Last time I priced up CPUs (admittedly a while ago, and I was looking at mid-to-highish-range rather than bleeding edge) Intel's offerings at the same price also won over slightly more expensive AMDs in benchmark races, but equivalent motherboard were so expensive that it tipped things back in favour of the AMD unit for my needs.
I took that $100 and put it towards a decent video card instead.
Processors really aren't that much of a big deal anymore...
It is surprising how many people do not understand this. Rather than getting a bleeding edge CPU (and mobo+RAM to go with it) you will often see much more benefit saving some on the CPU-et-al and spending a but more on the graphics card (if you are a gamer and haven't already specified something silly in that respect) or getting better drives (a reasonable SSD won't set you back too much, and can make make much more useful difference to everyday use than spending the extra on a bleeding-edge CPU).
They all use up to 125W. They will use something approaching that much if every core and other part (the memory controller and cache banks) are at full tilt. Under more normal loading conditions you will see considerable difference between some chips that have the same power requirements on paper.
But do many games take advantage of many cores? I have a four core box and when I've checked while running a game (CPU monitor running on the other screen) one or two cores are pretty busy and the other two are not working all that much (I assume some of the threads have their affinities set so they don't bounce between cores, otherwise I'd expect the load to look a bit more even). I usually recommend people go for two fast cores rather than 4 slower ones (or 4 faster ones rather than 6/8 slower ones if they are set on spending more on the CPU instead of following my other advice and putting the extra cash into the graphics card and/or considering an SSD drive) for fast home machines. Outside of artificial tests, scientific number-crunching, bitcoin, and server-side use (where things tend to take better advantage of multi-cpu/core arrangements), have you ever seen a 6-core CPU under full load?
I am still using XP Pro. SP3 from 2002 and Debian from 2005.
I hope those Debian installs are more up-to-date than that as no named release point from that year is still supported for security patches (the longer support cycle is why I've slowly moved from Debian to Ubuntu's LTS releases since Debian's release cycle sped up and the support windows shortened with them). But your point is perfectly valid: I run XPsp3 or Ubuntu 10.04 on my desktop/netbook and for the most part Ubuntu 10.04 or Debian/Stable server-side. Work is a little different as we are a Microsoft shop for what earns our money, but a number of our servers are Debian/OldStable (Lenny) with more recently installed/updated ones being Ubuntu 10.04.
Ubuntu need to decide whether they are "the Linux for the rest of us" or "the bleeding edge".
My interpretation is that they are trying to do both, with the stable and long-term supported versions every two years and three releases between that are more experimental. It makes sense to make major changes now, while they've still got another release to show improvement/refinement in (11.10) before heading towards the next LTS release (12.04). It is a pity that this isn't made clear on any of the promotional materials (though that could also mean that I am just plain wrong in my interpretation!).
I'm sticking with 10.04 for now. It'll still get security updates until early 2013, it is stable, works well for my needs, has uptodate enough packages and good stable PPAs available for the main things you might want to update (i.e. if you want to move to FF4 for the extra speed enhancements over 3.6.x - I might do that soon). I'll be keeping an eye on how things go over the next 11 months (until the next LTS release, 12.04) so a month or few after that is released I can try it in an informed manner and decide whether to upgrade or move on to another distro before the 10.04 support period expires.
Maybe I'm not being demanding enough of Ubuntu, but I really don't see why people are getting quite so worked up at this point.
Which is why I'm sticking with 10.04. It'll still get security updates until early 2013 and possibly beyond (depending on what the different desktop/server support windows turn out to mean in practise), it is stable, works well for my needs, has uptodate enough packages and good stable PPAs available for the main things you might want to update (i.e. if you want to move to FF4 for the extra speed enhancements over 3.6.x - I might do that soon).
Predictable and stable is what LTS releases are intended for. I've assumed that the releases in between are test-beds for people who need/want the bleeding edge and don't mind occasional minor scalping due to their position as pioneers. I'll be keeping an eye on how things go over the next 11 months (until the next LTS release, 12.04, so a month or few after that is release I can try it in an informed manner and decide whether to upgrade or move on to something else before the 10.04 support period expires.
The next release of Unity will likely fix the major problems.
Or not. But we'll find out in 12.04. I'm keeping my netbook at 10.04 at the moment (until I got the new one recently I was still running 9.10 which worked just fine for regular use of OO.o Write/Sheet and Firefox & Chromium (the four apps I use most on my portable device, others I use regularly being a syntax highlighting text editor, VLC for video playback, the default media player for MP3s, and a WinXP VM in VitualBox for basic testing of web related stuff in IE).
Maybe I'm not that demanding a user, but 10.04 seems perfectly fine and almost certainly will be for another year or more. The default packages are uptodate enough, there is a stable PPA for FF4 should I wish to upgrade from 3.6.x, and it'll get security patches for just short of another two years.
"Calm down people" is all I can suggest. If you really don't like it now stick with 10.04/10.10 or move to something else. When the next LTS release arrives (12.04) I'll give that a go from a live USB drive and if I don't like it I've still got a security-update supported setup to use and 12 months to pick an alternative to Ubuntu12.04 before that goes away.
It would need to be quite dynamic like that, but the current infrastructure isn't up to that sort of immediate feedback at the moment. To be useful the system would need to consider the wider grid rather than each stretch of road individually, and it couldn't be truly dynamic (dimmer people would get very confused and those trying to get out of a ticket would try claim the system was showing something else at the time) so some research would need to go into the optimum setup in terms of keeping the system stable, understandable, and auditable/accountable.
It might be worth checking your plugin options too is you use AdBlock, NoScript or similar - they may be inadvertently blocking some script in the site that is incorrectly identified (by some heuristic, or a blacklist updated by user intervention) as an unwanted bit of code.
Some sites deliberately mix their general script detection in with the ad servicing to discourage people from browsing their sits with adverts blocked, though I doubt this is the case with openstreetmap as it doesn't seem to have any ads at all at the moment.
This argument is essentially "it isn't safe for me to drive at 55 when everyone else is moving at 75". It isn't unsafe because of the law or because the specified maximum speed is 55, it is unsafe because everyone else is breaking the law. That argument can be countered with the standard school teacher response of "if everyone else put their head in fire, would you?". If everyone is breaking the law then by all means charge everyone for it. If you think the speed limit set by law is wrong campaign to have it lifter rather than just ignoring it and breaking the law. The speed cameras would not be "cash cows" if people didn't routinely ignore the speed limits.
Speed limits are not only set for safety in some places. Studies have shown that most road systems, once above a certain % of their carrying capacity, are most efficient (both in terms of average journey time for those taking part in the system and in terms of fuel efficiency) when the maximum speed is set to a value most people would find surprisingly low. This is mainly due to the fact it means people keep a more constant speed, with far less accelerating simply because the speed limit is higher then having to slow down again at the next obstruction (lights, slower moving traffic ahead, turning off into a slower road). Without this constant speed variation in individual vehicles less fuel would be used and there would be less "bunching" which can cause havoc with road system efficiency (meaning average journey times, and fuel waste, rise). Of course for optimum efficiency the speed limit would need to be more dynamic than the current fixed limits, rising on straight stretches at times when the roads are clear to traffic can move freely and safely+efficiently at a higher pace - but would require significant infrastructure investment to implement so may be a pretty bad optimisation in short/medium term.
The debate about speed cameras in high speed areas is interesting. If they were just there for the safety aspect then there may be a case for their being less of them, but there is also a case for speed limits being lower for efficiency reasons in many areas and there would be no way to implement that without the cameras to keep an eye on people.
One place where I would like to see *more* cameras (perhaps moving some of those that are currently monitoring high-speed areas?) is in slower zones where the issue is very much safety. I expect that cameras policing the 15 and 20mph zones near schools, parks, and other quiet residential areas would draw in less cash but would make more of an impact in terms of lives saved and injuries lessened. I've often seen people shoot past a local school here at far more than the posted (but not enforced, aside from the very occasional bobby with a radar gun) 20mph limit - when I had my motor bike I would sometimes be in that flow of traffic and be getting bibbed by the idiot behind me because I was moving at 20ish rather than the 30+ he thought more appropriate. The really irritating thing is that some of the people speeding were speeding away after dropping off their kids at the school (I'm sure they'd complain pretty indignantly if one day their snotty little sprog was skittled by a car or bike that was moving faster than the limit). An efficiency issue would be addressed by this too: all to often you see people putting their foot down at one end of a short street only to slam on the breaks at the other end before they turn, which is probably more wasteful than pushing up from 55 to 75 and back down again.
Well, he could hardly say "people understand the product well enough to make a well informed choice not to buy the thing" or "people just don't care as much about this gimmick as we hoped they would" could he? One of which is closer to the truth in most cases.
Protip: Clippy and the Spot the dog are the same thing.
Protip: Adding "Protip" doesn't make something true.
Protip: Stating that adding "Protip" doesn't make something true doesn't make that something false if it is in fact true. They were the same thing, or at least belonged to the same family of libraries. Both were implemented by some version of the "agents" library - different versions of the library perhaps but still the same line. The search helper only has a limited range of things it can offer as hints not because it is something different but because it is the calling programs responsibility to key the help prompts, not the libraries.
One of our clients had us implement the damn things as a hint system in an app because they liked the parrot character (http://rainbow.arch.scriptmania.com/merlin/msagents_download.html). Fools...
My way on new installs is to get a virtual XP going and use it to search for files on networks - case closed.
A less heavy alternative is to install a 3rd party app like Agent Ransack (free, bit not OSS), then you don't have a whole VM running (or how-ever the vXP is implemented) just to be able to search properly.
Searching for filenames is not the be all and end all of searching though. I often want to search for specific content.
Even the XP search doesn't do this right without some kicking (by default it won't find anything in HTML or XML files for instance and you have to use a registry tweak to get it to treat them as files it can search for text in), and there are other annoyances that I forget right now. Basically if you are searching by name or by file data or size it works fine but searching for content is broken even with full indexing turned on.
The first thing I do on any new Windows setup is install Agent Ransack (http://www.mythicsoft.com/page.aspx?type=agentransack&page=home) which unlike the XP+ file search actually works and also has extra features like boolean search (files containing X and Y, and so on) that I find very useful from time to time.
and even Windows computers running Safari 5 are being watched
And here was me about to install it for testing alongside other browsers. Consider that plan cancelled. I'll just have to assume that things work in Safari if they work everywhere else and if they don't, well, tough.
I believe there is a set of automated build loops for the kernel that are used to trap compile time and other non hardware specific errors, but effects such as this one are far harder to automatically detect.
That may lend credence to the theory I guessed at in another post above: a bug may have been introduced that stops certain power management options being detected and enabled.
Pah! Why on earth would a kernel be doing more work? That means they're doing something more inefficiently, that's a regression.
I didn't say it wasn't a regression. I said it probably wasn't an intentional regression (i.e. a change made knowing it would cause incompatibility or inefficiency with things that were compatible/efficient before).
It may not even be that more work is being done at all: it could be a problem somewhere in hardware initialisation code that is causing something to not have its power saving mode activated where it previously did: it is not at all inconceivable for a bit of code that usually runs once at boot to have a long lasting effect like this.
Assuming the 18% extra power consumption they found when the machine was idling is representative of what many users will see, than I expect non-power users will notice. If you have a laptop with only two hours of battery life (a netbook from a year or two ago like the AA1 with its standard battery for instance, or many fullsize laptops from a bit earlier) that is a full 21 minutes reduced run time on battery (36 minutes if you count the top, 30%, figure). People using their portable device on a long commute or other journey may well notice a difference, especially if they've not had chance to make sure it is fully charged before leaving.
This is presumably not an intentional regression though, more likely just some new/updated code that is causing the CPU to be more busy when the machine is effectively idle than it was previously. It isn't like someone said "hey, Linus, do you mind if I make the kernel eat more power?"!
A lot of people are saying they don't know why the iDevices are recording as much info and why it is kept for so long. There is an obvious purpose that many seem to have missed: warranty claim checking. It is known that the devices have ways to detect certain warranty voiding events (like having a small piece of litmus paper so they can tell if the device has been significantly exposed to liquid), so why not use this info to. While in most cases it will tell them nothing it could be used to check for obviously false claims. You were at home and it suddenly stopped working you say? Oddly enough your phone's records say you were up a mountain at that time...
Slashdot Mirror
Last time I priced up CPUs (admittedly a while ago, and I was looking at mid-to-highish-range rather than bleeding edge) Intel's offerings at the same price also won over slightly more expensive AMDs in benchmark races, but equivalent motherboard were so expensive that it tipped things back in favour of the AMD unit for my needs.
I took that $100 and put it towards a decent video card instead.
Processors really aren't that much of a big deal anymore...
It is surprising how many people do not understand this. Rather than getting a bleeding edge CPU (and mobo+RAM to go with it) you will often see much more benefit saving some on the CPU-et-al and spending a but more on the graphics card (if you are a gamer and haven't already specified something silly in that respect) or getting better drives (a reasonable SSD won't set you back too much, and can make make much more useful difference to everyday use than spending the extra on a bleeding-edge CPU).
They all use up to 125W. They will use something approaching that much if every core and other part (the memory controller and cache banks) are at full tilt. Under more normal loading conditions you will see considerable difference between some chips that have the same power requirements on paper.
But do many games take advantage of many cores? I have a four core box and when I've checked while running a game (CPU monitor running on the other screen) one or two cores are pretty busy and the other two are not working all that much (I assume some of the threads have their affinities set so they don't bounce between cores, otherwise I'd expect the load to look a bit more even). I usually recommend people go for two fast cores rather than 4 slower ones (or 4 faster ones rather than 6/8 slower ones if they are set on spending more on the CPU instead of following my other advice and putting the extra cash into the graphics card and/or considering an SSD drive) for fast home machines. Outside of artificial tests, scientific number-crunching, bitcoin, and server-side use (where things tend to take better advantage of multi-cpu/core arrangements), have you ever seen a 6-core CPU under full load?
Yep. See also The Sam Vimes "Boots" Theory of Economic Injustice: http://wiki.lspace.org/wiki/Sam_Vimes_Theory_of_Economic_Injustice
I am still using XP Pro. SP3 from 2002 and Debian from 2005.
I hope those Debian installs are more up-to-date than that as no named release point from that year is still supported for security patches (the longer support cycle is why I've slowly moved from Debian to Ubuntu's LTS releases since Debian's release cycle sped up and the support windows shortened with them). But your point is perfectly valid: I run XPsp3 or Ubuntu 10.04 on my desktop/netbook and for the most part Ubuntu 10.04 or Debian/Stable server-side. Work is a little different as we are a Microsoft shop for what earns our money, but a number of our servers are Debian/OldStable (Lenny) with more recently installed/updated ones being Ubuntu 10.04.
Ubuntu need to decide whether they are "the Linux for the rest of us" or "the bleeding edge".
My interpretation is that they are trying to do both, with the stable and long-term supported versions every two years and three releases between that are more experimental. It makes sense to make major changes now, while they've still got another release to show improvement/refinement in (11.10) before heading towards the next LTS release (12.04). It is a pity that this isn't made clear on any of the promotional materials (though that could also mean that I am just plain wrong in my interpretation!).
I'm sticking with 10.04 for now. It'll still get security updates until early 2013, it is stable, works well for my needs, has uptodate enough packages and good stable PPAs available for the main things you might want to update (i.e. if you want to move to FF4 for the extra speed enhancements over 3.6.x - I might do that soon). I'll be keeping an eye on how things go over the next 11 months (until the next LTS release, 12.04) so a month or few after that is released I can try it in an informed manner and decide whether to upgrade or move on to another distro before the 10.04 support period expires.
Maybe I'm not being demanding enough of Ubuntu, but I really don't see why people are getting quite so worked up at this point.
I need a predictable, stable system
Which is why I'm sticking with 10.04. It'll still get security updates until early 2013 and possibly beyond (depending on what the different desktop/server support windows turn out to mean in practise), it is stable, works well for my needs, has uptodate enough packages and good stable PPAs available for the main things you might want to update (i.e. if you want to move to FF4 for the extra speed enhancements over 3.6.x - I might do that soon).
Predictable and stable is what LTS releases are intended for. I've assumed that the releases in between are test-beds for people who need/want the bleeding edge and don't mind occasional minor scalping due to their position as pioneers. I'll be keeping an eye on how things go over the next 11 months (until the next LTS release, 12.04, so a month or few after that is release I can try it in an informed manner and decide whether to upgrade or move on to something else before the 10.04 support period expires.
The next release of Unity will likely fix the major problems.
Or not. But we'll find out in 12.04. I'm keeping my netbook at 10.04 at the moment (until I got the new one recently I was still running 9.10 which worked just fine for regular use of OO.o Write/Sheet and Firefox & Chromium (the four apps I use most on my portable device, others I use regularly being a syntax highlighting text editor, VLC for video playback, the default media player for MP3s, and a WinXP VM in VitualBox for basic testing of web related stuff in IE).
Maybe I'm not that demanding a user, but 10.04 seems perfectly fine and almost certainly will be for another year or more. The default packages are uptodate enough, there is a stable PPA for FF4 should I wish to upgrade from 3.6.x, and it'll get security patches for just short of another two years.
"Calm down people" is all I can suggest. If you really don't like it now stick with 10.04/10.10 or move to something else. When the next LTS release arrives (12.04) I'll give that a go from a live USB drive and if I don't like it I've still got a security-update supported setup to use and 12 months to pick an alternative to Ubuntu12.04 before that goes away.
It would need to be quite dynamic like that, but the current infrastructure isn't up to that sort of immediate feedback at the moment. To be useful the system would need to consider the wider grid rather than each stretch of road individually, and it couldn't be truly dynamic (dimmer people would get very confused and those trying to get out of a ticket would try claim the system was showing something else at the time) so some research would need to go into the optimum setup in terms of keeping the system stable, understandable, and auditable/accountable.
It might be worth checking your plugin options too is you use AdBlock, NoScript or similar - they may be inadvertently blocking some script in the site that is incorrectly identified (by some heuristic, or a blacklist updated by user intervention) as an unwanted bit of code.
Some sites deliberately mix their general script detection in with the ad servicing to discourage people from browsing their sits with adverts blocked, though I doubt this is the case with openstreetmap as it doesn't seem to have any ads at all at the moment.
This argument is essentially "it isn't safe for me to drive at 55 when everyone else is moving at 75". It isn't unsafe because of the law or because the specified maximum speed is 55, it is unsafe because everyone else is breaking the law. That argument can be countered with the standard school teacher response of "if everyone else put their head in fire, would you?". If everyone is breaking the law then by all means charge everyone for it. If you think the speed limit set by law is wrong campaign to have it lifter rather than just ignoring it and breaking the law. The speed cameras would not be "cash cows" if people didn't routinely ignore the speed limits.
Speed limits are not only set for safety in some places. Studies have shown that most road systems, once above a certain % of their carrying capacity, are most efficient (both in terms of average journey time for those taking part in the system and in terms of fuel efficiency) when the maximum speed is set to a value most people would find surprisingly low. This is mainly due to the fact it means people keep a more constant speed, with far less accelerating simply because the speed limit is higher then having to slow down again at the next obstruction (lights, slower moving traffic ahead, turning off into a slower road). Without this constant speed variation in individual vehicles less fuel would be used and there would be less "bunching" which can cause havoc with road system efficiency (meaning average journey times, and fuel waste, rise). Of course for optimum efficiency the speed limit would need to be more dynamic than the current fixed limits, rising on straight stretches at times when the roads are clear to traffic can move freely and safely+efficiently at a higher pace - but would require significant infrastructure investment to implement so may be a pretty bad optimisation in short/medium term.
The debate about speed cameras in high speed areas is interesting. If they were just there for the safety aspect then there may be a case for their being less of them, but there is also a case for speed limits being lower for efficiency reasons in many areas and there would be no way to implement that without the cameras to keep an eye on people.
One place where I would like to see *more* cameras (perhaps moving some of those that are currently monitoring high-speed areas?) is in slower zones where the issue is very much safety. I expect that cameras policing the 15 and 20mph zones near schools, parks, and other quiet residential areas would draw in less cash but would make more of an impact in terms of lives saved and injuries lessened. I've often seen people shoot past a local school here at far more than the posted (but not enforced, aside from the very occasional bobby with a radar gun) 20mph limit - when I had my motor bike I would sometimes be in that flow of traffic and be getting bibbed by the idiot behind me because I was moving at 20ish rather than the 30+ he thought more appropriate. The really irritating thing is that some of the people speeding were speeding away after dropping off their kids at the school (I'm sure they'd complain pretty indignantly if one day their snotty little sprog was skittled by a car or bike that was moving faster than the limit). An efficiency issue would be addressed by this too: all to often you see people putting their foot down at one end of a short street only to slam on the breaks at the other end before they turn, which is probably more wasteful than pushing up from 55 to 75 and back down again.
Well, he could hardly say "people understand the product well enough to make a well informed choice not to buy the thing" or "people just don't care as much about this gimmick as we hoped they would" could he? One of which is closer to the truth in most cases.
Protip: Clippy and the Spot the dog are the same thing.
Protip: Adding "Protip" doesn't make something true.
Protip: Stating that adding "Protip" doesn't make something true doesn't make that something false if it is in fact true. They were the same thing, or at least belonged to the same family of libraries. Both were implemented by some version of the "agents" library - different versions of the library perhaps but still the same line. The search helper only has a limited range of things it can offer as hints not because it is something different but because it is the calling programs responsibility to key the help prompts, not the libraries.
One of our clients had us implement the damn things as a hint system in an app because they liked the parrot character (http://rainbow.arch.scriptmania.com/merlin/msagents_download.html). Fools...
My way on new installs is to get a virtual XP going and use it to search for files on networks - case closed.
A less heavy alternative is to install a 3rd party app like Agent Ransack (free, bit not OSS), then you don't have a whole VM running (or how-ever the vXP is implemented) just to be able to search properly.
Searching for filenames is not the be all and end all of searching though. I often want to search for specific content.
Even the XP search doesn't do this right without some kicking (by default it won't find anything in HTML or XML files for instance and you have to use a registry tweak to get it to treat them as files it can search for text in), and there are other annoyances that I forget right now. Basically if you are searching by name or by file data or size it works fine but searching for content is broken even with full indexing turned on.
The first thing I do on any new Windows setup is install Agent Ransack (http://www.mythicsoft.com/page.aspx?type=agentransack&page=home) which unlike the XP+ file search actually works and also has extra features like boolean search (files containing X and Y, and so on) that I find very useful from time to time.
Well, Mr Nonymous didn't specify any of that information either, so it wouldn't serve any use for comparative purposes.
and even Windows computers running Safari 5 are being watched
And here was me about to install it for testing alongside other browsers. Consider that plan cancelled. I'll just have to assume that things work in Safari if they work everywhere else and if they don't, well, tough.
I believe there is a set of automated build loops for the kernel that are used to trap compile time and other non hardware specific errors, but effects such as this one are far harder to automatically detect.
That may lend credence to the theory I guessed at in another post above: a bug may have been introduced that stops certain power management options being detected and enabled.
Pah! Why on earth would a kernel be doing more work? That means they're doing something more inefficiently, that's a regression.
I didn't say it wasn't a regression. I said it probably wasn't an intentional regression (i.e. a change made knowing it would cause incompatibility or inefficiency with things that were compatible/efficient before).
It may not even be that more work is being done at all: it could be a problem somewhere in hardware initialisation code that is causing something to not have its power saving mode activated where it previously did: it is not at all inconceivable for a bit of code that usually runs once at boot to have a long lasting effect like this.
And I've just noticed the pun. Somebody mod me "too slow on the uptake".
Assuming the 18% extra power consumption they found when the machine was idling is representative of what many users will see, than I expect non-power users will notice. If you have a laptop with only two hours of battery life (a netbook from a year or two ago like the AA1 with its standard battery for instance, or many fullsize laptops from a bit earlier) that is a full 21 minutes reduced run time on battery (36 minutes if you count the top, 30%, figure). People using their portable device on a long commute or other journey may well notice a difference, especially if they've not had chance to make sure it is fully charged before leaving.
This is presumably not an intentional regression though, more likely just some new/updated code that is causing the CPU to be more busy when the machine is effectively idle than it was previously. It isn't like someone said "hey, Linus, do you mind if I make the kernel eat more power?"!
A lot of people are saying they don't know why the iDevices are recording as much info and why it is kept for so long. There is an obvious purpose that many seem to have missed: warranty claim checking. It is known that the devices have ways to detect certain warranty voiding events (like having a small piece of litmus paper so they can tell if the device has been significantly exposed to liquid), so why not use this info to. While in most cases it will tell them nothing it could be used to check for obviously false claims. You were at home and it suddenly stopped working you say? Oddly enough your phone's records say you were up a mountain at that time...