It'd be nice if perl, python, ruby etc can run a lot faster than now. I know some work has been done with python. But just hope things advance at a faster rate...
The Opteron is much cheaper than 75% of the Itanium.
So far I've seen a fair number of reports from people that the Opteron actually works much better than the Itanium for most real world tasks using real world system configurations. I haven't seen a single report saying otherwise (other than from PR announcements).
A dual core Opteron only requires 233 million transistors and a 200 square mm die.
Whereas a low-end 3MB cache McKinley Itanium 2 has 221 million transistors for _one_ core. The more recent 1.6GHz Itaniums use up to 410 million transistors taking 374 square mm (still one core)!
So I wouldn't say the Itanium 2 is a better architecture for floating point.
I bet that the Itanium 2 design is much better in parallelizable floating point tasks than non-parallelizable tasks. If I'm right, then that those very tasks will be fairly easy to run well across multiple cores anyway.
The IBM POWER5 has 276 million transistors and IMO it makes far better use of them than the Itanium (of course the IBM eServer p5's 36MB off-chip L3 cache might help a bit too;) ).
Given that, I'd say there is really very little reason to go Itanium (especially since HP's commitment to Tandem, OpenVMS seems rather questionable, a pity actually).
Either Intel is nerfing their Itaniums (by going single core with so many transistors for cache vs multicore with less cache) or their architecture just doesn't work well in practice. In any case, if you're going to pay a lot for non-x86, might as well go IBM POWER5.
Otherwise, the Opterons are pretty decent. I'll leave the Intel x86 figures for someone else to do:).
Strange that I don't see MS Office anywhere in their drop down.
Yeah yeah, MS Office isn't that great blahblah, but I use Open Office at work and it sucks - slow and bloated - it makes my 2GB 3GHz machine seem slow, launch times are terrible even the 2nd time round. I've had snappier performance from 2.5MHz machines.
As for Windows on Linux: if Linux is just the kernel as many people here like to keep saying, Microsoft could theoretically put something like Windows on top of it.
I wonder how suse, redhat, KDE and Gnome would like that...
Maybe some person would be happy that it's no longer GNU/Linux but Windows/Linux.;);).
"Easy. Order a ton of dell servers, pay for the next or try to get them to do same day shipping, order a huge EMC san, slip the installer a few hundreds so they do it that day. "
Do keep in mind a scenario where it's a disaster that doesn't just affect your company but others as well. If everyone else is also ordering a ton of servers I doubt everyone is going to get them the next day.
They were bound by legacy. It's easier to extend x86 to 16 registers than 32 registers. I believe there were spare unused bits in some x86 instructions that they could use to do 16 registers, I guess they couldn't find enough to do 32 registers efficiently.
Plus in a talk at a university (stanford?), one of their designers claimed that once you hit 16 or 32 registers you get diminishing returns. Most bang for the buck at 16.
Also, if you have lots of registers saving and restoring them starts to take more time... The Itanium approach of having tons and tons of registers doesn't appear to have given it that much benefit.
Why don't let people code in microops? That's because x86 is now a defacto layer of abstraction. Much like high level language vs machine code. You write in x86, if Intel/AMD etc figure faster ways to do x86, you gain without needing to rewrite or recompile your code.
Also, x86 seems a fair bit more compact than most RISC code. AFAIK, most RISC code is fixed size, whereas common x86 operations tend to be smaller than uncommon ones. You get a fair bit of performance gain from smaller code size. So think of x86 as compressed code with modern x86 cpus decompressing x86 on the fly to RISC, from low bandwidth (mem, 2nd level cache) to high bandwidth - on chip (or trace cache with P4s).
Lastly the Opterons perform as well as the RISC chips, so 16 registers, x86 legacy etc isn't really such a big disadvantage.
The opterons are doing it even with a comparable transistor count. Go look at the Itanium transistor count, sure the actual core is small, but how well does it perform in real life without those large caches? I haven't seen any real life tests or benchmarks of Itaniums with small caches, most have effectively _double_ the transistor counts of x86 chips (opteron or P4). For that number of transistors (400 million?) you can have a _dual_ core opteron.
Check this out though: http://cbcl.mit.edu/cbcl/news/files/caltech-6-05.h tml
I'd think it's more likely that that single cell only fires to a particular sequence of waves in the neurons around it.
Still, I found that quite interesting. What happens if that cell dies? Do you suddenly have to "relearn" that particular memory item?
Maybe that's why once in a while a common word appears unfamiliar?
I'm not a neuroscientist etc, but if that "single cell" thingy recognizing something very specific and high level is true and that some cells migrate to form long term memories, then it's not too far fetched that you have a single cell that fires for a specific long term memory.
I guess as you recreate the brain patterns of that memory, that cell fires when you get a match. I say this because another recent research states that as people recall stuff in the past, they recreate the brain states.
Well, I'm sure the transplanted organs would include some stem cells, and those stem cells can move about and generate new cells.
After all, it has been discovered that a mother can have her baby son's stem cells end up in her brain. Look it up. (So in a way, the Bible's concept of a man and woman becoming one flesh can be true even in a near literal sense).
I won't be too surprised if simple urges and food preferences changed, but I'd be surprised if detailed facts actually got transferred.
That said, apparently the process of forming long term memories involve brain cells moving about.
Assuming 4 drives and a smart RAID10 controller, reads can actually be faster than RAID5.
This is because you can actually interleave reads from mirrored drives to get better read performance for single sequential reads. 3ware has controllers that do that. In fact it should be better than stripes, because the same data is present on all mirrored drives so you can read the first chunk from whichever drive with its head closest to the chunk and schedule the subsequent chunks from the rest of the drives, thus minimizing latency and still maintaining throughput. Whereas with stripes, the first chunk will only be on one drive, so you have no choice of drives to read from.
So for sequential access you'd get the combined read throughput of four drives, and the write throughput of two drives.
Whereas with RAID5, you only get the read throughput of N-1 drives and the write throughput is poor (especially for sustained writes).
Of course if your raid controller isn't smart or you are using Linux software raid then for RAID10, you'd get single sequential read throughput of N/2.
Still, Linux software RAID is smart enough to balance reads in mirrors. Say you have a mirror with two drives A and B, and two sequential reads going on, one read typically ends up on A and the other on B. So in many multiuser environments the lack of read interleaving might not be such a big deal. I haven't done tests for RAID10 yet though.
In the future if videographic (photographic+audio) memory becomes a "default install" in humans, what are organizations going to do?
Only hire humans without videographic memory? Such organizations will be at a greater disadvantage then.
Install DRM into such humans to prevent them using their memories in certain cases? Ugh. Solution seems a lot worse than the problem.
Anyway, the article is _crap_. There are quite different issues, by mixing them all together the author just confuses the matter. Maybe the author has an agenda and is just scaremongering people for some strategic/tactical purpose but perhaps I'm being too cynical (but hey he apparently is Principal Analyst with 35+ years experience blahblahblah).
In contrast, I'm just some slashdotter, but here's my 2 cents worth:
The issues are: 1) Data loss Solution = backups, backups, backups.
2) Unauthorized access to data Solution = Only allow trusted AND competent personnel access to critical data, and use encryption accordingly.
3) How do you know who can be trusted (and is competent[1])? Solution = only time and proper testing/observation can tell.
If you still can't figure it out then perhaps you yourself aren't competent enough to run your organization. If someone can't be trusted with little you can't trust them with a lot.
Y'know even if someone nefarious sneaks into your organization behaving like a model member for 15 years, gradually gaining increasing amounts of trust, only to betray you in the last year, you'd have got 15 very good years from that person;). And you must be quite a special case for all that effort eh?
4) "Will you even know?" Solution = Watermark data, and maybe even generate fake databases.
Even better if you can give different employees different distinctive data. This way if data leaks, there's a chance you'd know that it's from your organization, and perhaps who it came from.
For example the To and CC fields in most SMTP based email don't count for anything, so you can actually send each group[2] a different email, and there are ways to mark the email without changing the meaning too much (or even at all). I'm not sure of any email client that does this, but it sure is possible. You may also use different key phrases when composing your messages.
[1] A person may have a higher than average level of integrity, but they could just be incompetent/weak in some areas, so you still can't trust them for some stuff.
For example, though your Aunt May might be a genuinely trustworthy person, she might be incapable of locking down her computer securely so that the messages you send to her will never be leaked, and she might be incompentent at judging her competence in such things. So it may be wise to just not send messages to her via her computer, and just pass information via other means.
[2] Say you start with 3 groups, and then you "rotate" people around through those groups. That way you might eventually find out who's the leaky untrustworthy/incompetent person. There are other ways too, go figure it out yourself.
By having the drive with the media, you have the ability to use the "media" with any compatible interface. There are magnitudes more computers with SATA/PATA/SCSI interfaces in the world than there are with LTO or DDS drives.
If you have a physically risky environment then sure tapes are better.
Looks like there's already a stable abi/api for linux network drivers- it's called NDIS.;)
If the Linux folks don't want to make one, then it sure looks like the defacto will continue to be NDIS.
Someone might do something similar for graphics and sound. Then it will be interesting to see if Desktop Linux will perform as well as Desktop Windows.
People here complain about Windows supporting viruses and trojans across decades and having the same bugs across generations of Windows. That's what happens when you have a culture of maintaining backward (and bug) compatibility. It's why I can still run many dos games (and apps) in windows 2000/XP. Apparently Windows actually turns on bug emulation for some software - it has a list of applications which rely on various bugs, and Windows will behave accordingly for that application.
In contrast with Linux, OSS you have things like "Is it linked to glib22 or glib23?" or "Just compile from the sources like everyone else", or "open source your app".
With OSS, if your app relies on a certain behaviour and it stops working with the new software release when the bug is fixed, you'd probably just get flamed if you complain.
Not so long ago when people were complaining that Python 2.4 requires some MS dll that can't be redistributed freely and as a result py2exe stuff doesn't work so well, the responses they got from the Python people were quite telling. They were saying stuff like "Just get your users to install Python 2.4 or some other package with that DLL and then install your software", or "It's just like flash and shockwave, you need to install the player to run it" or making useless negative/sarcastic comments. As if python has Macromedia's marketing and tie-in power. And some places don't allow you to run flash/shockwave either.
Same silly attitudes with Linux people. If the Linux devs strongly discourage people from making drivers with an opaque binary or don't expose a fairly stable interface, why should people take the extra effort to write drivers for Linux? It's not like Linux has > 60% market share. Companies like IBM will provide server drivers for Linux, but there are just so many distros and versions they can support.
I hear some people say if hardware vendors just open source their drivers, then things will be fine and be ported to future versions. That's bullshit, there is no guarantee that any of the kernel devs will take care of it.
"The fact that you do not know of these Christian holidays throws your Christianity into serious doubt and betrays your ignorance of what it is to be a Christian."
One does NOT need to know about Christian holidays in order to be a Christian.
The main part of being a Christian is knowing and following Christ. Everything else can be derived from that.
Not knowing when God's Son's birthday was actually (I doubt most Christians know roughly when it was even), does not exclude you from being part of the family.
You become/are part of the family not because you strictly follow the rules of the family (but of course following the rules would be nice;) ) and not because you memorize and observe all the customs and practices of the family.
They were allowed to eat from any tree except one.
I wonder how things would have turned out if they had started with the tree of life first and only then worked their way to the tree of the knowledge of good and evil;). D'oh.
Anyway, yes they did not die immediately if you talk about physical life and death.
But much of the Bible is about spiritual life and death - the relationship with God, etc.
I'm willing to bet there's are bunches of neurons modelling/mirroring/simulating the perceivable world in your brain. And that's used to help predict what might happen next, or what could happen.
Such a predictive ability is very useful to most animals. And I believe "models" are a fairly simple way to do things with neurons AND they can be used to solve more than a few problems - they are not a "premature optimization", unlike many of those simplified equations.
I suggest that most humans are able to tell whether a tree branch or even a plank or sheet of metal is unlikely to safely hold their weight, just from looking at the object and having a previous experience of the strength a sample of the material the object was made of.
I believe that sort of thing is more like modelling and simulation in your brain, than calculation and equations.
It is more like building a scaled down model of a river and a dam and pouring water in to see if something will overflow.
Sure your neurons probably do the equivalent of some analog math to simulate some stuff, but I believe "simulation" is more useful and accurate for what our brains do for this sort of thing. Otherwise it would be like someone saying that the water in the scaled down model of a dam is doing calculus. May be true in a way, but not very useful.
Lastly, I suspect a fair bit of consciousness is you modelling and predicting yourself. To quote Dune: "Solve thyself". BUT I don't think that's all of it because there're still a few other things;).
Still given what happens to the US will affect most of the world, and seeing stuff like Diebold, and masses fall for "9/11=Iraq", and much other dubious stuff being allowed to happen and even condoned is very worrying.
If it's just some small weak country, it's not a big worry for the rest of the world.
Not sure what can really be done at the moment, given the current situation is like being asked by a Magician to pick a card out of a preselected hand of cards. Just hope one can at least slow things down.
" a HDD depends on both the magnet medium and the conected electronics. With a tape, the medium is seperated from the physical components that make it go"
Uh, isn't that a significant minus for tapes. Especially since tapes aren't much cheaper than HDDs. And tape drives are very expensive.
If you've bought an LTO3 tape drive, and LTO4 tapes become cheap, you don't get a free LTO4 drive with each LTO4 tape.
Whereas if you have been using 250GB SATA drives as backup media, when 400GB SATA drives become cheap, you can switch to using them as backup media.
I forsee that SATA is more likely to be around and easily available longer than LTO3 will (or whatever tape standard). Look how long PATA and SCSI have been around and compare with DDS3, DDS4 etc.
So years later, as long as I have a SATA interface, I can still read from an old SATA drive which should have had rather little wear and tear if kept under controlled conditions.
Whereas LTOx or DDSx tape drives might be quite rare and even more expensive. Even if the HDDs physical stuff seizes up, finding someone who can recover your data from the platters will be not much harder than finding someone with a working DDS4 tape drive.
Also, if you use HDD as backup media, you can do backups/restores to/from more than one HDD at the same time. Multiple HDDs are much cheaper than multiple tape drives.
Advantages of using tape: You can drop a tape onto a carpeted floor without much worry.
Tape autoloaders are more common than HDD autoloaders;). So if you will have massive amounts of offline data, you have to go for tape.
Making noise about GWB is more to have a greater impact on what happens to us in the future than making noise about FDR.
Why are you even talking about FDR? If you don't like what FDR did and GWB is doing something similar, why'd you even need to bring up FDR? You can just go say GWB is doing something wrong.
I'm not a US citizen, but way too many US people seem to treat this Democrat vs Republican thing the same way those Pro-wrestling commentators do:
No matter what your team does it can do no wrong - even if they are blatantly cheating. "So what if my team is illegally using a chair, hey your team did that too in 2002".
I guess it's fine if it's pro-wrestling, but when it's about the Government and Leaders of the most powerful nation in the world, that sort of thing is so _STUPID_ that it is disgusting.
If someone you support is doing something really wrong, get some integrity and tell them it is wrong. If they are good people, YOU ARE DOING THEM A FAVOUR, and when it comes your leaders, you are doing YOURSELF a favour. If they are really bad people, don't bother telling them, just vote for someone better EVEN if he/she is not the same party you normally support.
Lastly, please make sure your voting systems work correctly. If you guys can spend billions of USD and thousands of lives on elections in Iraq, you should at least get something decent, rather than the dubious crap that Diebold has made. I personally find it strange that the most powerful country in the world picks its leaders using something with the quality of a failed high school project. Especially when they seem to think that free elections and democracy is so important...
HEY US CITIZENS, WAKE UP! WHAT IS WRONG WITH YOU PEOPLE???
If you can't get that right, outsource your elections to India if you have to. At least the Indians manage to get their elections working for their 1 billion citizens without too many riots or ending up in a civil war.
"i mean, look, i'm as attached to my genitals as anyone"
Most people do prefer to be totally attached to their genitals. Not just slightly attached.
Also, prioritizing genitals over brains appears to have worked fine for most species in the world. Genitals have a better track record for keeping a species around than brains.
You might be able to enter as the previous person;).
On many fingerprint readers there's usually this glass thingy where you put your finger. Usually the oils from the previous print will be left on the glass, by breathing on it, you can often fog up the glass to create an image which can be read by the reader.
With most biometric systems you need a human or something monitoring that no funny business like this is happening.
Anyway, you should just make a distinctive mold, register it and put it on your keychain.
The gaussian distribution of IQ is because the tests and scores are _designed_ to give that particular distribution given a typical population in the first place!
Would be interesting to see if you would get that same distribution if you put a random sample of Askhenazi Jews on those tests.
Even if you have a gaussian shaped distribution AND the average of a particular group is still stuck at average, you still can have a higher percentage of extreme cases.
All you need is a wider bell curve.
I postulate that this explains the difference between the sexes in humans - that the males have wider bell curves for many things (like intelligence) than females, so though the average male isn't more intelligent than the average female, there are more exceptionally stupid males, and more exceptionally smart males.
Anyway, despite what the "politically correct" people _say_, I believe that there's such a thing as race, while the differences are not as marked as breeds of dogs, there are still significant differences.
However, one should still hire based on merit and suitability[1], too bad if turns out a race is totally unsuitable for a job.
I don't see many NBA basketball players of pygmy descent, but I guess if there's a pygmy who can shoot 3 pointers from way out, there's a chance for them but still hard to shoot through the body of someone leaning over you. Maybe a "freak" pygmy with gigantism syndrome...
[1] It would be interesting if a white russian jew was picked to head the "National Association of Black Journalists" for instance. My guess is race matters for that organisation;).
Also if you are doing medical/bio studies on the differences between groups of humans, race is a reasonable _starting_point_. There are a great number of human "breeds" that could be classified as "black", of course in the US, most of the blacks there come from small subset of those breeds.
It's not the average who count for Nobel Prizes, major advances in tech, science, arts or even stuff like chess, tennis, F1 racing.
It's the extraordinary ones.
So it doesn't matter so much even if the average from a particular group is below average, the big deal is whether the "exceptionals" are really really good.
Of course it would mean that such a group with a "below average" average might not be as good at producing "cannonfodder" workers.
In reality, the curve is unlikely to go that way. But it is something to keep in mind, especially when so many people seem to like to compare the averages for these.
Sure there may be reasons to compare say the average 100 metre runners from various places. But AFAIK, it is more usual to compare the top runners.
Well, think of it this way, you probably have a better chance of getting promoted or raised based on merit in your current company than the other ones.
Unless you lack merit, this is normally a good thing.
Slashdot Mirror
It'd be nice if perl, python, ruby etc can run a lot faster than now. I know some work has been done with python. But just hope things advance at a faster rate...
The Opteron is much cheaper than 75% of the Itanium.
;) ).
:).
So far I've seen a fair number of reports from people that the Opteron actually works much better than the Itanium for most real world tasks using real world system configurations. I haven't seen a single report saying otherwise (other than from PR announcements).
A dual core Opteron only requires 233 million transistors and a 200 square mm die.
Whereas a low-end 3MB cache McKinley Itanium 2 has 221 million transistors for _one_ core. The more recent 1.6GHz Itaniums use up to 410 million transistors taking 374 square mm (still one core)!
So I wouldn't say the Itanium 2 is a better architecture for floating point.
I bet that the Itanium 2 design is much better in parallelizable floating point tasks than non-parallelizable tasks. If I'm right, then that those very tasks will be fairly easy to run well across multiple cores anyway.
The IBM POWER5 has 276 million transistors and IMO it makes far better use of them than the Itanium (of course the IBM eServer p5's 36MB off-chip L3 cache might help a bit too
AMD Opteron (TM) 180 2 cores, 1 chip, 2 cores/chip
spec cfp2000 rate base=32.3
spec cint2000 rate base=35.0
transistors = 233M x 1
AMD Opteron (TM) 280 4 cores, 2 chips, 2 cores/chip
spec cfp2000 rate base=68.7
spec cint2000 rate base=71.8
transistors = 233M x 2
AMD Opteron (TM) 880 8 cores, 4 chips, 2 cores/chip
spec cfp2000 rate base=129
spec cint2000 rate base=134
transistors = 233M x 4
IBM eServer p5 570 (1900 MHz, 4 CPU) 4 cores, 2 chips, 2 cores/chip (SMT on)
spec cfp2000 rate base=125
spec cint2000 rate base=74.4
transistors = 276M x 2
HP Integrity rx4640-8 (1.6GHz/9MB Itanium 2) 4 cores, 4 chips, 1 core/chip
spec cfp2000 rate base=77.9
spec cint2000 rate base=72.5
transistors = 410M x 4
Given that, I'd say there is really very little reason to go Itanium (especially since HP's commitment to Tandem, OpenVMS seems rather questionable, a pity actually).
Either Intel is nerfing their Itaniums (by going single core with so many transistors for cache vs multicore with less cache) or their architecture just doesn't work well in practice. In any case, if you're going to pay a lot for non-x86, might as well go IBM POWER5.
Otherwise, the Opterons are pretty decent. I'll leave the Intel x86 figures for someone else to do
Strange that I don't see MS Office anywhere in their drop down.
;) ;).
Yeah yeah, MS Office isn't that great blahblah, but I use Open Office at work and it sucks - slow and bloated - it makes my 2GB 3GHz machine seem slow, launch times are terrible even the 2nd time round. I've had snappier performance from 2.5MHz machines.
As for Windows on Linux: if Linux is just the kernel as many people here like to keep saying, Microsoft could theoretically put something like Windows on top of it.
I wonder how suse, redhat, KDE and Gnome would like that...
Maybe some person would be happy that it's no longer GNU/Linux but Windows/Linux.
"Easy. Order a ton of dell servers, pay for the next or try to get them to do same day shipping, order a huge EMC san, slip the installer a few hundreds so they do it that day. "
Do keep in mind a scenario where it's a disaster that doesn't just affect your company but others as well. If everyone else is also ordering a ton of servers I doubt everyone is going to get them the next day.
They were bound by legacy. It's easier to extend x86 to 16 registers than 32 registers. I believe there were spare unused bits in some x86 instructions that they could use to do 16 registers, I guess they couldn't find enough to do 32 registers efficiently.
Plus in a talk at a university (stanford?), one of their designers claimed that once you hit 16 or 32 registers you get diminishing returns. Most bang for the buck at 16.
Also, if you have lots of registers saving and restoring them starts to take more time... The Itanium approach of having tons and tons of registers doesn't appear to have given it that much benefit.
Why don't let people code in microops? That's because x86 is now a defacto layer of abstraction. Much like high level language vs machine code. You write in x86, if Intel/AMD etc figure faster ways to do x86, you gain without needing to rewrite or recompile your code.
Also, x86 seems a fair bit more compact than most RISC code. AFAIK, most RISC code is fixed size, whereas common x86 operations tend to be smaller than uncommon ones. You get a fair bit of performance gain from smaller code size. So think of x86 as compressed code with modern x86 cpus decompressing x86 on the fly to RISC, from low bandwidth (mem, 2nd level cache) to high bandwidth - on chip (or trace cache with P4s).
Lastly the Opterons perform as well as the RISC chips, so 16 registers, x86 legacy etc isn't really such a big disadvantage.
The opterons are doing it even with a comparable transistor count. Go look at the Itanium transistor count, sure the actual core is small, but how well does it perform in real life without those large caches? I haven't seen any real life tests or benchmarks of Itaniums with small caches, most have effectively _double_ the transistor counts of x86 chips (opteron or P4). For that number of transistors (400 million?) you can have a _dual_ core opteron.
Check this out though: http://cbcl.mit.edu/cbcl/news/files/caltech-6-05.h tml
i me_travel.php
;).
I'd think it's more likely that that single cell only fires to a particular sequence of waves in the neurons around it.
Still, I found that quite interesting. What happens if that cell dies? Do you suddenly have to "relearn" that particular memory item?
Maybe that's why once in a while a common word appears unfamiliar?
I'm not a neuroscientist etc, but if that "single cell" thingy recognizing something very specific and high level is true and that some cells migrate to form long term memories, then it's not too far fetched that you have a single cell that fires for a specific long term memory.
I guess as you recreate the brain patterns of that memory, that cell fires when you get a match. I say this because another recent research states that as people recall stuff in the past, they recreate the brain states.
http://www.seedmagazine.com/news/2005/12/mental_t
Sorry too lazy to do proper links
Well, I'm sure the transplanted organs would include some stem cells, and those stem cells can move about and generate new cells.
After all, it has been discovered that a mother can have her baby son's stem cells end up in her brain. Look it up. (So in a way, the Bible's concept of a man and woman becoming one flesh can be true even in a near literal sense).
I won't be too surprised if simple urges and food preferences changed, but I'd be surprised if detailed facts actually got transferred.
That said, apparently the process of forming long term memories involve brain cells moving about.
Assuming 4 drives and a smart RAID10 controller, reads can actually be faster than RAID5.
This is because you can actually interleave reads from mirrored drives to get better read performance for single sequential reads. 3ware has controllers that do that. In fact it should be better than stripes, because the same data is present on all mirrored drives so you can read the first chunk from whichever drive with its head closest to the chunk and schedule the subsequent chunks from the rest of the drives, thus minimizing latency and still maintaining throughput. Whereas with stripes, the first chunk will only be on one drive, so you have no choice of drives to read from.
So for sequential access you'd get the combined read throughput of four drives, and the write throughput of two drives.
Whereas with RAID5, you only get the read throughput of N-1 drives and the write throughput is poor (especially for sustained writes).
Of course if your raid controller isn't smart or you are using Linux software raid then for RAID10, you'd get single sequential read throughput of N/2.
Still, Linux software RAID is smart enough to balance reads in mirrors. Say you have a mirror with two drives A and B, and two sequential reads going on, one read typically ends up on A and the other on B. So in many multiuser environments the lack of read interleaving might not be such a big deal. I haven't done tests for RAID10 yet though.
I agree that RAID controllers fail way too often.
In the future if videographic (photographic+audio) memory becomes a "default install" in humans, what are organizations going to do?
;). And you must be quite a special case for all that effort eh?
Only hire humans without videographic memory? Such organizations will be at a greater disadvantage then.
Install DRM into such humans to prevent them using their memories in certain cases? Ugh. Solution seems a lot worse than the problem.
Anyway, the article is _crap_. There are quite different issues, by mixing them all together the author just confuses the matter. Maybe the author has an agenda and is just scaremongering people for some strategic/tactical purpose but perhaps I'm being too cynical (but hey he apparently is Principal Analyst with 35+ years experience blahblahblah).
In contrast, I'm just some slashdotter, but here's my 2 cents worth:
The issues are:
1) Data loss
Solution = backups, backups, backups.
2) Unauthorized access to data
Solution = Only allow trusted AND competent personnel access to critical data, and use encryption accordingly.
3) How do you know who can be trusted (and is competent[1])?
Solution = only time and proper testing/observation can tell.
If you still can't figure it out then perhaps you yourself aren't competent enough to run your organization. If someone can't be trusted with little you can't trust them with a lot.
Y'know even if someone nefarious sneaks into your organization behaving like a model member for 15 years, gradually gaining increasing amounts of trust, only to betray you in the last year, you'd have got 15 very good years from that person
4) "Will you even know?"
Solution = Watermark data, and maybe even generate fake databases.
Even better if you can give different employees different distinctive data. This way if data leaks, there's a chance you'd know that it's from your organization, and perhaps who it came from.
For example the To and CC fields in most SMTP based email don't count for anything, so you can actually send each group[2] a different email, and there are ways to mark the email without changing the meaning too much (or even at all). I'm not sure of any email client that does this, but it sure is possible. You may also use different key phrases when composing your messages.
[1] A person may have a higher than average level of integrity, but they could just be incompetent/weak in some areas, so you still can't trust them for some stuff.
For example, though your Aunt May might be a genuinely trustworthy person, she might be incapable of locking down her computer securely so that the messages you send to her will never be leaked, and she might be incompentent at judging her competence in such things. So it may be wise to just not send messages to her via her computer, and just pass information via other means.
[2] Say you start with 3 groups, and then you "rotate" people around through those groups. That way you might eventually find out who's the leaky untrustworthy/incompetent person. There are other ways too, go figure it out yourself.
By having the drive with the media, you have the ability to use the "media" with any compatible interface. There are magnitudes more computers with SATA/PATA/SCSI interfaces in the world than there are with LTO or DDS drives.
If you have a physically risky environment then sure tapes are better.
Looks like there's already a stable abi/api for linux network drivers- it's called NDIS. ;)
If the Linux folks don't want to make one, then it sure looks like the defacto will continue to be NDIS.
Someone might do something similar for graphics and sound. Then it will be interesting to see if Desktop Linux will perform as well as Desktop Windows.
People here complain about Windows supporting viruses and trojans across decades and having the same bugs across generations of Windows. That's what happens when you have a culture of maintaining backward (and bug) compatibility. It's why I can still run many dos games (and apps) in windows 2000/XP. Apparently Windows actually turns on bug emulation for some software - it has a list of applications which rely on various bugs, and Windows will behave accordingly for that application.
In contrast with Linux, OSS you have things like "Is it linked to glib22 or glib23?" or "Just compile from the sources like everyone else", or "open source your app".
With OSS, if your app relies on a certain behaviour and it stops working with the new software release when the bug is fixed, you'd probably just get flamed if you complain.
Not so long ago when people were complaining that Python 2.4 requires some MS dll that can't be redistributed freely and as a result py2exe stuff doesn't work so well, the responses they got from the Python people were quite telling. They were saying stuff like "Just get your users to install Python 2.4 or some other package with that DLL and then install your software", or "It's just like flash and shockwave, you need to install the player to run it" or making useless negative/sarcastic comments. As if python has Macromedia's marketing and tie-in power. And some places don't allow you to run flash/shockwave either.
Same silly attitudes with Linux people. If the Linux devs strongly discourage people from making drivers with an opaque binary or don't expose a fairly stable interface, why should people take the extra effort to write drivers for Linux? It's not like Linux has > 60% market share. Companies like IBM will provide server drivers for Linux, but there are just so many distros and versions they can support.
I hear some people say if hardware vendors just open source their drivers, then things will be fine and be ported to future versions. That's bullshit, there is no guarantee that any of the kernel devs will take care of it.
I have to recompile the vmware shims everytime I update the SuSE kernel. So far it works, but I don't have to do the same thing on Windows.
"The fact that you do not know of these Christian holidays throws your Christianity into serious doubt and betrays your ignorance of what it is to be a Christian."
;) ) and not because you memorize and observe all the customs and practices of the family.
One does NOT need to know about Christian holidays in order to be a Christian.
The main part of being a Christian is knowing and following Christ. Everything else can be derived from that.
Not knowing when God's Son's birthday was actually (I doubt most Christians know roughly when it was even), does not exclude you from being part of the family.
You become/are part of the family not because you strictly follow the rules of the family (but of course following the rules would be nice
There's another thing you should note.
;). D'oh.
The tree of life was also in the garden.
They were allowed to eat from any tree except one.
I wonder how things would have turned out if they had started with the tree of life first and only then worked their way to the tree of the knowledge of good and evil
Anyway, yes they did not die immediately if you talk about physical life and death.
But much of the Bible is about spiritual life and death - the relationship with God, etc.
Maybe hell is when you get to use God mode, but nobody wants to play with you anymore, _forever_.
I'm willing to bet there's are bunches of neurons modelling/mirroring/simulating the perceivable world in your brain. And that's used to help predict what might happen next, or what could happen.
;).
Such a predictive ability is very useful to most animals. And I believe "models" are a fairly simple way to do things with neurons AND they can be used to solve more than a few problems - they are not a "premature optimization", unlike many of those simplified equations.
I suggest that most humans are able to tell whether a tree branch or even a plank or sheet of metal is unlikely to safely hold their weight, just from looking at the object and having a previous experience of the strength a sample of the material the object was made of.
I believe that sort of thing is more like modelling and simulation in your brain, than calculation and equations.
It is more like building a scaled down model of a river and a dam and pouring water in to see if something will overflow.
Sure your neurons probably do the equivalent of some analog math to simulate some stuff, but I believe "simulation" is more useful and accurate for what our brains do for this sort of thing. Otherwise it would be like someone saying that the water in the scaled down model of a dam is doing calculus. May be true in a way, but not very useful.
Lastly, I suspect a fair bit of consciousness is you modelling and predicting yourself. To quote Dune: "Solve thyself". BUT I don't think that's all of it because there're still a few other things
Sorry bout my rant earlier.
Still given what happens to the US will affect most of the world, and seeing stuff like Diebold, and masses fall for "9/11=Iraq", and much other dubious stuff being allowed to happen and even condoned is very worrying.
If it's just some small weak country, it's not a big worry for the rest of the world.
Not sure what can really be done at the moment, given the current situation is like being asked by a Magician to pick a card out of a preselected hand of cards. Just hope one can at least slow things down.
" a HDD depends on both the magnet medium and the conected electronics. With a tape, the medium is seperated from the physical components that make it go"
;). So if you will have massive amounts of offline data, you have to go for tape.
Uh, isn't that a significant minus for tapes. Especially since tapes aren't much cheaper than HDDs. And tape drives are very expensive.
If you've bought an LTO3 tape drive, and LTO4 tapes become cheap, you don't get a free LTO4 drive with each LTO4 tape.
Whereas if you have been using 250GB SATA drives as backup media, when 400GB SATA drives become cheap, you can switch to using them as backup media.
I forsee that SATA is more likely to be around and easily available longer than LTO3 will (or whatever tape standard). Look how long PATA and SCSI have been around and compare with DDS3, DDS4 etc.
So years later, as long as I have a SATA interface, I can still read from an old SATA drive which should have had rather little wear and tear if kept under controlled conditions.
Whereas LTOx or DDSx tape drives might be quite rare and even more expensive. Even if the HDDs physical stuff seizes up, finding someone who can recover your data from the platters will be not much harder than finding someone with a working DDS4 tape drive.
Also, if you use HDD as backup media, you can do backups/restores to/from more than one HDD at the same time. Multiple HDDs are much cheaper than multiple tape drives.
Advantages of using tape:
You can drop a tape onto a carpeted floor without much worry.
Tape autoloaders are more common than HDD autoloaders
"But the constitution trumps FISA. FISA can't take powers away from the president that he is granted under the constitution."
BUT this is not just a card or table top game.
This is about how the top leaders of the most powerful country in the world get to behave.
Even if you are right and those are the rules AND how they are intended to be interpreted, is that how things should be?
Anyway, please explain which part of the constitution allows the President to ignore/override FISA.
Get your priorities right.
Making noise about GWB is more to have a greater impact on what happens to us in the future than making noise about FDR.
Why are you even talking about FDR? If you don't like what FDR did and GWB is doing something similar, why'd you even need to bring up FDR? You can just go say GWB is doing something wrong.
I'm not a US citizen, but way too many US people seem to treat this Democrat vs Republican thing the same way those Pro-wrestling commentators do:
No matter what your team does it can do no wrong - even if they are blatantly cheating. "So what if my team is illegally using a chair, hey your team did that too in 2002".
I guess it's fine if it's pro-wrestling, but when it's about the Government and Leaders of the most powerful nation in the world, that sort of thing is so _STUPID_ that it is disgusting.
If someone you support is doing something really wrong, get some integrity and tell them it is wrong. If they are good people, YOU ARE DOING THEM A FAVOUR, and when it comes your leaders, you are doing YOURSELF a favour. If they are really bad people, don't bother telling them, just vote for someone better EVEN if he/she is not the same party you normally support.
Lastly, please make sure your voting systems work correctly. If you guys can spend billions of USD and thousands of lives on elections in Iraq, you should at least get something decent, rather than the dubious crap that Diebold has made. I personally find it strange that the most powerful country in the world picks its leaders using something with the quality of a failed high school project. Especially when they seem to think that free elections and democracy is so important...
HEY US CITIZENS, WAKE UP! WHAT IS WRONG WITH YOU PEOPLE???
If you can't get that right, outsource your elections to India if you have to. At least the Indians manage to get their elections working for their 1 billion citizens without too many riots or ending up in a civil war.
Sheesh.
"i mean, look, i'm as attached to my genitals as anyone"
Most people do prefer to be totally attached to their genitals. Not just slightly attached.
Also, prioritizing genitals over brains appears to have worked fine for most species in the world. Genitals have a better track record for keeping a species around than brains.
You might be able to enter as the previous person ;).
On many fingerprint readers there's usually this glass thingy where you put your finger. Usually the oils from the previous print will be left on the glass, by breathing on it, you can often fog up the glass to create an image which can be read by the reader.
With most biometric systems you need a human or something monitoring that no funny business like this is happening.
Anyway, you should just make a distinctive mold, register it and put it on your keychain.
The gaussian distribution of IQ is because the tests and scores are _designed_ to give that particular distribution given a typical population in the first place!
;).
Would be interesting to see if you would get that same distribution if you put a random sample of Askhenazi Jews on those tests.
Even if you have a gaussian shaped distribution AND the average of a particular group is still stuck at average, you still can have a higher percentage of extreme cases.
All you need is a wider bell curve.
I postulate that this explains the difference between the sexes in humans - that the males have wider bell curves for many things (like intelligence) than females, so though the average male isn't more intelligent than the average female, there are more exceptionally stupid males, and more exceptionally smart males.
Anyway, despite what the "politically correct" people _say_, I believe that there's such a thing as race, while the differences are not as marked as breeds of dogs, there are still significant differences.
However, one should still hire based on merit and suitability[1], too bad if turns out a race is totally unsuitable for a job.
I don't see many NBA basketball players of pygmy descent, but I guess if there's a pygmy who can shoot 3 pointers from way out, there's a chance for them but still hard to shoot through the body of someone leaning over you. Maybe a
"freak" pygmy with gigantism syndrome...
[1] It would be interesting if a white russian jew was picked to head the "National Association of Black Journalists" for instance. My guess is race matters for that organisation
Also if you are doing medical/bio studies on the differences between groups of humans, race is a reasonable _starting_point_. There are a great number of human "breeds" that could be classified as "black", of course in the US, most of the blacks there come from small subset of those breeds.
It's not the average who count for Nobel Prizes, major advances in tech, science, arts or even stuff like chess, tennis, F1 racing.
It's the extraordinary ones.
So it doesn't matter so much even if the average from a particular group is below average, the big deal is whether the "exceptionals" are really really good.
Of course it would mean that such a group with a "below average" average might not be as good at producing "cannonfodder" workers.
In reality, the curve is unlikely to go that way. But it is something to keep in mind, especially when so many people seem to like to compare the averages for these.
Sure there may be reasons to compare say the average 100 metre runners from various places. But AFAIK, it is more usual to compare the top runners.
Well, think of it this way, you probably have a better chance of getting promoted or raised based on merit in your current company than the other ones.
Unless you lack merit, this is normally a good thing.