Although it also depends on the subject. In CS it is common to publish work three times, firstly at a workshop, then at a conference, and finally at a journal. Each level of the pyramid is happy as long as the work hasn't been that high before. Even before any of these it is common to release a tech report or an eprint to "get a flag in the ground". Part of the difference in culture comes from the turn around time on research.
The ethical issues are still the same though. Most "blind" review is not blind after a little googling, although preprints of the work do make that a little easier. Work in CS doesn't have such a binary quality control. There is an ordering between the different types of publications, but it isn't as important as the quality of the venue. I can think of some really prestigious workshops with 60:1 acceptance ratios against some pretty crappy journals that are 3:1.
Well, no actually you can't. If you've ever spent any time going through the 1000 page Intel Optimisation Guide for the x86 then you would know that they don't spell out all of the trade offs explicitly. They describe enough to point you in the right direction but they keep a lot back. Partially because the behaviour of these chips in certain usage patterns isn't even defined by the design - it's a side-effect of several other parts of the chip design interacting. So the best that you can do is suck it and see - and in general it changes not between major ISA revisions but on individual models.
Now, if you're Intel then you have the time and the money to work out exactly how to exploit these tradeoffs to schedule threads effectively. But you don't want to give that away for free. From the (very scanty) marketing bullshit that was linked to, it would appear that they've appear an Intel-specific threading library (probably with a POSIX interface). Separate to this is a profiling tool and a multi-threaded debugger (the latter of which is non-trivial). While any debugger will let you skip across threads allowing you do it in a deterministic manner to look for race hazards is much harder.
The analysis tools sound nice, but the bolton library is nothing special. It's purely to win a few synthetic benchmarks and gain some marketshare for ICC and therefore more "Made for Intel" applications in the market. I'm cynical about the library because what is broken about the threading model in C/C++ would take more than a library to fix. It would require redesigning the language down to the ground and choosing a different set of control constructs.
So finally, when you claim that it's because Intel has "better" coders. You don't know what you're talking about. I know a few guys who code GCC for a living, and they are grade A coders. It is because Intel has moved the goalposts. It's not so much that GCC targets multiple architectures, it's that they are trying to stick to (relatively) standard C where-as Intel is willing to redefine where the semantic gap sits if they can squeeze out a little more performance. Their attitude is screw portable code - talking across different compiler vendors here, rather than chip vendors. If what they need to squeeze into their compiler is no longer "C" strictly speaking, then they don't care. The gcc guys do.
Ah yes, and portable code can be a smaller window than you expect. That weighty 1000 page Intel document is sitting comfortable next to the AMD equivalent, which differs in suprising places.
if a simple search for my real name can present the searcher with a look into my private life because some "friend" feels it necessary to catalogue the names of everyone in their photos.
You and the other poster seem to think that you're arguing about whether or not people should post photos online. But reading your comments as a third-party it seems that you both disagree over what a photo actually is. When you're hanging out with your friends in a private environment you have some expectation of privacy. As soon as somebody takes a photograph they have violated that privacy straight away. Before flickr people assume that this violation wasn't a big deal - only people in your group of friends would see the photo anyway.
The reason that the other poster feels it isn't a BFG (I assume) is that once that photo exists your privacy has been violated anyway. Flickr is just the icing on the cake. From what you've written I assume that you don't see the original violation, and consider the posting of the image to be the loss of privacy. There are arguments to be made on both sides here, but my gut reaction would be with the other poster - any photograph is a violation of privacy, although mostly we would allow the violation.
Consider photos taken with an ex-lover. When they become an ex what are the photos? They have become a violation of privacy long after the fact. People do instinctively understand this on some level and hence many people are uneasy being photographed in certain situations, whether making love or just hitting a bong at a party.
The real problem here is the vampire that wrote the paper. He has seen a way to expand his market so that he can take a pound of flesh from every "privacy" dispute over online photographs that contain people (ie 90% of them). The easiest way to fix this problem is not to create an online registry - first thing that we do is we shoot all of the lawyers, then we can argue about privacy.
Why didn't they test in a mastering studio? Earbuds? Sheesh, seriously. Not only are they testing for something that is at or beyond the limits of human perception, but they are also degrading the signal confounding the test.
Gosh that's a hard question! I always plug my ipod into a mastering studio input when I'm using it. It's quite hard to carry around, find space for on the bus etc... but thank god I don't earbud!
Errr, do you reckon it was a "real world" test? Checking if it is worth it for people who but music off iTunes to play on their ipods. Knowing the effect of the codec in an ideal environment doesn't really do it, does it?
Seeing as people are posting weird things they've found as replies here. Take a look at this.
It looks pretty unremarkable but move one step east and see what happens? I guess that either their stitching chokes or the gps on the car flaked out while they drove through the carpark that you see just off the road. It's a bit weird.
Given that it is an example of an embarrasingly parallel algorithm, like ray-tracing. No, it is not the best example. The difficulty in parallel programming is trying to divide up the work when your sub-problems are not entirely independent of each other.
It's odd, from the insightful mod and the people answering you, it would appear that you refuted his comment. But he said that we read comments sequentially. Are you saying that you skim from some letters in the words in this comment onto another comment and then back? Even if the actual reading is a parallel process, it is sequentialised when we turn the squiggles into concepts. Much like your later argument about addition, sure it's parallel at the bit-level but it is still sequential when you try and combine those word-level operations together.
I think you missed my point. It wasn't contained in a single line, but in the post as a whole. The current ISP situation in the UK is fucked. My ISP is above average, but they don't make it into the "good" category. I was just pointing out that the headline 1Gb cap is extreme and 100x worse than what is on offer.
So yes, they are not much better, but with a higher cap. Now that I've replied to each of the replies to my message it is definitely time to accept that I'm not getting any more work done tonight and go home...
This is really sad to hear. We had their service a couple of years ago and it rocked. They were tolerent of people running servers, had bandwidth coming out of their arse, uncapped lines and (not completely crap) Tv. The only reason we switched to DSL was because we moved flat into a road with a sunken level pavement - basement flat "under" the road with stairs leading down. The telewest installation guys took one look at it and told us they wouldn't touch it with a bargepole.
No, neither of those. It's an old well established ISP, it's just that after begin with two ISPs in a row that started with great customer service, then got really popular and then went down the tubes (Plusnet was one of them) I'm reluctant to advertise for these guys.
Yeah, the submitter was talking about the uk market. I don't know anything about caps internationally. Surely there must be a lot of fibre running into Oz by now. Isn't that limit just history now?
All UK ISP's who are on wholesale packages with BT do this - they are charged by BT per gig that goes upstream onto BT's network. Their entire business model is as a middleman who buys a bundle of bandwidth and increases the value by choppng it up into smaller pieces with more of a margin. This is no secret as the submission seems to think, as any UK ISP what a "contended" service is and they will explain it.
As you point out this model breaks when people really do want "unlimited" bandwidth for things like IPTV. The real crime that I can see is to allow ISPs to advertise an "unlimited" service that clearly has limits. Answering a few points that earlier posters have made; if it were made illegal to advertise a crippled service as unlimited then the market would fragment into low bandwidth cheap package deals, and premium connections for those that would pay to get higher bandwidth. This would be a good thing for everyone, and the only reason that it hasn't happened is that Ofcom hasn't smacked down the ISPs for lying to their customers about what they are selling.
Another poster (below) mentions the dark fibre of the dot-com era. It's out there, and it's being used. Telewest are not on a wholesale bundle deal with BT. They peer at Telehouse and their network has thousands of miles of dark fibre should they require more bandwidth. As a result they operate completely differently to the DSL ISPs and don't throttle their traffic. You buy a 10Mb connection, you get a 10Mb connection. The only contention is from other subscribers on your segment; ie Telewest customers in the same street.
As far as the 1Gb caps mentioned in the submission go. They are an extreme, aimed at customers who don't know the market. My ISP (who I won't name as their customer service is currently excellent) has a "fair usage policy" like most. Their unofficial caps start when you regularly download 100s of Gigs a month. Below that limit I can max out my 8Mb line without trouble, and bittorrent is excellent.
But he didn't mention eavesdropping, he mentioned man in the middle attacks. Just like a quantum link this is vulnerable to man in the middle attacks when used without a separate authenticated channel.
(What I'm about to describe is true of the 7800GTX and below. I haven't got my hands on an 8800 to play with and I think that the new memory architecture fixes this problem).
On a graphics card you can't implement a scatter operation effectively. So in a sieving operation a scatter might do something like: for p in primes
for i is 1 to n
sieve[i*p] += log(p)
This is simplifying greatly, but the point is that the memory locations that you touch in the sieving loop are not uniform (linear) outputs, and scatter through the address space. This is a fundamental part of the sieving process, and to implement this you either need real scatter on your architecture or you fake it with sorting / routing. If you take the faking it approach then you lose a factor of 10000x against a CPU on a reasonable window size and set of primes.
Basically, this problem just won't fit in the limited problem class of stream programs that can execute on a GPU.
Or you could say that a bunch of companies who buy bandwidth in bulk, and sell it in small pieces can't cut their margins too tight without going out of business. Either way, it's a matter of perception.
However, video p2p services don't have to suffer this way. The service provider is being shit by not preferring local peers over distant peers. If they recoded their applications to take explicit measures to route the majority of traffic within an ISP's address block then it would escape traffic shaping and throttling which happen at the interface to the network.
So the ISPs wouldn't lose money, and the punters could watch their porn. So whose fault is it now?
The whole point of a hierarchical naming scheme was to spread the load around and remove a centralised point from the network. At the moment 99% of websites are.com and the extension has become meaningless. If URLs were actually split into domains that made sense it would be easier for people to remember web addresses...
Yes it can be quite annoying. I can understand the flipside of the coin - there is a certain limit to how long a page can be before it becomes unreadable. But I agree that material shouldn't be expunged, when pages are oversized they should split into subpages so that the wiki grows in breadth as well as depth.
I didn't know about the nocache tags, that is pretty offensive. Given that wiki is the first (reasonably successful) stab at a hitchhikers guide (yes, yes h2g2 was much earlier but less successful) it seems almost criminal to intentionally destory the early drafts. They could be priceless to future historians, and disk space is cheap.
Slashdot Mirror
Although it also depends on the subject. In CS it is common to publish work three times, firstly at a workshop, then at a conference, and finally at a journal. Each level of the pyramid is happy as long as the work hasn't been that high before. Even before any of these it is common to release a tech report or an eprint to "get a flag in the ground". Part of the difference in culture comes from the turn around time on research.
The ethical issues are still the same though. Most "blind" review is not blind after a little googling, although preprints of the work do make that a little easier. Work in CS doesn't have such a binary quality control. There is an ordering between the different types of publications, but it isn't as important as the quality of the venue. I can think of some really prestigious workshops with 60:1 acceptance ratios against some pretty crappy journals that are 3:1.
Gee, that makes what you have to say insightful, I guess I'll read the rest of your zzzzzzzzzzzzzzzzzzzzz
Well, no actually you can't. If you've ever spent any time going through the 1000 page Intel Optimisation Guide for the x86 then you would know that they don't spell out all of the trade offs explicitly. They describe enough to point you in the right direction but they keep a lot back. Partially because the behaviour of these chips in certain usage patterns isn't even defined by the design - it's a side-effect of several other parts of the chip design interacting. So the best that you can do is suck it and see - and in general it changes not between major ISA revisions but on individual models.
Now, if you're Intel then you have the time and the money to work out exactly how to exploit these tradeoffs to schedule threads effectively. But you don't want to give that away for free. From the (very scanty) marketing bullshit that was linked to, it would appear that they've appear an Intel-specific threading library (probably with a POSIX interface). Separate to this is a profiling tool and a multi-threaded debugger (the latter of which is non-trivial). While any debugger will let you skip across threads allowing you do it in a deterministic manner to look for race hazards is much harder.
The analysis tools sound nice, but the bolton library is nothing special. It's purely to win a few synthetic benchmarks and gain some marketshare for ICC and therefore more "Made for Intel" applications in the market. I'm cynical about the library because what is broken about the threading model in C/C++ would take more than a library to fix. It would require redesigning the language down to the ground and choosing a different set of control constructs.
So finally, when you claim that it's because Intel has "better" coders. You don't know what you're talking about. I know a few guys who code GCC for a living, and they are grade A coders. It is because Intel has moved the goalposts. It's not so much that GCC targets multiple architectures, it's that they are trying to stick to (relatively) standard C where-as Intel is willing to redefine where the semantic gap sits if they can squeeze out a little more performance. Their attitude is screw portable code - talking across different compiler vendors here, rather than chip vendors. If what they need to squeeze into their compiler is no longer "C" strictly speaking, then they don't care. The gcc guys do.
Ah yes, and portable code can be a smaller window than you expect. That weighty 1000 page Intel document is sitting comfortable next to the AMD equivalent, which differs in suprising places.
You and the other poster seem to think that you're arguing about whether or not people should post photos online. But reading your comments as a third-party it seems that you both disagree over what a photo actually is. When you're hanging out with your friends in a private environment you have some expectation of privacy. As soon as somebody takes a photograph they have violated that privacy straight away. Before flickr people assume that this violation wasn't a big deal - only people in your group of friends would see the photo anyway.
The reason that the other poster feels it isn't a BFG (I assume) is that once that photo exists your privacy has been violated anyway. Flickr is just the icing on the cake. From what you've written I assume that you don't see the original violation, and consider the posting of the image to be the loss of privacy. There are arguments to be made on both sides here, but my gut reaction would be with the other poster - any photograph is a violation of privacy, although mostly we would allow the violation.
Consider photos taken with an ex-lover. When they become an ex what are the photos? They have become a violation of privacy long after the fact. People do instinctively understand this on some level and hence many people are uneasy being photographed in certain situations, whether making love or just hitting a bong at a party.
The real problem here is the vampire that wrote the paper. He has seen a way to expand his market so that he can take a pound of flesh from every "privacy" dispute over online photographs that contain people (ie 90% of them). The easiest way to fix this problem is not to create an online registry - first thing that we do is we shoot all of the lawyers, then we can argue about privacy.
Gosh that's a hard question! I always plug my ipod into a mastering studio input when I'm using it. It's quite hard to carry around, find space for on the bus etc
Errr, do you reckon it was a "real world" test? Checking if it is worth it for people who but music off iTunes to play on their ipods. Knowing the effect of the codec in an ideal environment doesn't really do it, does it?
To be fair, with windows you don't have to twist the lock... a strong fart on the way past would do it.
Seeing as people are posting weird things they've found as replies here. Take a look at this.
It looks pretty unremarkable but move one step east and see what happens? I guess that either their stitching chokes or the gps on the car flaked out while they drove through the carpark that you see just off the road. It's a bit weird.
Given that it is an example of an embarrasingly parallel algorithm, like ray-tracing. No, it is not the best example. The difficulty in parallel programming is trying to divide up the work when your sub-problems are not entirely independent of each other.
Cough... Occam.
It's odd, from the insightful mod and the people answering you, it would appear that you refuted his comment. But he said that we read comments sequentially. Are you saying that you skim from some letters in the words in this comment onto another comment and then back? Even if the actual reading is a parallel process, it is sequentialised when we turn the squiggles into concepts. Much like your later argument about addition, sure it's parallel at the bit-level but it is still sequential when you try and combine those word-level operations together.
Ok, but to be fair you must have worked in Kent...
Awesome. You replied to his comment, without actually referring to anything that he said or inferred - and got modded up for it.
You sir, have won the greater slashdot irony game. The moderator of your comment gets a close second.
I think you missed my point. It wasn't contained in a single line, but in the post as a whole. The current ISP situation in the UK is fucked. My ISP is above average, but they don't make it into the "good" category. I was just pointing out that the headline 1Gb cap is extreme and 100x worse than what is on offer.
So yes, they are not much better, but with a higher cap. Now that I've replied to each of the replies to my message it is definitely time to accept that I'm not getting any more work done tonight and go home...
This is really sad to hear. We had their service a couple of years ago and it rocked. They were tolerent of people running servers, had bandwidth coming out of their arse, uncapped lines and (not completely crap) Tv. The only reason we switched to DSL was because we moved flat into a road with a sunken level pavement - basement flat "under" the road with stairs leading down. The telewest installation guys took one look at it and told us they wouldn't touch it with a bargepole.
No, neither of those. It's an old well established ISP, it's just that after begin with two ISPs in a row that started with great customer service, then got really popular and then went down the tubes (Plusnet was one of them) I'm reluctant to advertise for these guys.
Yeah, the submitter was talking about the uk market. I don't know anything about caps internationally. Surely there must be a lot of fibre running into Oz by now. Isn't that limit just history now?
All UK ISP's who are on wholesale packages with BT do this - they are charged by BT per gig that goes upstream onto BT's network. Their entire business model is as a middleman who buys a bundle of bandwidth and increases the value by choppng it up into smaller pieces with more of a margin. This is no secret as the submission seems to think, as any UK ISP what a "contended" service is and they will explain it.
As you point out this model breaks when people really do want "unlimited" bandwidth for things like IPTV. The real crime that I can see is to allow ISPs to advertise an "unlimited" service that clearly has limits. Answering a few points that earlier posters have made; if it were made illegal to advertise a crippled service as unlimited then the market would fragment into low bandwidth cheap package deals, and premium connections for those that would pay to get higher bandwidth. This would be a good thing for everyone, and the only reason that it hasn't happened is that Ofcom hasn't smacked down the ISPs for lying to their customers about what they are selling.
Another poster (below) mentions the dark fibre of the dot-com era. It's out there, and it's being used. Telewest are not on a wholesale bundle deal with BT. They peer at Telehouse and their network has thousands of miles of dark fibre should they require more bandwidth. As a result they operate completely differently to the DSL ISPs and don't throttle their traffic. You buy a 10Mb connection, you get a 10Mb connection. The only contention is from other subscribers on your segment; ie Telewest customers in the same street.
As far as the 1Gb caps mentioned in the submission go. They are an extreme, aimed at customers who don't know the market. My ISP (who I won't name as their customer service is currently excellent) has a "fair usage policy" like most. Their unofficial caps start when you regularly download 100s of Gigs a month. Below that limit I can max out my 8Mb line without trouble, and bittorrent is excellent.
But he didn't mention eavesdropping, he mentioned man in the middle attacks. Just like a quantum link this is vulnerable to man in the middle attacks when used without a separate authenticated channel.
Actually with the way that their sandbox sits behind the kernel it sounds more like WINE, but in reverse...
No you can't, and no it doesn't work.
(What I'm about to describe is true of the 7800GTX and below. I haven't got my hands on an 8800 to play with and I think that the new memory architecture fixes this problem).
On a graphics card you can't implement a scatter operation effectively. So in a sieving operation a scatter might do something like:
for p in primes
for i is 1 to n
sieve[i*p] += log(p)
This is simplifying greatly, but the point is that the memory locations that you touch in the sieving loop are not uniform (linear) outputs, and scatter through the address space. This is a fundamental part of the sieving process, and to implement this you either need real scatter on your architecture or you fake it with sorting / routing. If you take the faking it approach then you lose a factor of 10000x against a CPU on a reasonable window size and set of primes.
Basically, this problem just won't fit in the limited problem class of stream programs that can execute on a GPU.
Or you could say that a bunch of companies who buy bandwidth in bulk, and sell it in small pieces can't cut their margins too tight without going out of business. Either way, it's a matter of perception.
However, video p2p services don't have to suffer this way. The service provider is being shit by not preferring local peers over distant peers. If they recoded their applications to take explicit measures to route the majority of traffic within an ISP's address block then it would escape traffic shaping and throttling which happen at the interface to the network.
So the ISPs wouldn't lose money, and the punters could watch their porn. So whose fault is it now?
You should try that comparison with Verizon customer services...
"Ok, so voice is 6 cents per byte, and SMS is 4 cents per byte. That's what it says on your calling plan, that's what we're charging you"
Why would that be a bad thing?
.com and the extension has become meaningless. If URLs were actually split into domains that made sense it would be easier for people to remember web addresses...
The whole point of a hierarchical naming scheme was to spread the load around and remove a centralised point from the network. At the moment 99% of websites are
I'm fairly sure that you missed his point.
It's not that other ways exist - it is that Second Life is shit.
This was a public service announcement, honest.
Yes it can be quite annoying. I can understand the flipside of the coin - there is a certain limit to how long a page can be before it becomes unreadable. But I agree that material shouldn't be expunged, when pages are oversized they should split into subpages so that the wiki grows in breadth as well as depth.
I didn't know about the nocache tags, that is pretty offensive. Given that wiki is the first (reasonably successful) stab at a hitchhikers guide (yes, yes h2g2 was much earlier but less successful) it seems almost criminal to intentionally destory the early drafts. They could be priceless to future historians, and disk space is cheap.