A quick google for PGP whole disk encryption yield PGP's spec page
Which probably means that the scheme is your typical PGP your symmetric key...
So it seems that an AES acceleration, such as the VIA PadLock, could potentially mitigate the performance issues. X-bit labs just had a minor blurb recently about how the Via Nano with PadLock trounced the Core 2 Quad...
It just seems like an abrupt end to what they could've really explored, and I was suspicious as to why they clammed up like that.
This should come to no surprise, but damn, they rather air censored stories of explosive pants, but not even get the RFID through.
The thing is this stuff is so hackable, it's not much of a secret anyways. Every Defcon features something about RFIDs, so what good do they think is security through obscurity?
I think that's just a silly way to prune people for no good reason.
There are lots of office with inter-office gossip, and inter-office politics. Does that me those people are ready to betray trade secrets to outsiders?
It's easy for one to say that one have cracked passwords on his friends machine for shits and giggles. Often, the same people wouldn't dream of doing the same thing for say their workplace, unless it was part of the workplace audit. Slippery slope arguments aren't valid logically.
Maybe a lot of these Chief Execs in transition are paid well to prevent trouble, but people often forget about the sysadmin, the guy who's responsible for the machines and all the data's in it?
It's only human nature that once betrayed, they won't feel loyalty. I've seen that sysadmins are often burdened with responsibility at the level of some of these Chief Execs, and these CEx would cause major trouble for the sysadmin, and not provide the proper budget, but the sysadmin would never be compensated appropriately. That's just starting off at the wrong foot, if they expected loyalty from the sysadmin.
For me, the failure from kinetic shock has become a bigger thing since I had 2 iPods failed.
I didn't drop them, but I kept them in my jacket, in a protective holster, but I wasn't as careful with the jacket as with the iPod.
Consumers rarely crank databases or Final Cut Pro on a laptop, and if they think the laptop could last for 3 years or so, the reliability of the SSD would become a factor.
Those of us who need lots of sequential reads or absolute battery life would do a lot more research, and generally not "consumers" as we might think.
Given that cooked food is easier to absorb, then what would the raw food proponents say about that. These people keep saying that the nutrients in raw food would break down during the cooking process, yet cooking food may actually help fuel humans much better than animals that eat food raw.
OLPC has lost its way a long time ago --- at their design phase. In previous post on related topics, I've made no secret that I'm displeased with Sugar, and the whole design process (it has no real focus initially).
Now with a friend of mine who actually owns the OLPC, and having my hands on it, it's safe to conclude that Sugar is crap.
I've seen non-technical and technical types struggle with the GUI. The GUI is pretty slow, despite the boost in the hardware standards, going from IIRC Geode GX 500 (433MHz) to a Geode LX, and doubling the RAM.
There should've been clear design focus, treat the laptop as what it is (a specialized, embedded-system grade laptop), and design the software around it.
Instead, the Asus Eee PC with their distro is doing much better as a computing device than the OLPC. Had there been effort in a) designing educational software, and b) screw the "language-less" GUI, and just work on something that's a lot more intuitive, OLPC could've been a success.
The OLPC has been overbudget (it isn't a $100 laptop, not close), and underwhelming. I see that Negroponte flirting with Windows as more of an effort to extract some utility out of the hardware. Whether he's doing the right thing, or he's betraying the ideals, or whatever, I personally don't really care anymore.
I've seen a lot of WD drives die, and I wouldn't trust it anyways. If I'm going to build a new rig, I still won't use WD drives, no matter how fast they are. They're still offering subpar warranty AFAIK.
I worked as help desk at a bioinformatics research facility, with roughly 200 people, and I can fit the number of power users that I could remotely trust to run their own machine in one hand. And 3 of them have gone over our heads - one wiped his own RHEL Linux (not that I'm a fan, but it's managed) with his own Ubuntu install, causing us grief when we change settings. He also cause a Kent State Computing Science PhD (who's more like a n00b who can't type his password right) to demand the "same" setup, burning up weeks of time for 2 out of 4 IT staff, myself included. The other 2 would routinely try to install pirated software on work computers.
And we do try to install software in time for our users. We would try to allocate the right software in time, and if there's no reasonable way to do it (i.e. the user can't get the funding), we try to offer alternatives. In the past, yes, the IT department had been sluggish, but the majority of them have left, and we do try to provide good service.
Apparently, in a bioinformatics research facility, most of the staff who do research don't know jack about computers, or how to maintain them. If the users are allowed to manage their own machine, I would spend so much time fixing machines, I would want to jump off the building.
Thank god I left that place. It was bad enough with the existing setup. To think that most users can maintain their machines is pure folly.
If you don't board first, you don't always get space for the luggage. Typically, first class people board together. Given that 2 first class people on the same row boards at nearly the same time, one of them is guaranteed to find a luggage compartment that isn't empty.
Suppose late guy's row number n > current boarding row number c.
He would be stuck waiting for the people who are in row number c to get settled, then he could proceed pass them to be back of the plane at row n, and put his luggage away, and perhaps barge into his seats.
In the other case, where n c, then he simply has to find his place in the line-up and board properly.
Alternatively, you could just stall the guy, proceed as normal, so that they could go to the back of the plane with no one putting their luggages in the compartments.
So their reasoning was that by embedded a GPU-style core, which tends to do vector/matrix ops very well, you can basically map all the SIMD instructions to the GPU-style core. If you modify the GPU-style core so that it can execute x86 style operations, albeit slowly, you could write schedulers that would dump the memory bound ops onto that core. They could also extend the x86_64 ISA to allow for quick matrix op in the GPGPU.
Mind you ATI was probably not their first pick, but they failed in negotiating with nVidia.
In regards to gluing the cores together, the reason why Intel's method works well is that most of the application that uses their 2x2 core CPUs aren't well multi-threaded. Once they are, and you schedule 2 threads with high level of memory sharing on separate cores, it would start snooping caches of the other die through the FSB, and that's a very high penalty. Curiously, the K8 architecture with HyperTransport would mitigate performance issues for cross core snooping, had they just glued 2 K8s together on the same package.
AMD is fundamentally more technologically driven, and Intel is more business driven. A lot of what they're doing is actually prepping themselves for on-going research. e.g. Research has shown that the core-independent clocking in Phenom/Barcelona will allow for finer-grained control in power usage. Schedulers that are power-aware, and has a power-cap (e.g. the CPU cannot exceed a certain Wattage), will be able to squeeze more performance out of core-independent clocking, than chip-wide clocking. That means, suppose that Phenom and C2D performs exactly the same, other than the Phenom can clock the cores independently, then a scheduler with a strict wattage bound would be able to squeeze much more performance out of Phenom than the C2D without exceeding the bound.
Their purchase of ATi, and probably to embed GPGPU, is somewhat inline with research on heterogeneous systems, where the cores perform differently. A decent scheduler that would assign the right process to the right core can actually squeeze more performance than your typical homogeneous multicore occupying the same area.
Curiously, Intel is taking a long long time to actually integrate the memory controller, and ditching FSB. They are also taking their sweet time making a "true quad core", and haven't been able to clock the cores independently, even in their Core 2 Duo. Instead, they have been just shrinking and shrinking the transistor side, and leveraging their benefits, using newer techniques to combat transistor leakage due to the shrink. Making true quad cores aren't easy. Making cores that clock independently isn't easy either - the processor would have multiple clock domains, and there are huge synchronization problems to overcome. Making true quad cores with core-independent clocking would be exceptionally hard. For Intel, a giant in semiconductor manufacturing comparing to AMD (AMD is always at least 12 months behind on feature-size shrink), they have yet to produce anything that resembles that.
Say what you will about the Core 2 Quads, and how they trounce what AMD offers. Energy management will become more and more prominent issue than just pure performance, and Intel can't bank on feature size shrink anymore. They've got the hafnium high-K to help mitigate problems on Penryn, but transistor size and speed will peak soon, and they would have to move to true multicore, and probably heterogeneous multicore. AMD is still not completely out at the server market - HyperTransport does make SMP much easier (and far less contention compared to FSB), and as long as they survive their immediate problems, I don't see the going away soon.
See: Single-ISA Heterogeneous Multi-Core Architectures for Multithreaded Workload Performance, and a bunch of other papers...
Your drive died while running Ubuntu has very little to do with the OS itself. I had a 80GB seagate 7200.10 ATA that failed while running *gasp* OpenBSD, do I blame OpenBSD? Even the slashdotters who regularly flames de Raadt won't blame him or his OS for drive failure.
Drive failure happens, even within warranty period. That's why if you choose your parts, choose those that have a long warranty period, and if someone's important enough, keep a backup somewhere.
However, I do, in general agree with you on the problems of Linux. It isn't the perception of free. Joe and Jane Blow will get a computer that either comes with an OS, or get their friend's smart-alecky kid to install something. If that smart-alecky kid isn't a Linux proponent, Windows will get slapped on. Even if he is, chances are Windows will be installed because he doesn't want to explain Linux's quirks to tech-illiterate folks.
Even when you overcome that, they would occasionally stick a Windows-only install CD into the Linux box, and wonder "WTF? Why won't this install". Wine may now be "beta", but is there an easy setup to detect Windows apps and try and run them? Does Wine have good compatibility to the plethora of Windows software out there?
Reason why machines like eeePC works is that the applications that most people would use it for are included, and they don't expect to install more software (e.g. there isn't a CDROM drive on there). I'm not saying that Linux has to adapt Window's GUI mannerisms - people can adjust to that. But Linux needs to be even more adaptable to Windows. There's a very strong compatibility problem that Linux needs to combat.
Either that, or people would start offering Linux with a Windows VM installed in it. And simply tell Mr. and Mrs. Blow that if their software doesn't run, start the Windows VM, and if something went wrong, they can roll back to a snapshot of the VM.
The curious thing is that the Isaiah is heading towards OoO, whereas Intel's going to build the first in-order chip since the Pentium in Silverthorn.
C7 already has a good track-record for small form factor, low power, and providing acceptable performance at that category. IMO with the OoO they're heading more towards the laptop market, and I think they could've done something at least less conventional with the design.
Imagine that they modified the C7-M in-order execution core to a 4-way, fine grain interleaved multithreading, and have 2 cores. The existing C7-M has a short pipe, so pipeflushes aren't as penalizing. At the clockspeed that they're starting at (2GHz), each thread would have acceptable performance for your typical workload. And as OSes are becoming more thread happy (OSX is definitely one of them), such design would be at least something different than ordinary. It would be like having a cut down Sun Niagara in your laptop.
The current design would make it work decently well for low end laptop and desktops, but I can't help but think that the core now has a bunch of stuff that they can't exactly turn off - I haven't heard of a CPU that could switch off its OoO and retire queue, and the die size has increased significantly compared to the C7.
Thank you for saying what is exactly needed. I was on Helldesk for the last 8 months, and I've observed that one of the guy that I work closely with would routinely put on his headphones if he need to work on something with some concentration. Whenever this mildly-annoying guy who chats way too long comes by and finds him for something, he basically can't really get back into the groove for a couple of hours.
There were countless time where I would be testing things, or just going though some long and arduous procedures that demanded my attention, and I would be interrupted by someone with a stupid, inane question, simply because they think they're better than everyone else, and refuses to use the ticketing system. We could spend all the time we want educating users on how to submit requests, or spend time writing how-tos to minimize routine questions, only to have the same users to come back and interrupt us, with little respect of the effort we've gone through to ask them to follow procedure. But no, if a cheque is lost in the mail, you would have to go through their procedures and there's little you could do about it.
I was working in an grant-based bioinformatics research facility, in which the people could be mostly separated into administration, research/development, lab techs, and a smaller group of computer and engineer techs. I simply cannot deal with how we kept hiring researchers who can't even handle a computer. There's this PHD from Kent state at Computing Science that could barely type, and thought that we changed his password while he neglected to hunt-and-peck his password correctly. This was after I've spent a few days building his personal machine according to someone else's unsupported (and unsanctioned) Ubuntu box, and joined it properly to our systems.
You call us having unreasonable demand? Go work helldesk for a while. I was horrifyingly underpaid - especially after factoring Co-op fees into my pay equation; I dealt with the stupidest computer users who are actually doing bioinformatics research for a cancer agency; and our requests are constantly disrespected.
All of that, and probably if they ever had a budget crunch, we would've been the first to go.
From this hard drive recovery guy, if your drive is multiplatter, you can just take it apart, twist the disks, simply because the servo data won't align again, and the drive would be unrecoverable. For single platter, you could just take it apart and expose the surface to sandpaper, magnet, knife, etc to physically damage the platter enough to make the drive not worth recovering. Besides, the opening of the drive would expose the platter to dust, and that and of itself could make the drive full of errors, given enough dust.
Re:The classmate hardware SUCKS, at least...
on
Negroponte vs Intel
·
· Score: 1
"I question some of the OLPC's intent, but their hardware design blows away that Intel POS its not even funny."
Having good hardware design doesn't mean jack if the software is slow as hell. Their intention of creating a GUI that can be recoded was a poor decision. They had no respect to the fact that the hardware should've been treated as an embedded system, and for all the builds that I've tried, either on emulator, or on test boards, sucks.
The Asus EeePC has demonstrated that it can be done, and it can be done well. OLPC's direction (or lack of) caused a whole group at my university to drop out of interest.
"would you rather give $200 of rice to a dictator that the people will never see, or try and get them a machine that can help spread education and freedom to peoples all over the world?" With the sluggish UI based on Sugar, I wonder how much education the kids would have. The idea isn't bad, but the program started without focus. They refused to acknowledge that the machine's an embedded system and chose a language that has a high memory footprint for a major part of the system. All of that based on the reasoning that we want to teach the kids programming... What's more important in education? The basics that they're not learning now? or the programming that depends on the basics that they're not really getting.
As much of a troll Dvorak is, I think he might luck out on this one. I'll take my chances and drop a $200 bag of rice. What's a dictator going to do with excessive rice in storage anyways?
Really? You have printer problems with Ubuntu Gutsy? What printer do you use?
And what is with the OSX interface that infuriates you? At least I can name them. I don't like the stoplight adjustments. I understand the Red X, and the Yellow -, but I never understood the Green +, since it never really does what I want. I also don't like the use of Finder to connect to server, and if it fails, it takes down the whole Finder to it. But other than that, there isn't any GUI elements that really pisses me off, and I've been using a Mac at work for about 8 months now. Would I use a Mac even with my gripes? Certainly.
And I do agree with you on the Vista bit, I went to microsoft.com, and that's the listed price from them.
However, I wouldn't mind hearing you qualify the gripes that you have. I personally don't have much problems with Ubuntu - I use a Samsung ML-1710, and the only gripe with printing is that it seems to print multiple pages as multiple jobs, and OSX gripes are listed above.
"THIS is torture - using pain compliance to subdue a subject who has been rendered harmless by the situation, or who was always harmless, but resisting arrest as best they are able (if that)." Uh... Tasers are meant to override your skeletal muscles so that you won't be able to fight back. Pain is more or less a side-effect from the use of applying electroshock.
"Just like XP flopped when people were complaining for ages that thousands of applications wouldn't work on it, very few DOS programs wouldn't work and it seemingly didn't offer enough benefits to counter-act this?"
Application compatibility get fixed, and legacy DOS program can be run on legacy OS. Which is why eventually WinXP was somewhat more accepted. Also I found out the hard way that Win2k doesn't offer remoe desktop, which can be convenient, and reason enough why I nuke Win2k boxes for WinXP.
Windows Vista, however, wouldn't even run ok on a laptop with 1GB of RAM. One of the big shots at work got a neat little Sony Vaio and it was loaded with Vista Business something, with 1GB of RAM, and I've never seen the hard drive light turn off. As soon as we found that Sony offered WinXP drivers, and I wiped the box to WinXP, the thing is instantly speedier.
I might buy new boxes loaded with min 2GB for WinXP, but that's mostly because 2GB of RAM is in the sweet spot. WinXP can work with 512MB quite handily.
How did Vista chews up so much resources is beyond me, and most IT professional, when confronted with a case of Vista in person, would conclude that they want none of it. We still have P3 beige boxes and P4 2GHz with 512MB of RAM at best, and we have 0 intention to move to Vista.
When our ACs that powers the server room (1 dedicated, and 1 that cools the offices, and supplements the server) both went out, the temperature of the server room rose rather quickly. It hit 30C pretty fast, and we do have a cluster at work.
Raising the set-point, especially for high density computing, is folly, especially for those who needs a cluster, but don't pay someone to babysit the server room 24/7. In that article, if it is indeed the case that a 5kW rack will shutdown within 3 minutes, even having a person present may not be enough to prevent an unplanned shutdown.
I'm afraid when we're talking about PEBKAC, we aren't talking your average user who can't tell you which part is which on the inside of the computer; we're talking about someone like a Computing Science post-doc who mistypes their password using the index finger pecking technique.
An average user can be taught to at least use an anti-virus, anti-spyware and firewall, given some time and effort. They might even fork out cash for software.
You probably don't want to look inside any of your computers if you're going to be stupid enough to link what's going on in Myanmar with the Chinese - you'll find plenty of made in China parts inside. Also, by your logic, don't buy anything Made in America, Made in UK, etc...
Slashdot Mirror
A quick google for PGP whole disk encryption yield PGP's spec page
Which probably means that the scheme is your typical PGP your symmetric key...
So it seems that an AES acceleration, such as the VIA PadLock, could potentially mitigate the performance issues.
X-bit labs just had a minor blurb recently about how the Via Nano with PadLock trounced the Core 2 Quad...
It just seems like an abrupt end to what they could've really explored, and I was suspicious as to why they clammed up like that.
This should come to no surprise, but damn, they rather air censored stories of explosive pants, but not even get the RFID through.
The thing is this stuff is so hackable, it's not much of a secret anyways. Every Defcon features something about RFIDs, so what good do they think is security through obscurity?
I think that's just a silly way to prune people for no good reason.
There are lots of office with inter-office gossip, and inter-office politics. Does that me those people are ready to betray trade secrets to outsiders?
It's easy for one to say that one have cracked passwords on his friends machine for shits and giggles. Often, the same people wouldn't dream of doing the same thing for say their workplace, unless it was part of the workplace audit. Slippery slope arguments aren't valid logically.
Maybe a lot of these Chief Execs in transition are paid well to prevent trouble, but people often forget about the sysadmin, the guy who's responsible for the machines and all the data's in it?
It's only human nature that once betrayed, they won't feel loyalty. I've seen that sysadmins are often burdened with responsibility at the level of some of these Chief Execs, and these CEx would cause major trouble for the sysadmin, and not provide the proper budget, but the sysadmin would never be compensated appropriately. That's just starting off at the wrong foot, if they expected loyalty from the sysadmin.
For me, the failure from kinetic shock has become a bigger thing since I had 2 iPods failed.
I didn't drop them, but I kept them in my jacket, in a protective holster, but I wasn't as careful with the jacket as with the iPod.
Consumers rarely crank databases or Final Cut Pro on a laptop, and if they think the laptop could last for 3 years or so, the reliability of the SSD would become a factor.
Those of us who need lots of sequential reads or absolute battery life would do a lot more research, and generally not "consumers" as we might think.
Given that cooked food is easier to absorb, then what would the raw food proponents say about that. These people keep saying that the nutrients in raw food would break down during the cooking process, yet cooking food may actually help fuel humans much better than animals that eat food raw.
OLPC has lost its way a long time ago --- at their design phase.
In previous post on related topics, I've made no secret that I'm displeased with Sugar, and the whole design process (it has no real focus initially).
Now with a friend of mine who actually owns the OLPC, and having my hands on it, it's safe to conclude that Sugar is crap.
I've seen non-technical and technical types struggle with the GUI. The GUI is pretty slow, despite the boost in the hardware standards, going from IIRC Geode GX 500 (433MHz) to a Geode LX, and doubling the RAM.
There should've been clear design focus, treat the laptop as what it is (a specialized, embedded-system grade laptop), and design the software around it.
Instead, the Asus Eee PC with their distro is doing much better as a computing device than the OLPC. Had there been effort in a) designing educational software, and b) screw the "language-less" GUI, and just work on something that's a lot more intuitive, OLPC could've been a success.
The OLPC has been overbudget (it isn't a $100 laptop, not close), and underwhelming. I see that Negroponte flirting with Windows as more of an effort to extract some utility out of the hardware. Whether he's doing the right thing, or he's betraying the ideals, or whatever, I personally don't really care anymore.
I've seen a lot of WD drives die, and I wouldn't trust it anyways.
If I'm going to build a new rig, I still won't use WD drives, no matter how fast they are. They're still offering subpar warranty AFAIK.
I worked as help desk at a bioinformatics research facility, with roughly 200 people, and I can fit the number of power users that I could remotely trust to run their own machine in one hand. And 3 of them have gone over our heads - one wiped his own RHEL Linux (not that I'm a fan, but it's managed) with his own Ubuntu install, causing us grief when we change settings. He also cause a Kent State Computing Science PhD (who's more like a n00b who can't type his password right) to demand the "same" setup, burning up weeks of time for 2 out of 4 IT staff, myself included. The other 2 would routinely try to install pirated software on work computers.
And we do try to install software in time for our users. We would try to allocate the right software in time, and if there's no reasonable way to do it (i.e. the user can't get the funding), we try to offer alternatives. In the past, yes, the IT department had been sluggish, but the majority of them have left, and we do try to provide good service.
Apparently, in a bioinformatics research facility, most of the staff who do research don't know jack about computers, or how to maintain them. If the users are allowed to manage their own machine, I would spend so much time fixing machines, I would want to jump off the building.
Thank god I left that place. It was bad enough with the existing setup. To think that most users can maintain their machines is pure folly.
Suppose late guy's row number n > current boarding row number c.
He would be stuck waiting for the people who are in row number c to get settled, then he could proceed pass them to be back of the plane at row n, and put his luggage away, and perhaps barge into his seats.
In the other case, where n c, then he simply has to find his place in the line-up and board properly.
Alternatively, you could just stall the guy, proceed as normal, so that they could go to the back of the plane with no one putting their luggages in the compartments.
So their reasoning was that by embedded a GPU-style core, which tends to do vector/matrix ops very well, you can basically map all the SIMD instructions to the GPU-style core.
If you modify the GPU-style core so that it can execute x86 style operations, albeit slowly, you could write schedulers that would dump the memory bound ops onto that core. They could also extend the x86_64 ISA to allow for quick matrix op in the GPGPU.
Mind you ATI was probably not their first pick, but they failed in negotiating with nVidia.
In regards to gluing the cores together, the reason why Intel's method works well is that most of the application that uses their 2x2 core CPUs aren't well multi-threaded. Once they are, and you schedule 2 threads with high level of memory sharing on separate cores, it would start snooping caches of the other die through the FSB, and that's a very high penalty. Curiously, the K8 architecture with HyperTransport would mitigate performance issues for cross core snooping, had they just glued 2 K8s together on the same package.
AMD is fundamentally more technologically driven, and Intel is more business driven. A lot of what they're doing is actually prepping themselves for on-going research. e.g. Research has shown that the core-independent clocking in Phenom/Barcelona will allow for finer-grained control in power usage. Schedulers that are power-aware, and has a power-cap (e.g. the CPU cannot exceed a certain Wattage), will be able to squeeze more performance out of core-independent clocking, than chip-wide clocking. That means, suppose that Phenom and C2D performs exactly the same, other than the Phenom can clock the cores independently, then a scheduler with a strict wattage bound would be able to squeeze much more performance out of Phenom than the C2D without exceeding the bound.
Their purchase of ATi, and probably to embed GPGPU, is somewhat inline with research on heterogeneous systems, where the cores perform differently. A decent scheduler that would assign the right process to the right core can actually squeeze more performance than your typical homogeneous multicore occupying the same area.
Curiously, Intel is taking a long long time to actually integrate the memory controller, and ditching FSB. They are also taking their sweet time making a "true quad core", and haven't been able to clock the cores independently, even in their Core 2 Duo. Instead, they have been just shrinking and shrinking the transistor side, and leveraging their benefits, using newer techniques to combat transistor leakage due to the shrink. Making true quad cores aren't easy. Making cores that clock independently isn't easy either - the processor would have multiple clock domains, and there are huge synchronization problems to overcome. Making true quad cores with core-independent clocking would be exceptionally hard. For Intel, a giant in semiconductor manufacturing comparing to AMD (AMD is always at least 12 months behind on feature-size shrink), they have yet to produce anything that resembles that.
Say what you will about the Core 2 Quads, and how they trounce what AMD offers. Energy management will become more and more prominent issue than just pure performance, and Intel can't bank on feature size shrink anymore. They've got the hafnium high-K to help mitigate problems on Penryn, but transistor size and speed will peak soon, and they would have to move to true multicore, and probably heterogeneous multicore. AMD is still not completely out at the server market - HyperTransport does make SMP much easier (and far less contention compared to FSB), and as long as they survive their immediate problems, I don't see the going away soon.
See: Single-ISA Heterogeneous Multi-Core Architectures for Multithreaded Workload Performance, and a bunch of other papers...
Your drive died while running Ubuntu has very little to do with the OS itself. I had a 80GB seagate 7200.10 ATA that failed while running *gasp* OpenBSD, do I blame OpenBSD? Even the slashdotters who regularly flames de Raadt won't blame him or his OS for drive failure.
Drive failure happens, even within warranty period. That's why if you choose your parts, choose those that have a long warranty period, and if someone's important enough, keep a backup somewhere.
However, I do, in general agree with you on the problems of Linux. It isn't the perception of free. Joe and Jane Blow will get a computer that either comes with an OS, or get their friend's smart-alecky kid to install something. If that smart-alecky kid isn't a Linux proponent, Windows will get slapped on. Even if he is, chances are Windows will be installed because he doesn't want to explain Linux's quirks to tech-illiterate folks.
Even when you overcome that, they would occasionally stick a Windows-only install CD into the Linux box, and wonder "WTF? Why won't this install". Wine may now be "beta", but is there an easy setup to detect Windows apps and try and run them? Does Wine have good compatibility to the plethora of Windows software out there?
Reason why machines like eeePC works is that the applications that most people would use it for are included, and they don't expect to install more software (e.g. there isn't a CDROM drive on there). I'm not saying that Linux has to adapt Window's GUI mannerisms - people can adjust to that. But Linux needs to be even more adaptable to Windows. There's a very strong compatibility problem that Linux needs to combat.
Either that, or people would start offering Linux with a Windows VM installed in it. And simply tell Mr. and Mrs. Blow that if their software doesn't run, start the Windows VM, and if something went wrong, they can roll back to a snapshot of the VM.
The curious thing is that the Isaiah is heading towards OoO, whereas Intel's going to build the first in-order chip since the Pentium in Silverthorn.
C7 already has a good track-record for small form factor, low power, and providing acceptable performance at that category. IMO with the OoO they're heading more towards the laptop market, and I think they could've done something at least less conventional with the design.
Imagine that they modified the C7-M in-order execution core to a 4-way, fine grain interleaved multithreading, and have 2 cores. The existing C7-M has a short pipe, so pipeflushes aren't as penalizing. At the clockspeed that they're starting at (2GHz), each thread would have acceptable performance for your typical workload. And as OSes are becoming more thread happy (OSX is definitely one of them), such design would be at least something different than ordinary. It would be like having a cut down Sun Niagara in your laptop.
The current design would make it work decently well for low end laptop and desktops, but I can't help but think that the core now has a bunch of stuff that they can't exactly turn off - I haven't heard of a CPU that could switch off its OoO and retire queue, and the die size has increased significantly compared to the C7.
Thank you for saying what is exactly needed.
I was on Helldesk for the last 8 months, and I've observed that one of the guy that I work closely with would routinely put on his headphones if he need to work on something with some concentration. Whenever this mildly-annoying guy who chats way too long comes by and finds him for something, he basically can't really get back into the groove for a couple of hours.
There were countless time where I would be testing things, or just going though some long and arduous procedures that demanded my attention, and I would be interrupted by someone with a stupid, inane question, simply because they think they're better than everyone else, and refuses to use the ticketing system. We could spend all the time we want educating users on how to submit requests, or spend time writing how-tos to minimize routine questions, only to have the same users to come back and interrupt us, with little respect of the effort we've gone through to ask them to follow procedure. But no, if a cheque is lost in the mail, you would have to go through their procedures and there's little you could do about it.
I was working in an grant-based bioinformatics research facility, in which the people could be mostly separated into administration, research/development, lab techs, and a smaller group of computer and engineer techs. I simply cannot deal with how we kept hiring researchers who can't even handle a computer. There's this PHD from Kent state at Computing Science that could barely type, and thought that we changed his password while he neglected to hunt-and-peck his password correctly. This was after I've spent a few days building his personal machine according to someone else's unsupported (and unsanctioned) Ubuntu box, and joined it properly to our systems.
You call us having unreasonable demand? Go work helldesk for a while. I was horrifyingly underpaid - especially after factoring Co-op fees into my pay equation; I dealt with the stupidest computer users who are actually doing bioinformatics research for a cancer agency; and our requests are constantly disrespected.
All of that, and probably if they ever had a budget crunch, we would've been the first to go.
I was glad to walk out of there after 8 months.
From this hard drive recovery guy, if your drive is multiplatter, you can just take it apart, twist the disks, simply because the servo data won't align again, and the drive would be unrecoverable.
For single platter, you could just take it apart and expose the surface to sandpaper, magnet, knife, etc to physically damage the platter enough to make the drive not worth recovering.
Besides, the opening of the drive would expose the platter to dust, and that and of itself could make the drive full of errors, given enough dust.
"I question some of the OLPC's intent, but their hardware design blows away that Intel POS its not even funny."
Having good hardware design doesn't mean jack if the software is slow as hell. Their intention of creating a GUI that can be recoded was a poor decision. They had no respect to the fact that the hardware should've been treated as an embedded system, and for all the builds that I've tried, either on emulator, or on test boards, sucks.
The Asus EeePC has demonstrated that it can be done, and it can be done well. OLPC's direction (or lack of) caused a whole group at my university to drop out of interest.
As any one use rpm lately? Even with the yum front-end, it's still a PITA.
"would you rather give $200 of rice to a dictator that the people will never see, or try and get them a machine that can help spread education and freedom to peoples all over the world?"
With the sluggish UI based on Sugar, I wonder how much education the kids would have.
The idea isn't bad, but the program started without focus. They refused to acknowledge that the machine's an embedded system and chose a language that has a high memory footprint for a major part of the system. All of that based on the reasoning that we want to teach the kids programming... What's more important in education? The basics that they're not learning now? or the programming that depends on the basics that they're not really getting.
As much of a troll Dvorak is, I think he might luck out on this one. I'll take my chances and drop a $200 bag of rice. What's a dictator going to do with excessive rice in storage anyways?
Really? You have printer problems with Ubuntu Gutsy? What printer do you use?
And what is with the OSX interface that infuriates you? At least I can name them. I don't like the stoplight adjustments. I understand the Red X, and the Yellow -, but I never understood the Green +, since it never really does what I want. I also don't like the use of Finder to connect to server, and if it fails, it takes down the whole Finder to it. But other than that, there isn't any GUI elements that really pisses me off, and I've been using a Mac at work for about 8 months now.
Would I use a Mac even with my gripes? Certainly.
And I do agree with you on the Vista bit, I went to microsoft.com, and that's the listed price from them.
However, I wouldn't mind hearing you qualify the gripes that you have. I personally don't have much problems with Ubuntu - I use a Samsung ML-1710, and the only gripe with printing is that it seems to print multiple pages as multiple jobs, and OSX gripes are listed above.
"THIS is torture - using pain compliance to subdue a subject who has been rendered harmless by the situation, or who was always harmless, but resisting arrest as best they are able (if that)."
Uh... Tasers are meant to override your skeletal muscles so that you won't be able to fight back. Pain is more or less a side-effect from the use of applying electroshock.
"Just like XP flopped when people were complaining for ages that thousands of applications wouldn't work on it, very few DOS programs wouldn't work and it seemingly didn't offer enough benefits to counter-act this?"
Application compatibility get fixed, and legacy DOS program can be run on legacy OS. Which is why eventually WinXP was somewhat more accepted. Also I found out the hard way that Win2k doesn't offer remoe desktop, which can be convenient, and reason enough why I nuke Win2k boxes for WinXP.
Windows Vista, however, wouldn't even run ok on a laptop with 1GB of RAM. One of the big shots at work got a neat little Sony Vaio and it was loaded with Vista Business something, with 1GB of RAM, and I've never seen the hard drive light turn off.
As soon as we found that Sony offered WinXP drivers, and I wiped the box to WinXP, the thing is instantly speedier.
I might buy new boxes loaded with min 2GB for WinXP, but that's mostly because 2GB of RAM is in the sweet spot. WinXP can work with 512MB quite handily.
How did Vista chews up so much resources is beyond me, and most IT professional, when confronted with a case of Vista in person, would conclude that they want none of it. We still have P3 beige boxes and P4 2GHz with 512MB of RAM at best, and we have 0 intention to move to Vista.
When our ACs that powers the server room (1 dedicated, and 1 that cools the offices, and supplements the server) both went out, the temperature of the server room rose rather quickly. It hit 30C pretty fast, and we do have a cluster at work.
Raising the set-point, especially for high density computing, is folly, especially for those who needs a cluster, but don't pay someone to babysit the server room 24/7. In that article, if it is indeed the case that a 5kW rack will shutdown within 3 minutes, even having a person present may not be enough to prevent an unplanned shutdown.
I'm afraid when we're talking about PEBKAC, we aren't talking your average user who can't tell you which part is which on the inside of the computer; we're talking about someone like a Computing Science post-doc who mistypes their password using the index finger pecking technique.
An average user can be taught to at least use an anti-virus, anti-spyware and firewall, given some time and effort. They might even fork out cash for software.
You probably don't want to look inside any of your computers if you're going to be stupid enough to link what's going on in Myanmar with the Chinese - you'll find plenty of made in China parts inside.
Also, by your logic, don't buy anything Made in America, Made in UK, etc...