I just wrote to my MP telling him that I'm disgusted by Cameron's attempt to use a tragedy in this way. I hope that other UK readers will do the same. If you've never written to your MP before, Write to them is run by mySociety and makes it very easy.
This particular point about email actually makes sense. They've decoupled the sending and receiving accounts. You can have multiple of each and you can specify multiple sending addresses associated with a receiving account (not well documented, but a comma-separated list in the email address box) and you can have a preferred SMTP server for each IMAP/POP server, but still fall back to others when that's unavailable. Adding a new SMTP server is easy: Preferences, accounts, Outgoing mail server (SMTP) drop-down, edit server list, and then you're in the interface for creating new ones.
What it does reliable when you open the Mac Book and more or less in +/- one second remove the power cord: crash. Always.
That sounds like you may have a hardware defect. I've done that a few times by accident recently and not had a problem.
The gestures aren't too bad - you can turn them off. Since the introduction of OS X, there's been some crap that's only for in-store demos and you want to turn off to get real work done. The real issues are more subtle, such as around 10.7 they removed the much-bigger shadow on the foreground window because it 'looked ugly' and then removed most of the other visual clues that a window is foreground, measurably increasing the likelihood of users thinking that the wrong window is foreground. There are lots of things like this, where the UI has slowly regressed and, if anyone bothers to run user studies, they can clearly measure the regressions. Unfortunately, they've also improved a load of things and so there's no simple ordering of OS X versions by usability: each one introduces improvements and regressions.
As someone who does -j32 on our existing systems and sees almost linear speedup, I disagree. If you've got a decent amount of RAM, you won't hit the storage at all - everything will be in the buffer cache. Memory bandwidth is not an issue either - compiling generally has good locality of reference and so the cache works well.
SMT addresses two problems. The first is that, when a thread is completely stalled waiting for memory, you can try to run another thread. This doesn't happen that often with superscalar out-of-order cores, as you can also schedule other instructions from the same thread if there are no dependencies. The second is that a fetch granule from one thread may not be able to issue instructions to all of the execution units in a single cycle, giving you more to choose from.
The down side of SMT is that you increase contention on various resources. Cache is the biggest one, as sharing the cache between 4 or 8 threads increases contention a lot. Things like the cache and TLB depend on associativity and you can't arbitrarily scale associativity and still get single-cycle access, so you're going to hit diminishing returns quite quickly. Most of these things also increase power consumption a lot as you try to scale them.
In particular, on a modern system, register renaming takes a lot of the die area, and the more thread you have, the more rename registers you need to get the full throughput.
I'm slightly surprised at that, as last time I looked a lot of Android users were still on 2.x, which ships with Android Browser, not Chrome. I'm finding it very hard to read that graph as it doesn't provide totals for all versions of each browser, but it looks as if Android Browser still has around 5%, UC for Android (which I'd never heard of) has over 7%, so Chrome at 16% only has a bit more than 50% of the Android market share.
Not sure why this is off-topic. I started using Firefox on the phone because it was the first Android browser to offer a sane set of cookie management options (i.e. something beyond 'allow all' or 'block all', though it was restricted to this in the first couple of Android releases for some reason). With the self-destructing cookies plugin, it actually does what I want with respect to privacy. Most importantly though, it avoids a monoculture. Android has a huge market share and the idea of a bug in one browser being able to exploit the vast majority of all mobile phones is terrifying. Unfortunately, as with IE on Windows, enough apps use the Android WebView that there's a good chance that something else will run JavaScript with V8 even if you uninstall Chrome.
I just care about the Play Store. I can no longer purchase applications from the Play Store
For me, this has been a really good reason not to buy any applications from Google Play. I set up a new throw-away account for each device to download the free apps that I can only find in Google Play, but most of what I run is now from F-Droid. I've donated to a few of those projects, because I know that their apps won't randomly stop working at the whim of a third party.
The P4 got very high clock rates by having a lot of very short pipeline stages. You can push clock rates a lot higher than they are today if you split your pipelines into more stages, each of which does less. The P4 had some piplelines that were over 30 stages long. A 30-stage pipeline means that if you put an operation in at one end, you get the result 30 cycles later. If you want to saturate the execution units, then you need to find 29 more operations to start that don't depend on the results of the first (or any of the others). This is incredibly rare. It's even more rare when you consider that a typical superscalar chip has half a dozen or so separate pipelines. The P4 could have up to 140 instructions in flight at any given time. To fully saturate the pipelines, you need to have code that has 140 totally independent operations at a time. This is why Intel added Hyperthreading: there were always a lot of pipeline slots that had nothing going into them.
As to the lack of increase in clock frequency, that's due to the end of Dennard Scaling. Read the Wikipedia article for more information, but basically shrinking the die size stopped giving you a free power consumption reduction about 10 years ago. This is why Dark Silicon is the industry buzzword de jour: Moore's law still means that you can double the number of transistors on your chip from generation to generation, but you only get a small increase in the number that you can have powered at any given time.
There are two things wrong with this. The first is that no one does comparative benchmarks using Intel's compiler. Intel does, but they don't benchmark AMD chips, so you can't use them for Intel. The second is that a number of other x86 vendors have advertised features in CPUID and then either not implemented them at all, or implemented them in slow microcoded paths. Before Intel's compiler did the vendor ID checking, they were criticised for using features that were fast on their chips and very slow on others.
It's not quite so clear cut. Both vendors' chips are superscalar and so have multiple pipelines. Intel chips have around 5 independent integer pipelines, plus the AVX ones. Very few workloads will saturate them all. This was the main motivation for IBM's SMT: if you have two threads then you can get a lot closer to saturating the execution units than with one. Of course, you're still likely to suffer a bit because of contention on a few units, which is why Hyperthreading often isn't an overall performance win on Intel.
AMD implementations add more integer pipelines than any single-threaded workload is likely to be able to use, so they're more likely to get a speedup, but it's still not guaranteed if both workloads are very heavily biased towards a the same few pipelines.
As an opensource developer, do you really see someone as choosing to use another project as a problem?
Nope. Open source and proprietary alike care about contributors, not users. In off-the-shelf proprietary software, the (non-pirate) users are all contributors, because they all pay money. In open source, there are a lot of ways to contribute (code, testing, bug reports, documentation, artwork, and so on, as well as money), but users only matter inasmuch as they're real or potential contributors. Some users are negative contributors - they make a lot of demands, but give nothing back. The project is usually better off without them.
If the company is providing a significant contribution, then they may well expect something in return. Often, it's just the continued survival of the project, but if they want more then you should make sure that both parties agree on what the contribution is buying. If they're going to throw money at you, then they might reasonable assume that they've paid for some of your time to be spent on issues that matter to them. If you don't want to accept these terms, then having them use another project is probably better for both of you.
There was a vulnerability found in Xen a couple of weeks ago that allowed any PV guest running on any version of Xen released in the last 7 years, to map the whole of physical memory and tamper with the contents of any other VMs. This is, what, the fourth such exploit in Xen in the last year? Many of the others came from QEMU code, which was never intended to be used in security-critical situations. VMWare and HyperV almost certainly have similar issues, though they may not be so public about announcing them.
You may trust Amazon more than your own badly treated and underpaid IT staff, but do you also trust all of Amazon's customers more than your IT staff? Even if you're using VMs for your local deployment, the odds are that you aren't running VMs from third parties on the same physical hardware as your own VMs. If you host your stuff in someone else's cloud, then you aren't just trusting them, you're trusting either that their hypervisor is bug-free or that their other customers are all trustworthy.
I've only ever played Candy Crush on an aeroplane. They're selling it as part of the seat-back entertainment system, which must bring in a fair bit (the input latency is quite painful, so it's not a great idea, which is a bit surprising, as Mario 3 worked well in seat-back entertainment systems 20 years ago). I don't know what other devices they license their games for, but they may have a few more unexpected revenue sources.
Social Networks are the graphs of personal relationships. The term has been used in sociology for a long time. Social Media refers to advertising platforms that provide surrogate friends in an attempt to collect the most personal information possible about a person to sell to advertisers and anyone else with spare money. So, yes, please don't confuse them.
It is the MINIX from the book, if you have the third edition (published in 2006). And, if you haven't, then I suggest buying / borrowing a copy, along with Modern Operating Systems (by the same author), both of which are well worth reading. If you want to know a bit more detail about how a more mainstream (for a given value of mainstream) OS works, then the new Design and Implementation of the FreeBSD Operating System is well worth a read. If you want to know how a kernel designed for mobile devices should be designed, then Symbian OS Internals: Real-time Kernel Programming is a great reference.
And I really should have made all of those Amazon Affiliate links, but I'm too lazy...
They might hit problems if the words that they're matching on are correlated with speech patterns from ethnic groups. If you can find a set of words that black people are far more likely than white people to type into Facepalm posts then you can try claiming that it's not racial discrimination if you use them, but good luck convincing a court of that.
You don't have to pay interest, just get a no fee credit card, use it for some of your purchases and pay it off every month
People who don't do this and are proud of not using credit seem to be indicating that they're pretty poor at financial planning to me. I got a credit card as soon as I was legally able to (which had a very low credit limit, as I was a student with no real income). By the time I was earning, it had a big enough limit to cover all of my monthly expenses (though, unfortunately, I couldn't pay rent with it) and 1% cash back. By using the credit card, I get to have one more month's worth of salary in my savings account at any given time and I get 1% back on everything that I spend. I don't understand why anyone would prefer to use a debit card, which has neither of these advantages, but if I were lending money I'd look at the people who did and think that they're not so good at understanding personal finance.
Because saving money and paying off credit is basically the same thing: making sure you have left over money at the end of the month to achieve what you want/need to achieve
Nope, regular savings and paying off credit are the same things. This is why having a savings account that has a regular fixed payment increases your credit score, but one that you just dump money into periodically does not.
I think the truth is more likely closer to what others mention here: why would banks give credit to people who hardly need it, and will never ever be late in paying or transfer balances to the next month? that's where all their money comes from, sounds like a terrible customer to do business with!
It depends a lot on the kind of credit. For personal loans, the interest rates are so high that you want people who can pay it back, because you're making 10-20% in a year on the total loan. For mortgages, the cost of repossessing a house and then selling it can be very high, so you really want people who are able to pay back the loan. For credit cards, you want a mixture of people, some who will have a high throughput and rack up a load of transaction fees from the merchant and who can justify to the merchants that they'd lose business if they didn't accept cards (I put about £20K on my credit card each year, mostly business expenses, and get very good customer service - everything I've complained about has been resolved immediately and followed up with a goodwill payment to apologise), and some who will serve as long-term income sources by running up a smallish debt that they can't pay and effectively paying the interest in perpetuity.
My bank doesn't appear to even share my income with themselves. When I applied for a mortgage, one of the things that I had to prove was that I'd been paying my rent regularly for the last three months (yes, they really did only care about three months). I thought this would be easy: I'd been paying by standing order, from an account with them, so they could easily look at when the standing order was set up, and the fact that it had been paid every month for the last few years. Apparently they couldn't - I had to take three months of bank statements into the branch. It was then that I understood why subprime mortgages had been such an issue.
Just imagine the chinese coming and cashing in your debts. Your whole country would be out their houses.
The Fed would print enough dollars to pay them. There would be inflation, which would devalue the dollar. Most people would no longer be able to afford 'cheap' Chinese imports, because they'd now be very expensive and the Chinese economy would crash from having one of its two biggest markets suddenly disappear. Exports from the USA would suddenly become very cheap. Of, and all of the rich folk who had moved a load of their capital to Chinese investments would make a huge loss. Guess which of these reasons is why it hasn't happened.
Slashdot Mirror
I just wrote to my MP telling him that I'm disgusted by Cameron's attempt to use a tragedy in this way. I hope that other UK readers will do the same. If you've never written to your MP before, Write to them is run by mySociety and makes it very easy.
What it does reliable when you open the Mac Book and more or less in +/- one second remove the power cord: crash. Always.
That sounds like you may have a hardware defect. I've done that a few times by accident recently and not had a problem.
The gestures aren't too bad - you can turn them off. Since the introduction of OS X, there's been some crap that's only for in-store demos and you want to turn off to get real work done. The real issues are more subtle, such as around 10.7 they removed the much-bigger shadow on the foreground window because it 'looked ugly' and then removed most of the other visual clues that a window is foreground, measurably increasing the likelihood of users thinking that the wrong window is foreground. There are lots of things like this, where the UI has slowly regressed and, if anyone bothers to run user studies, they can clearly measure the regressions. Unfortunately, they've also improved a load of things and so there's no simple ordering of OS X versions by usability: each one introduces improvements and regressions.
As someone who does -j32 on our existing systems and sees almost linear speedup, I disagree. If you've got a decent amount of RAM, you won't hit the storage at all - everything will be in the buffer cache. Memory bandwidth is not an issue either - compiling generally has good locality of reference and so the cache works well.
The down side of SMT is that you increase contention on various resources. Cache is the biggest one, as sharing the cache between 4 or 8 threads increases contention a lot. Things like the cache and TLB depend on associativity and you can't arbitrarily scale associativity and still get single-cycle access, so you're going to hit diminishing returns quite quickly. Most of these things also increase power consumption a lot as you try to scale them.
In particular, on a modern system, register renaming takes a lot of the die area, and the more thread you have, the more rename registers you need to get the full throughput.
I'm slightly surprised at that, as last time I looked a lot of Android users were still on 2.x, which ships with Android Browser, not Chrome. I'm finding it very hard to read that graph as it doesn't provide totals for all versions of each browser, but it looks as if Android Browser still has around 5%, UC for Android (which I'd never heard of) has over 7%, so Chrome at 16% only has a bit more than 50% of the Android market share.
Not sure why this is off-topic. I started using Firefox on the phone because it was the first Android browser to offer a sane set of cookie management options (i.e. something beyond 'allow all' or 'block all', though it was restricted to this in the first couple of Android releases for some reason). With the self-destructing cookies plugin, it actually does what I want with respect to privacy. Most importantly though, it avoids a monoculture. Android has a huge market share and the idea of a bug in one browser being able to exploit the vast majority of all mobile phones is terrifying. Unfortunately, as with IE on Windows, enough apps use the Android WebView that there's a good chance that something else will run JavaScript with V8 even if you uninstall Chrome.
I just care about the Play Store. I can no longer purchase applications from the Play Store
For me, this has been a really good reason not to buy any applications from Google Play. I set up a new throw-away account for each device to download the free apps that I can only find in Google Play, but most of what I run is now from F-Droid. I've donated to a few of those projects, because I know that their apps won't randomly stop working at the whim of a third party.
To be fair to the space shuttle, it went a really high number of miles between services!
The P4 got very high clock rates by having a lot of very short pipeline stages. You can push clock rates a lot higher than they are today if you split your pipelines into more stages, each of which does less. The P4 had some piplelines that were over 30 stages long. A 30-stage pipeline means that if you put an operation in at one end, you get the result 30 cycles later. If you want to saturate the execution units, then you need to find 29 more operations to start that don't depend on the results of the first (or any of the others). This is incredibly rare. It's even more rare when you consider that a typical superscalar chip has half a dozen or so separate pipelines. The P4 could have up to 140 instructions in flight at any given time. To fully saturate the pipelines, you need to have code that has 140 totally independent operations at a time. This is why Intel added Hyperthreading: there were always a lot of pipeline slots that had nothing going into them.
As to the lack of increase in clock frequency, that's due to the end of Dennard Scaling. Read the Wikipedia article for more information, but basically shrinking the die size stopped giving you a free power consumption reduction about 10 years ago. This is why Dark Silicon is the industry buzzword de jour: Moore's law still means that you can double the number of transistors on your chip from generation to generation, but you only get a small increase in the number that you can have powered at any given time.
Sony or Microsoft might buy their chip facilities since they do use AMD chips as the CPU/GPU in their consoles
Given that AMD spun off its chip facilities a few years ago, I think we can probably ignore your market analysis.
There are two things wrong with this. The first is that no one does comparative benchmarks using Intel's compiler. Intel does, but they don't benchmark AMD chips, so you can't use them for Intel. The second is that a number of other x86 vendors have advertised features in CPUID and then either not implemented them at all, or implemented them in slow microcoded paths. Before Intel's compiler did the vendor ID checking, they were criticised for using features that were fast on their chips and very slow on others.
It's not quite so clear cut. Both vendors' chips are superscalar and so have multiple pipelines. Intel chips have around 5 independent integer pipelines, plus the AVX ones. Very few workloads will saturate them all. This was the main motivation for IBM's SMT: if you have two threads then you can get a lot closer to saturating the execution units than with one. Of course, you're still likely to suffer a bit because of contention on a few units, which is why Hyperthreading often isn't an overall performance win on Intel.
AMD implementations add more integer pipelines than any single-threaded workload is likely to be able to use, so they're more likely to get a speedup, but it's still not guaranteed if both workloads are very heavily biased towards a the same few pipelines.
As an opensource developer, do you really see someone as choosing to use another project as a problem?
Nope. Open source and proprietary alike care about contributors, not users. In off-the-shelf proprietary software, the (non-pirate) users are all contributors, because they all pay money. In open source, there are a lot of ways to contribute (code, testing, bug reports, documentation, artwork, and so on, as well as money), but users only matter inasmuch as they're real or potential contributors. Some users are negative contributors - they make a lot of demands, but give nothing back. The project is usually better off without them.
If the company is providing a significant contribution, then they may well expect something in return. Often, it's just the continued survival of the project, but if they want more then you should make sure that both parties agree on what the contribution is buying. If they're going to throw money at you, then they might reasonable assume that they've paid for some of your time to be spent on issues that matter to them. If you don't want to accept these terms, then having them use another project is probably better for both of you.
There was a vulnerability found in Xen a couple of weeks ago that allowed any PV guest running on any version of Xen released in the last 7 years, to map the whole of physical memory and tamper with the contents of any other VMs. This is, what, the fourth such exploit in Xen in the last year? Many of the others came from QEMU code, which was never intended to be used in security-critical situations. VMWare and HyperV almost certainly have similar issues, though they may not be so public about announcing them.
You may trust Amazon more than your own badly treated and underpaid IT staff, but do you also trust all of Amazon's customers more than your IT staff? Even if you're using VMs for your local deployment, the odds are that you aren't running VMs from third parties on the same physical hardware as your own VMs. If you host your stuff in someone else's cloud, then you aren't just trusting them, you're trusting either that their hypervisor is bug-free or that their other customers are all trustworthy.
No mod points for me, but the video linked by the parent explains exactly why it's so successful and deserves moderating up.
I've only ever played Candy Crush on an aeroplane. They're selling it as part of the seat-back entertainment system, which must bring in a fair bit (the input latency is quite painful, so it's not a great idea, which is a bit surprising, as Mario 3 worked well in seat-back entertainment systems 20 years ago). I don't know what other devices they license their games for, but they may have a few more unexpected revenue sources.
Social Networks are the graphs of personal relationships. The term has been used in sociology for a long time. Social Media refers to advertising platforms that provide surrogate friends in an attempt to collect the most personal information possible about a person to sell to advertisers and anyone else with spare money. So, yes, please don't confuse them.
And I really should have made all of those Amazon Affiliate links, but I'm too lazy...
They might hit problems if the words that they're matching on are correlated with speech patterns from ethnic groups. If you can find a set of words that black people are far more likely than white people to type into Facepalm posts then you can try claiming that it's not racial discrimination if you use them, but good luck convincing a court of that.
You don't have to pay interest, just get a no fee credit card, use it for some of your purchases and pay it off every month
People who don't do this and are proud of not using credit seem to be indicating that they're pretty poor at financial planning to me. I got a credit card as soon as I was legally able to (which had a very low credit limit, as I was a student with no real income). By the time I was earning, it had a big enough limit to cover all of my monthly expenses (though, unfortunately, I couldn't pay rent with it) and 1% cash back. By using the credit card, I get to have one more month's worth of salary in my savings account at any given time and I get 1% back on everything that I spend. I don't understand why anyone would prefer to use a debit card, which has neither of these advantages, but if I were lending money I'd look at the people who did and think that they're not so good at understanding personal finance.
Because saving money and paying off credit is basically the same thing: making sure you have left over money at the end of the month to achieve what you want/need to achieve
Nope, regular savings and paying off credit are the same things. This is why having a savings account that has a regular fixed payment increases your credit score, but one that you just dump money into periodically does not.
I think the truth is more likely closer to what others mention here: why would banks give credit to people who hardly need it, and will never ever be late in paying or transfer balances to the next month? that's where all their money comes from, sounds like a terrible customer to do business with!
It depends a lot on the kind of credit. For personal loans, the interest rates are so high that you want people who can pay it back, because you're making 10-20% in a year on the total loan. For mortgages, the cost of repossessing a house and then selling it can be very high, so you really want people who are able to pay back the loan. For credit cards, you want a mixture of people, some who will have a high throughput and rack up a load of transaction fees from the merchant and who can justify to the merchants that they'd lose business if they didn't accept cards (I put about £20K on my credit card each year, mostly business expenses, and get very good customer service - everything I've complained about has been resolved immediately and followed up with a goodwill payment to apologise), and some who will serve as long-term income sources by running up a smallish debt that they can't pay and effectively paying the interest in perpetuity.
My bank doesn't appear to even share my income with themselves. When I applied for a mortgage, one of the things that I had to prove was that I'd been paying my rent regularly for the last three months (yes, they really did only care about three months). I thought this would be easy: I'd been paying by standing order, from an account with them, so they could easily look at when the standing order was set up, and the fact that it had been paid every month for the last few years. Apparently they couldn't - I had to take three months of bank statements into the branch. It was then that I understood why subprime mortgages had been such an issue.
Maybe London could leave the EU and the UK, and then everyone would be happy.
Just imagine the chinese coming and cashing in your debts. Your whole country would be out their houses.
The Fed would print enough dollars to pay them. There would be inflation, which would devalue the dollar. Most people would no longer be able to afford 'cheap' Chinese imports, because they'd now be very expensive and the Chinese economy would crash from having one of its two biggest markets suddenly disappear. Exports from the USA would suddenly become very cheap. Of, and all of the rich folk who had moved a load of their capital to Chinese investments would make a huge loss. Guess which of these reasons is why it hasn't happened.