As soon as YOU are part of the control stucture, your whims and foilbles, like that of persons in control in gov't become mandates for the rest of us.
Leave our packets alone and we'll think about getting gov't off your back.
(BTW, you're right about QoS - I think forcing you to treat some packets as special is unreasonable too)
Sure takes the edge of 64-bitness when big memory addressablilty is stifled by a platform that only takes a MAX of 1Gb of RAM. No, Virginia, 64-bit systems aren't faster unless they can use big memory (more than 4GB).
In terms of Opteron not hitting the high end of the market, that was simply never their intended market.
But, unlike Xeon and Opteron, Itanium WAS intended to hit this market and as long as there's other less attractive cpus being sold out there like Power and SPARC and as long as commodity processors like Xeon/Operon can't fill that role, Itanium has a market. Either you ARE arguing that Xeon with 64 bit addressing OR Opteron will make Itanium unneeded or you're not. Please decide.
But I wouldn't assume Intel doesn't think there's a role for both architectures.
The original question was: would there be a role for Itanium in a world where Intel offered extended addressing for Xeon? My point was not about Itanium's price/perf but it's pure performance. If Opterons were faster in absolute terms then they would be in the Top Ten perf list not just the Price/Perf list. Instead, unlike the million+ tpm the best Itanium system supported, the best Opteron result was a piddly 20k (Because Racksaver is only showing results from ONE cpu. -if it scaled, maybe showing MORE than one cpu would make sense...) Anyway, if your solution really needs the high thoughput, it doesn't matter how cheap it is if you can't do the job. Price/perf only applies when you meet job criteria.
Your point about the Itanium system not existing, (like your 50-Opteron-box) is interesting. The Superdome certainly exists, (if not, what did they run the test on?) it's just not shipping yet - as opposed to the totally hypothetical Opteron system that exists only in your head. No one (that I know of) has built a large scale SMP or NUMA (non-clustered) architecture with Opterons (Hypertransport makes small sub-8-way SMP and clustering much more attractive) so they are not playing in this space yet. I'm skeptical about whether this would be worth the effort for Opterons -unlike Itaniums which were designed to be very 32-way+ scalable (explict parallelism, very large caches etc.) As for the performance of your multi-way Opteron system, you may want to read some CompSci texts on how SMP scaling works (hint: simply mulitplying 1 proc perf by number of procs is NOT the way it works). And one would think that the big system vendors who are adopting Opteron, IBM & Sun are not going to vandalize their own lucrative large-scale Power & SPARC markets by building inexpensive scaled Opteron solutions. If they don't, who will? Racksaver? Again, people will try to solve some of these problems with clusters, but there really are applications for which a large single image partition is what's really needed.
As long as some users need higher perf, bigger high-end iron, Itanium will do better. And as long as there is an installed base of IBM Power, Sun SPARC and other underperforming, overpriced (YES, relative to Itanium) RISC systems out there, there are customers with applications that need that large SMP perf and Itanium has a role that Xeon and Opteron aren't filling - even with or without memory extensions.
[IRONY]If you choose to live in a country where nine grand a year doesn't go very far, that's your problem, right? It's not that US techs are paid too much, it's that the cost-of-living in US (and Europe too, for that mattter) is innordinately high by world standards. Until this problem is rectified, US workers can use stopgaps like living with your parents and eating at the homeless shelter and shopping at the foodbank occasionally.[/IRONY]
The online trade press is (deliberately?) ignoring this , but on app benchmarks (not just HPC) Itanium is kicking butt.
Check out the Top Ten Performance TPC-C benchmarks for on OLTP and 5 of the top 10 systems are Itanium and ALL of the the top 3. An HP Superdome with 64 Itaniums running Oracle 10g was the first ever system to do OVER 1 MILLION transactions per minute.
NO ARCHITECTURE, Opteron, Xeon, Power, (Certainly not SPARC or MIPS) can touch that right now.
Slashdot Mirror
One positive effect from this is that the writeable top CD disks will have to be free from manufacturer logos either etched or printed. I'd almost pay 10 cents more per disk right now for if I could find really "blank" disks with no marks on them. Verbatim used to have some wonderful pure white surfaces with NOTHING on them, but I can't seem to find these any more...
I'm tired of advertiseing for a disc maker every time a give a CD away...
Stay tuned for news at 7...
...any more than IBM would ditch Power4/5 architecture, just because they have a commodity market x86 chip with 64-bit address extensions (Opteron).
In the 'big iron' enterprise market against RISC where Itanium is beating everything handily (check out the latest TPC-C list Top 10 where Itanium holds spots 1,3,4,7,10 (5 out of the Top 10 are Itanium systems running a mix of Linux, HP-UX and Windows on HP and NEC systems), Itanium is gradually out-selling all of the big RISC opponents like Power4. Note that IBM is certainly not spending the money to put up an Opteron cluster for the TPC-C test(no 32-way or 64-way scaled solutions for it on the horizon) even if they got good enough results (which they wouldn't) if they can't beat Itanium 2 right now with the high-margin Power 4. No doubt they'll have a run at Itanium again this year with Power 5.
But there's no way that Opteron OR a 64-bit Xeon plays in the big high thoughput space, so people that assume Intel would get rid of Itanium simply don't know what they're talking about.
As for Itanium not selling, That's funny. Itanium sold over 100,000 cpus last year which is a big number for the enterprise server market (That's more than some other major RISC processors sold in 2003 (like Power 4)). If you don't believe me Google "Itanium" "100,000" and "Otellini" and you'll see lots of links to Intel pres Paul Otellini's announcement back in Nov that Intel would ship over 100,000 Itanium processors in 2003.
The post says:
Barrett is quoted saying, 'There will be one operating system that will support all (64-bit) extended systems.'
WRONG. Barrett did not say this, a Microsoftie did. (Makes sense since Intel doesn't make OSes)
The article says:
Intel's approach is compatible with AMD's, the Microsoft representative said. "There will be one operating system that will support all (64-bit) extended systems," the representative said.
Note that this refers to "(64-bit) extended systems" -the pure, native (non-extended) Itanium 64-bit version of Windows Server 2003 which is already shipping is, a will always be different from the x86-extended code base since it takes advantage of the EPIC instruction set.
They bought it from Appeal
Sorry I don't want to delete the post. Despite your alleged implications the situation, surely you're not denying that had the problem occured on a Windows platform many slashdotters would make reflex comments about the platform whether the problem lay with the coders or not?
I certainly don't concede your suggestion that RISC/UNIX systems are so stable that it would be clear that it was bad coding. There's lot's of examples of flaws in various *NIX OSs and physical issues on RISC servers (ask current Sun users about system quality and reliability on the low-end V series boxes)
BTW, I spend about 50% of my time with Windows environments and 50% with *UNIX/Linux environments. My single most consistant observation is that quality of end-user management disipline by managers and adminstrators is the most likely source of system failures and variance here is far more important than the particular platform.
In this vein, the closest I'll come to your position is that Microsoft made it easy (via quick GUIs and reams of pre-baked defaults) for poorly trained people to poorly deploy Windows systems - that do run. *nix/Linux systems are just hard enough to get installed that if you really don't know what you're doing you can't put them into production at all. Economics ensures that some fraction of the business community will go to the lowest cost option that seems to work (badly or not)-which is why there's so many poor Windows installations out there. Because there CAN be.
...to say everybody must migrate RIGHT NOW to Server 2003 and XP. ;-)
Seriously, esp once the security patches hit XP with SP2 this summer, expect Ms to start suggesting that anything older is not to be trusted...
Had this been a Windows-based system, the torrent of comments about how unreliable the OS and platform fundementally was would be huge.
Funny, just because this ships for "industrial strength" AIX / Solaris RISC systems (see specs on pg 8), I don't see any cheap, reflexive comments about the platform.
I guess the message here is that good or bad code can be written for any architecture.
I can't believe how many people jump on anything to do with Intel or AMD and turn it into flame wars about cpus. This story is about optical comm gear breakthoughs and has NOTHING to do with processors.
Proof:
Google "Itanium" "100,000" and "Otellini"
And you'll see dozen's of links to Intel pres Paul Otellini's announcement back in Nov 2003 that Intel would ship over 100,000 Itanium processors in 2003.
It would be pretty hard to lie about that, Pan, esp if your assumption that they shipped less than they did in 2002 was true.
Sorry to burst your bubble.
"What really signals that Itanium is doomed is the fact that no one is buying it."
That's funny. Itanium sold over 100,000 cpus last year which is a big number for the enterprise server market. That's more than some other major RISC processors sold in 2003 (like Power 4).
Just because it's too expensive for you and your LAN-party gamer buddies to buy on your allowance, doesn't mean serious businesses doing big serious tasks wouldn't we willing to spend a lot of money (but less than they do on IBM and Sun and the other big RISC platforms...) for Itanium
...any more than IBM would ditch Power4/5 architecture, just because they have a commodity market x86 chip with 64-bit address extensions (Opteron).
In the 'big iron' enterprise market against RISC where Itanium is beating everything handily (check out the latest TPC-C list Top 10 where Itanium holds spots 1,2,3,4 (6 out of the Top 10 are Itanium systems running a mix of Linux, HP-UX and Windows on HP and NEC systems), Itanium is gradually out-selling all of the big RISC opponents like Power4. Note that IBM is certainly not spending the money to put up an Opteron cluster (no 32-way or 64-way scaled solutions for it on the horizon) even if they got good enough results (which they wouldn't) if they can't beat Itanium 2 right now with the high-margin Power 4. No doubt they'll have a run at Itanium again this year with Power 5.
But there's no way that Opteron OR a 64-bit Xeon plays in the big high thoughput space, so people that assume Intel would get rid of Itanium simply don't know what they're talking about.
Ah yes, when will someone release a simple distribution that uses non-GNU components so we can tell RMS:
SEE! There CAN be Linux without GNU, so calling it all GNU/Linux is wrong!
Howver, HT as Intel has implemented it, will reduce overall latency running more than one single threaded app, not just multi-threaded apps. Who has a system with ONLY ONE THREAD executing?
...is that the used-car salesman knows when he's lying.
There's a really interesting philosopical point here, BTW. If you are chartered to (or are pretending to know) something that you don't really understand, can you really claim that you didn't lie (because you didn't realize what you said was false) or do you have a responsibility to be correct if you offer yourself as an authority on a subject?
...since you implicitly want to do a search and seizure on any encypted packets crossing your network which may be hiding illicit VPN tunnels. As soon as YOU are part of the control stucture, your whims and foilbles, like that of persons in control in gov't become mandates for the rest of us.
Leave our packets alone and we'll think about getting gov't off your back.
(BTW, you're right about QoS - I think forcing you to treat some packets as special is unreasonable too)
Sure takes the edge of 64-bitness when big memory addressablilty is stifled by a platform that only takes a MAX of 1Gb of RAM.
No, Virginia, 64-bit systems aren't faster unless they can use big memory (more than 4GB).
Thanks for conceding my original point:
In terms of Opteron not hitting the high end of the market, that was simply never their intended market.
But, unlike Xeon and Opteron, Itanium WAS intended to hit this market and as long as there's other less attractive cpus being sold out there like Power and SPARC and as long as commodity processors like Xeon/Operon can't fill that role, Itanium has a market. Either you ARE arguing that Xeon with 64 bit addressing OR Opteron will make Itanium unneeded or you're not. Please decide.
But I wouldn't assume Intel doesn't think there's a role for both architectures.
Let's give some credit to the underlying architecture.
Bravo
As the article says about lawyers, it 'turns out you can't leave Earth without them.'
This would be worth persuing if we can simply send ALL the lawyers to the moon...
The original question was: would there be a role for Itanium in a world where Intel offered extended addressing for Xeon? My point was not about Itanium's price/perf but it's pure performance. If Opterons were faster in absolute terms then they would be in the Top Ten perf list not just the Price/Perf list. Instead, unlike the million+ tpm the best Itanium system supported, the best Opteron result was a piddly 20k (Because Racksaver is only showing results from ONE cpu. -if it scaled, maybe showing MORE than one cpu would make sense...) Anyway, if your solution really needs the high thoughput, it doesn't matter how cheap it is if you can't do the job. Price/perf only applies when you meet job criteria.
Your point about the Itanium system not existing, (like your 50-Opteron-box) is interesting. The Superdome certainly exists, (if not, what did they run the test on?) it's just not shipping yet - as opposed to the totally hypothetical Opteron system that exists only in your head. No one (that I know of) has built a large scale SMP or NUMA (non-clustered) architecture with Opterons (Hypertransport makes small sub-8-way SMP and clustering much more attractive) so they are not playing in this space yet. I'm skeptical about whether this would be worth the effort for Opterons -unlike Itaniums which were designed to be very 32-way+ scalable (explict parallelism, very large caches etc.) As for the performance of your multi-way Opteron system, you may want to read some CompSci texts on how SMP scaling works (hint: simply mulitplying 1 proc perf by number of procs is NOT the way it works). And one would think that the big system vendors who are adopting Opteron, IBM & Sun are not going to vandalize their own lucrative large-scale Power & SPARC markets by building inexpensive scaled Opteron solutions. If they don't, who will? Racksaver? Again, people will try to solve some of these problems with clusters, but there really are applications for which a large single image partition is what's really needed.
As long as some users need higher perf, bigger high-end iron, Itanium will do better. And as long as there is an installed base of IBM Power, Sun SPARC and other underperforming, overpriced (YES, relative to Itanium) RISC systems out there, there are customers with applications that need that large SMP perf and Itanium has a role that Xeon and Opteron aren't filling - even with or without memory extensions.
"World Gains 500,000 jobs from US"
[IRONY]If you choose to live in a country where nine grand a year doesn't go very far, that's your problem, right? It's not that US techs are paid too much, it's that the cost-of-living in US (and Europe too, for that mattter) is innordinately high by world standards. Until this problem is rectified, US workers can use stopgaps like living with your parents and eating at the homeless shelter and shopping at the foodbank occasionally.[/IRONY]
The online trade press is (deliberately?) ignoring this , but on app benchmarks (not just HPC) Itanium is kicking butt.
Check out the Top Ten Performance TPC-C benchmarks for on OLTP and 5 of the top 10 systems are Itanium and ALL of the the top 3. An HP Superdome with 64 Itaniums running Oracle 10g was the first ever system to do OVER 1 MILLION transactions per minute.
NO ARCHITECTURE, Opteron, Xeon, Power, (Certainly not SPARC or MIPS) can touch that right now.
I suspect Sun will work very hard to keep people from comparing SPARC hardware with Opteron units, since the RISC systems will get killed.
But Sun customers should insist on seeing this comparison.
http://www.intel.com/research/spotlights/terahertz bkgdr.htm
Not really clear if it has actually been RUN at a Terahertz, but it's implied to he capable technology