Uhh, try using the CURRENT tarriff, and not the one that expired three years ago! Here is the current tarriff for blank media. Note that even though it only starts out as a proposal, these are the real tarriffs being collected, as per this decision.
Actual tarriff on blank data CDs in Canada is $0.21.
The current tarriff is only $0.21 per CD. The proposed increases to this tarriff would see that number rise to $0.59 per CD. Under the current rules, roughly half of the money you pay to buy a data CD for that latest Linux ISO is going to the recording artists of Canada because you making a data CD is hurting their livelyhood so badly. In the updated version, 3/4 of that Linux ISO will go to help our starving record execs.
Ohh, and as for who gets the money for your Linux ISOs...
"Since no inventory of privately copied tracks exists, distribution is based on representative samples of radio airplay and album sales, which are given equal weight in the distribution."
Better hope that the artists you like get lots of radio play, but chances are that most of our money is going to Celine Dion and Shania Twain.
He might not be able toback it up, but I certainly can. Here are current tarrifs and here are the proposed tarrifs. Note that the first set of proposed increases to tarriffs were somewhat shot down in favor of the ones listed above.
For those of you who want to go straight to the original, here is the official proposal. Note that in this proposal, they wanted to charge a full $21/GB for hard-drives in MP3 players. This actually would have raised the price of an iPod or similar such device (20GB hard drive) by a full $410. Given that the current going price for a 20GB iPod is roughly $500 to $600 CDN, this would have come close to doubling the price.
In the updated levy, the charge would be ~$110 for a 20GB MP3 player.
Actually that brings MS and Apple even for the past month at 1 a piece (Microsoft had a buffer overrun in the Workstation service).
Ohh, and both MS and Apple have had a security vulnerbility for their browser this month on top of the OS vulnerbilities listed above.
Linux doesn't seem to have had any new security vulnerbilities announced this month, though a few security fixes are filtering through for vulnerbilities announced in October. Both WinXP and OS X also had some similar fixes for earlier bugs.
Long story short, it doesn't matter what OS you run, you WILL have security vulnerbilities. Patch your OS and use a firewall already!
Umm.. Microsoft does not currently offer any anti-spam products.
On the other hand, they pay probably hundreds of millions of dollars to deal with the spam problem. Hotmail.com receives at least a billion spam messages a day, and probably closer to 2 billion. Do you know how much money it costs Microsoft to pay for that bandwidth? How many extra servers they have to pay for? Not to mention how much they're already paying third-parties anti-spam software companies?
And that isn't even starting to touch the huge amount of lost productivity at Microsoft caused by their employees receiving boatloads of spam (think: computer industry company with tends of thousands of employees, you better believe that MS recieves a lot of spam).
There is no way that MS is ever going to cover the expense of their receiving spam by any products. Trust me, MS wants spam to stop just as much as any other major corporation that has to deal with the problem.
How, pray tell, would that affect a spammer?! You don't honestly think that a spammer uses an ISPs legitmate e-mailer servers do you? Let me guess, you also think that those spams with "hotmail.com" from addresses were actually sent from Hotmail too?
Face it, spammers are criminals, and expecting simple laws to stop them is hugely naive when the multitude of existing laws that spammers break continue to go unenforced.
His name was Gordon Moore, and he worked for Intel... err, more to the point, he founded Intel along with Andy Grove and Robert Noyce. Of course, like pretty much all the important players in the semiconductor industry, he was working at Fairchild at the time that he released his paper. What he said was:
"The complexity for minimum component costs has increased at a rate of roughly a factor of two per year"
He predicted that this would hold true for 10 years, from 1965 to 1975. Of course, this prediction fairly quickly changed to doubling the transistor count every 18 months, since the 12 month prediction didn't hold true for very long. Now we're doubling transistor count roughly every 24 months.
FWIW anyone who's interested can take a quick read of Moore's whitepaper where this "law" came from.
Err, no. AMD just announced a new fab in Dresden that will produce chips at a 65nm fab process, just as they have been saying all along.
Of course, AMD's fab won't be ready until 2006, so Intel will have a lead, but that's nothing new. Intel is pretty much always 6 months ahead of AMD (and everyone else for that matter) at bringing out a new manufacturing process. Being the biggest semiconductor manufacturer with the largest budget and most buying power has it's advantages you know.
There are a number of things going on here, but a few important things to think of.
First off, with previous shrinking of the manufacturing process you could run the processor at a lower voltage. Most 500nm chips ran at 3.3V, 350nm chips ran at 2.8V, 250nm chips ran at 2.0V, 180nm chips ran at 1.75V and 130nm chips now run mostly at 1.55V. As you can see pretty quickly though, the difference in voltage isn't as much as it used to be, and with 90nm production, that difference is pretty much zero, most 90nm chips will probably run at about the same 1.55V of current (130nm) chips.
Second, the Prescott does add a bit more than just cache. Alongside the new SSE3 instructions, Intel is also making some fairly major changes to the P4 core, fixing some of the potential problem areas. I haven't heard 100% official confirmation, but apparently it adds a barrel shifter (should be very noticeable for D.net clients if they do) and fixes some of the scheduling issues with multiplications. These changes are going to result in extra transistors, and extra transistors means more power.
Also, the Prescott is supposed to improve hyperthreading. This is a good thing from a performance standpoint, it means that you'll get more of a performance boost from running two threads at once. The downside is that it means that the processor pipeline will be packed more fully, again increasing power consumption.
In short, there's lots to consider, no easy equation to get you the power consumption of the new chip.
Originally Moore's law stated that transistor density would double every 12 months. That was fairly quickly changed to say that it would double every 18 months. The current "law" states that transistor density doubles about every 24 months.
Long story short, we haven't really been following Moore's law for a little while, though we do continue to double the amount of bits we can stuff onto a piece of silicon at a fairly rapid pace. Intel's plan to bring out 65nm chips before the end of 2005 continues this trend.
FWIW IBM is also looking to bring out 65nm chips by late 2005/early 2006, while AMD is hoping to get their 65nm fab process up and running in their new fab early in 2006. TSMC and UMC are likely to follow in mid-2006, though I haven't heard any official comments from either.
FWIW this guy also got a brief mention in this article.
Not only did he make death threats against the spammers, apparently he had been warned to stop by local authorities, but continued.
It's one thing to be pissed off at someone because they're annoying, but to continue making death threats against that person even after being told to stop by police is quite another thing.
Meant to provide one with my original post, but got distracted. Here it's mentioned in a CNet story about it. I think AMD actually stated the feature size would be 65nm in their conference call, though I didn't listen in.
Shouldn't really be a big surprise though, 2003 was when everyone was supposed to be switching to 90nm production (though it looks like it's actually going to happen mid to late 2004), and normally new feature sizes come out about every two to three years. So, with the new fab opening up in 2006, a 65nm feature size is right on track with expectations. Actually I think Intel is hoping to get to 65nm before the end of 2005, but then again, they also wanted to start shipping chips with 90nm features months ago.
As others have mentioned, Intel just has their chips packaged in Malaysia, the Philippines, etc. Similarly AMD has almost all of their chips packaged in Malaysia. FWIW most of Intel's actually chip fabs are in the US. Oregon to be exact. They also have a couple plants in Ireland and in Isreal. To the best of my knowledge they do not have any plants in south-east Asia.
In any case, here's a few numbers for you.. A typical plant these days costs about $2.5 billion dollars to build. Equipment in that plant is about another billion dollars and has a useful life of about 5 years if you can stretch it. The plant as a whole has a useful life of about 10 years before it needs a major overhaul.
So, simple bit of math tells us that you're looking at a fixed capital cost of somewhere around $4.5 billion. This gives you a rate of depresiation of roughly $1.2 million dollars a day.
Now, if you pay an average of $100,000/year to the ~1000 employees of the plant, you're looking at roughly $275,000 a day. As you can see, this isn't all that large of a number when you just compare it to depreciation, let along the (rather high) cost of electricity to keep the plant up and running and the huge cost of all the raw materials for producing and cleaning the chips. In short, your cost of labor is VERY low as a proportion of the whole business, but the quality of your employees has a VERY direct relation to the quality of the product (most importantly yields and speed bins) that is coming out of the fab.
Long story short, you REALLY want to place your fab where you can get skilled workers, not where labor is cheap.
Of course, the determining factor on where the fab is built is always which government gives the best grants and loans to build the fab.
Transmeta's problem is that their chips actually don't consume less power than a number of Intel chips. The Transmeta chips have a maximum power draw of about 7-10W. Intel has several ULV chips that are in the same range. Now Transmeta has had a bit of an advantage in it's dynamic power consumption strategies, but the difference isn't huge, especially not when compared to the Pentium M processor.
The real question about the success of the Efficeon is whether or not it will be able to offer more performance for fewer dollars than Intel's ~800MHz Ultra Low Voltage Celeron processors. These chips consume very similar amounts of power and are both fairly low-cost chips.
FWIW the main reason why Transmeta powered laptops have used so little power as compared to Intel-based laptops is that the Transmeta ones tend to ship with very small, low powered screens, while the Intel-based ones tend to use larger screens. With the LCD screen consuming up to 25W on some laptops, any difference there can make a big difference. Things like hard drives and graphics controllers also play a major role in the total power consumption of a laptop. Most small to mid-sized laptops have a maximum power consumption of up around 50-60W. With the processor taking 7-10W of that it's an important part, but by no means the only piece of the puzzle.
The IBM PowerPC 970 (aka G5) does not use "a lot less power" than similar x86 chips. They're well within the same order of magnitude, and in fact the G5 and the AMD Athlon64/Opteron consume basically the same amount of power at the same clock speed. The Pentium4 consumes a bit more, but not by much. IBM doesn't bother documenting the power consumption of their processors (actually they don't bother publicly providing much useful documentation at all), but they listed the "typical" power of the 1.8GHz G5 at 48W. Given that "typical" power is usually around 75% of the maximum power consumption, and scaling things up to 2.0GHz, you're probably looking at about 70W of power. A bit lower than the 82W of power that the top-end P4's consume, but not by much.
That beign said, you're on the right track with the PowerPC idea. IBM's Blue Gene/L design uses and offshoot of the PowerPC 440 processor and manages some VERY impressive performance/W, probably a whole heck of a lot more than any of Transmeta chips are going to manage.
Re:What kind of marketing garbage is this crap?!
on
New Linux TPC-H Record Set
·
· Score: 5, Informative
The Itanium's have 512KB of L2 cache and 3MB of L3 cache, with it's L3 cache being faster and having lower latency than the L2 cache of the old Xeons.
Xeons are fine chips, but the 900MHz Xeon is totally outdated. A new 2.8GHz XeonMP system with 2MB of L3 cache would probably also be about 3.5 times faster on this test than the old 900MHz Xeon.
Wow, talk about using a NARROW definition of "largest Linux machine". There are larger Linux systems, some using thousands of processors in a cluster. There are also other companies out there building systems that can have up to 64 processors per node like this SGI system. But SGI is using a combination of software and hardware to get global shared memory instead of a more traditional cluster design.
Of course, Linux's NUMA optimizations are still in their infancy (they are pretty much non-existant in the 2.4.x kernel and still a work-in-progress for the 2.6.x kernel), so using NUMA over internode communication link is probably going to kill your scalability.
Holy crap this story is useless! Go to the TCP-H site and actually look at the results, it really is nothing even remotely impressive.
- It's NOT the fastest TCP-H result, it's the fastest LINUX TCP-H 1000GB result. Actually it's the ONLY Linux TCP-H 1000GB result. 5th of 8 overall
- It's not even offering very good bang for your buck, coming in 5th of 8 for Price/QphH ($156 US according to today's currency exchange). The only systems it managed to beat are two outdated systems (both from HP) and an old price for a Fujitsu system, quoted in euro (the same system offers the same performance but a lower price on a newer entry quoted in US $).
In short, if anything this suggests that Linux is a BAD choice for this work! The performance isn't there and the cost is high.
Where things get REALLY bad though is the claim that this is "3.5 times faster" than a system running IBM's DB2. This is just 100% pure bullshit! The new Linux/Oracle system runs 1.3GHz Itanium2 processors and Oracle 10g. The HP/Windows/DB2 system runs 900MHz Xeon processors and runs DB2 7.2 (8.1 is current version). What's more, the Oracle/Linux system isn't even 3.5 times faster, it's just 3.5 times faster PER PROCESSOR! Great, your brand-spanking new Itanium2 is 3.5 times faster than four year old Xeon 900MHz chips. Whoopie!
Note: if you do want to see impressive Linux results, look at what IBM is doing with their Opteron cluster and DB2 running under SuSE Linux. They turned in the top results in the two TPC-H tests they entered (100GB and 300GB).
We'll probably start seeing some mobile G5 chips once IBM starts producing the chips on a 90nm fab process (down from the current 130nm fab process). Dell and crew have proven that you CAN put a processor that consumes 50-70W of power in a notebook and still cool it (the "Mobile P4" chips consume about that much, as oppsed to the "Mobile P4-M" or "Pentium-M" chips that are much lower powered offerings), however it's not that great of an idea.
Once IBM shrinks the die down on a 90nm fab process though, their power consumption should drop a fair bit, making it fairly practical to put one of these chips in a notebook. FWIW 90nm production of the PPC 970 should start happening fairly soon (within the next 3-6 months?). At that point in time, all of your dream notebook should be an option except for the upgradable PCI-X graphics card. Upgradable graphics and notebooks almost never happen, and PCI-X is pointless for graphics (AGP 8x offers the same bandwidth as the fastest PCI-X speeds but it does so at a lower cost and without sharing the bus with all the other PCI-X devices).
Slashdot Mirror
Here's a list:
IBM BIOS: IBM PCs and laptops
AMI BIOS: umm.. I don't think anyone uses them anymore
Pheonix BIOS: everyone else
Since Pheonix bought out Award, they are basically the only player in the BIOS market.
Uhh, try using the CURRENT tarriff, and not the one that expired three years ago! Here is the current tarriff for blank media. Note that even though it only starts out as a proposal, these are the real tarriffs being collected, as per this decision.
Actual tarriff on blank data CDs in Canada is $0.21.
The current tarriff is only $0.21 per CD. The proposed increases to this tarriff would see that number rise to $0.59 per CD. Under the current rules, roughly half of the money you pay to buy a data CD for that latest Linux ISO is going to the recording artists of Canada because you making a data CD is hurting their livelyhood so badly. In the updated version, 3/4 of that Linux ISO will go to help our starving record execs.
Ohh, and as for who gets the money for your Linux ISOs...
"Since no inventory of privately copied tracks exists, distribution is based on representative samples of radio airplay and album sales, which are given equal weight in the distribution."
Better hope that the artists you like get lots of radio play, but chances are that most of our money is going to Celine Dion and Shania Twain.
He might not be able toback it up, but I certainly can. Here are current tarrifs and here are the proposed tarrifs. Note that the first set of proposed increases to tarriffs were somewhat shot down in favor of the ones listed above.
For those of you who want to go straight to the original, here is the official proposal. Note that in this proposal, they wanted to charge a full $21/GB for hard-drives in MP3 players. This actually would have raised the price of an iPod or similar such device (20GB hard drive) by a full $410. Given that the current going price for a 20GB iPod is roughly $500 to $600 CDN, this would have come close to doubling the price.
In the updated levy, the charge would be ~$110 for a 20GB MP3 player.
Actually that brings MS and Apple even for the past month at 1 a piece (Microsoft had a buffer overrun in the Workstation service).
Ohh, and both MS and Apple have had a security vulnerbility for their browser this month on top of the OS vulnerbilities listed above.
Linux doesn't seem to have had any new security vulnerbilities announced this month, though a few security fixes are filtering through for vulnerbilities announced in October. Both WinXP and OS X also had some similar fixes for earlier bugs.
Long story short, it doesn't matter what OS you run, you WILL have security vulnerbilities. Patch your OS and use a firewall already!
Umm.. Microsoft does not currently offer any anti-spam products.
On the other hand, they pay probably hundreds of millions of dollars to deal with the spam problem. Hotmail.com receives at least a billion spam messages a day, and probably closer to 2 billion. Do you know how much money it costs Microsoft to pay for that bandwidth? How many extra servers they have to pay for? Not to mention how much they're already paying third-parties anti-spam software companies?
And that isn't even starting to touch the huge amount of lost productivity at Microsoft caused by their employees receiving boatloads of spam (think: computer industry company with tends of thousands of employees, you better believe that MS recieves a lot of spam).
There is no way that MS is ever going to cover the expense of their receiving spam by any products. Trust me, MS wants spam to stop just as much as any other major corporation that has to deal with the problem.
How, pray tell, would that affect a spammer?! You don't honestly think that a spammer uses an ISPs legitmate e-mailer servers do you? Let me guess, you also think that those spams with "hotmail.com" from addresses were actually sent from Hotmail too?
Face it, spammers are criminals, and expecting simple laws to stop them is hugely naive when the multitude of existing laws that spammers break continue to go unenforced.
His name was Gordon Moore, and he worked for Intel... err, more to the point, he founded Intel along with Andy Grove and Robert Noyce. Of course, like pretty much all the important players in the semiconductor industry, he was working at Fairchild at the time that he released his paper. What he said was:
"The complexity for minimum component costs has increased at a rate of roughly a factor of two per year"
He predicted that this would hold true for 10 years, from 1965 to 1975. Of course, this prediction fairly quickly changed to doubling the transistor count every 18 months, since the 12 month prediction didn't hold true for very long. Now we're doubling transistor count roughly every 24 months.
FWIW anyone who's interested can take a quick read of Moore's whitepaper where this "law" came from.
Err, no. AMD just announced a new fab in Dresden that will produce chips at a 65nm fab process, just as they have been saying all along.
Of course, AMD's fab won't be ready until 2006, so Intel will have a lead, but that's nothing new. Intel is pretty much always 6 months ahead of AMD (and everyone else for that matter) at bringing out a new manufacturing process. Being the biggest semiconductor manufacturer with the largest budget and most buying power has it's advantages you know.
There are a number of things going on here, but a few important things to think of.
First off, with previous shrinking of the manufacturing process you could run the processor at a lower voltage. Most 500nm chips ran at 3.3V, 350nm chips ran at 2.8V, 250nm chips ran at 2.0V, 180nm chips ran at 1.75V and 130nm chips now run mostly at 1.55V. As you can see pretty quickly though, the difference in voltage isn't as much as it used to be, and with 90nm production, that difference is pretty much zero, most 90nm chips will probably run at about the same 1.55V of current (130nm) chips.
Second, the Prescott does add a bit more than just cache. Alongside the new SSE3 instructions, Intel is also making some fairly major changes to the P4 core, fixing some of the potential problem areas. I haven't heard 100% official confirmation, but apparently it adds a barrel shifter (should be very noticeable for D.net clients if they do) and fixes some of the scheduling issues with multiplications. These changes are going to result in extra transistors, and extra transistors means more power.
Also, the Prescott is supposed to improve hyperthreading. This is a good thing from a performance standpoint, it means that you'll get more of a performance boost from running two threads at once. The downside is that it means that the processor pipeline will be packed more fully, again increasing power consumption.
In short, there's lots to consider, no easy equation to get you the power consumption of the new chip.
Originally Moore's law stated that transistor density would double every 12 months. That was fairly quickly changed to say that it would double every 18 months. The current "law" states that transistor density doubles about every 24 months.
Long story short, we haven't really been following Moore's law for a little while, though we do continue to double the amount of bits we can stuff onto a piece of silicon at a fairly rapid pace. Intel's plan to bring out 65nm chips before the end of 2005 continues this trend.
FWIW IBM is also looking to bring out 65nm chips by late 2005/early 2006, while AMD is hoping to get their 65nm fab process up and running in their new fab early in 2006. TSMC and UMC are likely to follow in mid-2006, though I haven't heard any official comments from either.
Worst PCs? How about ANYTHING that was made by Packard Bell?
FWIW this guy also got a brief mention in this article.
Not only did he make death threats against the spammers, apparently he had been warned to stop by local authorities, but continued.
It's one thing to be pissed off at someone because they're annoying, but to continue making death threats against that person even after being told to stop by police is quite another thing.
Meant to provide one with my original post, but got distracted. Here it's mentioned in a CNet story about it. I think AMD actually stated the feature size would be 65nm in their conference call, though I didn't listen in.
Shouldn't really be a big surprise though, 2003 was when everyone was supposed to be switching to 90nm production (though it looks like it's actually going to happen mid to late 2004), and normally new feature sizes come out about every two to three years. So, with the new fab opening up in 2006, a 65nm feature size is right on track with expectations. Actually I think Intel is hoping to get to 65nm before the end of 2005, but then again, they also wanted to start shipping chips with 90nm features months ago.
Take a look at the photo in the top-left corner of the linked press release from AMD.
:>
Could that guy possibly be any more German?!
65nm initially, with plans to switch to the next generation (45nm?) a few years down the road.
Considering Cyrix no longer exists, probably never.
Good thing too since the entire valley flooded not too long ago!
As others have mentioned, Intel just has their chips packaged in Malaysia, the Philippines, etc. Similarly AMD has almost all of their chips packaged in Malaysia. FWIW most of Intel's actually chip fabs are in the US. Oregon to be exact. They also have a couple plants in Ireland and in Isreal. To the best of my knowledge they do not have any plants in south-east Asia.
In any case, here's a few numbers for you.. A typical plant these days costs about $2.5 billion dollars to build. Equipment in that plant is about another billion dollars and has a useful life of about 5 years if you can stretch it. The plant as a whole has a useful life of about 10 years before it needs a major overhaul.
So, simple bit of math tells us that you're looking at a fixed capital cost of somewhere around $4.5 billion. This gives you a rate of depresiation of roughly $1.2 million dollars a day.
Now, if you pay an average of $100,000/year to the ~1000 employees of the plant, you're looking at roughly $275,000 a day. As you can see, this isn't all that large of a number when you just compare it to depreciation, let along the (rather high) cost of electricity to keep the plant up and running and the huge cost of all the raw materials for producing and cleaning the chips. In short, your cost of labor is VERY low as a proportion of the whole business, but the quality of your employees has a VERY direct relation to the quality of the product (most importantly yields and speed bins) that is coming out of the fab.
Long story short, you REALLY want to place your fab where you can get skilled workers, not where labor is cheap.
Of course, the determining factor on where the fab is built is always which government gives the best grants and loans to build the fab.
Transmeta's problem is that their chips actually don't consume less power than a number of Intel chips. The Transmeta chips have a maximum power draw of about 7-10W. Intel has several ULV chips that are in the same range. Now Transmeta has had a bit of an advantage in it's dynamic power consumption strategies, but the difference isn't huge, especially not when compared to the Pentium M processor.
The real question about the success of the Efficeon is whether or not it will be able to offer more performance for fewer dollars than Intel's ~800MHz Ultra Low Voltage Celeron processors. These chips consume very similar amounts of power and are both fairly low-cost chips.
FWIW the main reason why Transmeta powered laptops have used so little power as compared to Intel-based laptops is that the Transmeta ones tend to ship with very small, low powered screens, while the Intel-based ones tend to use larger screens. With the LCD screen consuming up to 25W on some laptops, any difference there can make a big difference. Things like hard drives and graphics controllers also play a major role in the total power consumption of a laptop. Most small to mid-sized laptops have a maximum power consumption of up around 50-60W. With the processor taking 7-10W of that it's an important part, but by no means the only piece of the puzzle.
The IBM PowerPC 970 (aka G5) does not use "a lot less power" than similar x86 chips. They're well within the same order of magnitude, and in fact the G5 and the AMD Athlon64/Opteron consume basically the same amount of power at the same clock speed. The Pentium4 consumes a bit more, but not by much. IBM doesn't bother documenting the power consumption of their processors (actually they don't bother publicly providing much useful documentation at all), but they listed the "typical" power of the 1.8GHz G5 at 48W. Given that "typical" power is usually around 75% of the maximum power consumption, and scaling things up to 2.0GHz, you're probably looking at about 70W of power. A bit lower than the 82W of power that the top-end P4's consume, but not by much.
That beign said, you're on the right track with the PowerPC idea. IBM's Blue Gene/L design uses and offshoot of the PowerPC 440 processor and manages some VERY impressive performance/W, probably a whole heck of a lot more than any of Transmeta chips are going to manage.
The Itanium's have 512KB of L2 cache and 3MB of L3 cache, with it's L3 cache being faster and having lower latency than the L2 cache of the old Xeons.
Xeons are fine chips, but the 900MHz Xeon is totally outdated. A new 2.8GHz XeonMP system with 2MB of L3 cache would probably also be about 3.5 times faster on this test than the old 900MHz Xeon.
Wow, talk about using a NARROW definition of "largest Linux machine". There are larger Linux systems, some using thousands of processors in a cluster. There are also other companies out there building systems that can have up to 64 processors per node like this SGI system. But SGI is using a combination of software and hardware to get global shared memory instead of a more traditional cluster design.
Of course, Linux's NUMA optimizations are still in their infancy (they are pretty much non-existant in the 2.4.x kernel and still a work-in-progress for the 2.6.x kernel), so using NUMA over internode communication link is probably going to kill your scalability.
Holy crap this story is useless! Go to the TCP-H site and actually look at the results, it really is nothing even remotely impressive.
- It's NOT the fastest TCP-H result, it's the fastest LINUX TCP-H 1000GB result. Actually it's the ONLY Linux TCP-H 1000GB result. 5th of 8 overall
- It's not even offering very good bang for your buck, coming in 5th of 8 for Price/QphH ($156 US according to today's currency exchange). The only systems it managed to beat are two outdated systems (both from HP) and an old price for a Fujitsu system, quoted in euro (the same system offers the same performance but a lower price on a newer entry quoted in US $).
In short, if anything this suggests that Linux is a BAD choice for this work! The performance isn't there and the cost is high.
Where things get REALLY bad though is the claim that this is "3.5 times faster" than a system running IBM's DB2. This is just 100% pure bullshit! The new Linux/Oracle system runs 1.3GHz Itanium2 processors and Oracle 10g. The HP/Windows/DB2 system runs 900MHz Xeon processors and runs DB2 7.2 (8.1 is current version). What's more, the Oracle/Linux system isn't even 3.5 times faster, it's just 3.5 times faster PER PROCESSOR! Great, your brand-spanking new Itanium2 is 3.5 times faster than four year old Xeon 900MHz chips. Whoopie!
Note: if you do want to see impressive Linux results, look at what IBM is doing with their Opteron cluster and DB2 running under SuSE Linux. They turned in the top results in the two TPC-H tests they entered (100GB and 300GB).
We'll probably start seeing some mobile G5 chips once IBM starts producing the chips on a 90nm fab process (down from the current 130nm fab process). Dell and crew have proven that you CAN put a processor that consumes 50-70W of power in a notebook and still cool it (the "Mobile P4" chips consume about that much, as oppsed to the "Mobile P4-M" or "Pentium-M" chips that are much lower powered offerings), however it's not that great of an idea.
Once IBM shrinks the die down on a 90nm fab process though, their power consumption should drop a fair bit, making it fairly practical to put one of these chips in a notebook. FWIW 90nm production of the PPC 970 should start happening fairly soon (within the next 3-6 months?). At that point in time, all of your dream notebook should be an option except for the upgradable PCI-X graphics card. Upgradable graphics and notebooks almost never happen, and PCI-X is pointless for graphics (AGP 8x offers the same bandwidth as the fastest PCI-X speeds but it does so at a lower cost and without sharing the bus with all the other PCI-X devices).