I think around 10-20Mb/s is where you start to hit diminishing returns. That's enough to stream HD and not completely kill everything else that you're doing. I remember as a student we decided to pay extra to get the entire 1Mb/s that our cable company offered (their standard package was 512Kb/s) for a shared house. It was a huge difference from the modem (56Kb/s in theory, 33.6Kb/s if I was lucky, 28.8Kb/s most of the time) that I was used to. I stayed on their fastest (and most expensive) plan until it hit 10Mb/s, and then stayed there as that gradually became their middle plan and then their slowest one. They upgraded me to 20, then 30 and 50Mb/s and I didn't really notice much difference except when I ran speed tests (the main thing was that their 30Mb/s plan, when I was throttled to 25% maximum speed for excess use, was still fast enough to stream HD video). I now have FTTP, but I'm on my provider's slowest plan, which is 54Mb/s. At this point, I notice upstream differences a lot more than downstream (copying 2TB to off-site backups took a few months), but that's not a number ISPs like to advertise.
At work, I have a GigE local network that goes to a faster upstream connection such that the GigE link is the bottleneck if I connect to a relatively close Internet peer. I *very* rarely notice the difference between that and the speed of my home connection (though it is nice when I install a game that happens to be on a local CDN node and it's fast enough to warm up my SSD), most of the time the bottleneck isn't my connection.
The point at which you hit diminishing returns is slowly edging up. 10 years ago, 10Mb/s was ample for anything. Now it's right at the slow end of generally useful. It's hard to tell the difference between 50, 100, or 300Mb/s unless you're actively monitoring the network.
A lot of the fraud was solved in the rest of the world by a simple change to the merchant banking rules: merchants may not take the card out of sight of the customer. If you want people to pay at the table in a restaurant, you come around with a wireless card reader. This removes 99% of the opportunities for skimming and it means that if a merchant does take the card away it's so unusual that the customer will likely remember it when they discover fraudulent transactions and can easily report the source. It's weird visiting the US and seeing that it's still standard to allow waiters to take the card away into a back room where they can make a clone and bring back the original.
That's the theory. Unfortunately, one of the flaws in the EMV protocol is that the authentication is unidirectional. The card must authenticate itself to the bank, but the bank doesn't have to authenticate itself to the card. This makes it comparatively easy to MITM the transaction. It's a shame that the US waited over 20 years until the EMV protocol had been thoroughly analysed and numerous flaws identified and then deployed it.
Graphite holds together, but only just. Pencils work because a tiny amount of shear force is enough to cause layers of it to come off (and that's the direction that you'd be fighting if you tried to pull two tubes apart that were stuck in this way). A child can pull a lump of graphite apart.
This has always been the problem with potential space-elevator materials. It's relatively easy to make something that's strong enough over a very short distance, but none of the proposed materials can either be synthesised in a single long chunk (yet?) or can be woven together to form a rope that maintains anything like the same tensile strength.
'Pro' products have become a synonym for Expensive,
I'm typing this on a late 2013 15" Retina MacBook Pro. I just looked up the order and it cost £2,540.21 for a quad-core 2.6GHz Haswell i7, 16GB of RAM, the GeForce 750M with 2GB of RAM, and a 1TB SSD, plus an external DVD drive and a few dongles. This was basically the top of the range for a MBP at the time (November 2013). It's lasted well and it's only the most recent generation that looks as if it's actually faster by a useful amount.
The new generation is, as you say, really expensive. £2,699.00 buys you the 2.6GHz model (6 core, so a decent speed improvement over the five-year-old model), but with only a 512GB SSD and the same amount of RAM. The markup on larger SSDs is insane. I can buy a 1TB NVMe drive for under £300, but Apple charges £360 for the upgrade to 1TB. They were charging about the same amount five years ago. Upgrading the disk and the RAM so that they're the smallest sizes that are more than my five-year-old model has brings the price to £4,139. That's more than I'm willing to pay for such a small improvement.
I think that's a mischaracterisation. Steve Jobs cared a lot about usability, but only for his specific use cases. Under Jobs, doing anything that he did on an Apple product was always efficient and streamlined. Doing things that he never did varied from 'works well' to 'isn't possible at all', often falling in the 'kind of works but is really buggy' part of the spectrum.
Since his death, the best-case usability of Apple products has definitely dropped, though I'm not sure where the average has ended up.
I just tried and in the zero-click info box at the top of DDG I get the exchange rate along with a little drop-down for each currency so that I can change either one. For 'highly technical searches', DDG partners with a couple of source-code indexing domain-specific search engines and so typically points me to the documentation for whatever I'm looking for in the zero-click box too.
I switched some years ago. If I don't find the result, I try with !b and !g (forwards to Bing and Google, respectively). Both Bing and Google will give me a load of results when DDG tells me it can't find anything, but I've yet to see an instance of those results actually corresponding to the thing that I'm searching for. I've no idea why Google thinks that giving me pages and pages of results that don't contain my search term is actually helpful.
If they wanted to actually contribute to the community then they would have made an LLVM frontend
They did. LLVM is not currently a great fit for garbage collected languages, though the LLILC team worked with some of the Azure folks to improve it in this regard. LLVM also suffered from longer compile times (important for a JIT). The Roslyn architecture makes it easy to incrementally replace the JIT, because it supports trying to compile individual functions with a new JIT and falling back to the old one if it doesn't support all of the functionality required. The LLILC team made use of this when trying to bootstrap.
I'm in the UK and get a fair number of scam calls. I've had the same telephone number for 20 years. Not sure about Australia, but in the UK, mobile numbers are allocated with a small number of non-geographical area codes, so if you guess a random number in those area codes there's a very high probability that you'll get a real number. You can also easily find out the blocks allocated to different carriers, so I get a lot of scammers claiming to be from the company that I used when I first had the number (I've since ported it to three other mobile operators, and each time the people phoning up claiming to be from the first ones are more and more ludicrous). I get a lot of scammers telling me I was miss-sold PPI (ironic, because I wasn't eligible for PPI as a result of being self employed during that scandal).
When has wealth not been concentrated? It has _always_ been concentrated. I'd argue that it's less concentrated today. Back in the day, some asshole of a King owned everything, including your LIFE.
We are far from the worst period in history, but that's not exactly something to be proud of. I read an article last week that looked at the amount of money made by labour versus the amount made by owning capital in the UK over the 20th century. The labour percentage was up at around 70% a few decades ago and is now closer to 50%.
Wealth will always concentrate unless you are willing to strip humans of freedom and free will.. One guy will always be just a tiny bit better at his business... He'll make just a tad more profit, save an iota more on expenses.. Slowly his fortune will grow.
Most people don't object to someone who is better at their job earning more. They object to the people with huge incomes that result from the things that they own, not from the things that they do.
Someone did run the numbers for the UK a few years ago. They assumed a £10K/year payment (which is enough to live on outside of London - very comfortably in some parts of the country - and might help reduce the housing pressure in London), set the tax-free allowance to the same as the UBI amount (so you didn't pay tax on the UBI, but you did on every pound earned after that) and shuffled the tax bands around to make it revenue neutral (i.e. they absorbed unemployment benefits and so on, but assumed that the changes in taxes must raise enough to pay for it). As I recall, anyone currently on £20K/year or less would be better off, anyone earning more would be worse off. I'd be paying a noticeably larger tax bill each year, but I'm okay with that in exchange for a social safety net that means that no one starves because they can't find employment.
Do you think it's easier to spot fraud in a system with thousands of rules and exceptions to the rules that decides who gets paid what and how much, or in a system where everyone receives exactly the same amount?
Which is exactly what you'd expect. You pump more energy into a chaotic system, you will see more variation at both extremes. Imagine you have a slightly off-axis spinning top, travelling along a roughly straight path. Now you spin it faster, do you expect it to veer off to the left? No, you'd expect it to oscillate wildly until it eventually falls over and reaches a stable equilibrium.
The quoted post in the link from TFS does not paint her in a very good light. The response to her initial thing is:
Really interesting thread to read!
However, allow me to disagree *slightly*. I dont[sic] believe the issue lies in the MMORPG genre itself (as your wording seemingly suggest[sic]). I believe the issue lies in the constraints of the Living Story's narrative design; (1 of 3)
That sounds like a polite disagreement. Her response was:
Today in being a female game dev:
"Allow me--a person who does not work with you--explain to you how you do your job"
This was a personal attack, a mischaracterisation of the post (at least the first one, I've not seen 2/3 or 3/3). As a game developer talking to a fan, she was the one in the relative position of power and she uses this to belittle someone.
Her Twitter profile says:
Game producer, writer, editor, howling maenad. ArenaNet Narrative team. Obsessed with lionesses. Salty language. I block often. I won't play demure for you.
i.e. she is explicitly associating herself with ArenaNet. There is no statement that her views do not necessarily reflect the views of her company, she is representing them in public by belittling and insulting their customers. She also seems to think 'play demure' means 'interact like a reasonable human being'.
I'd be willing to bet that if she had a gender-neutral avatar and removed the gendered terminology from her posts and profile then people would think she was a male asshat.
The link in TFS shows only the first post that she responded to, but from the quoted section someone politely suggested that she was generalising from her personal experience to an entire industry and she then immediately launched a personal attack in response. If you do that in private, you're an asshat. If you do that after publicly associating your online persona with your company, there's grounds for disciplinary action. Worse, her response explicitly drew attention to her link with her employer.
Twitter makes no difference, if you go around saying 'as an employee of FooCorp, I am an expert in this' via any communication channel, then if you subsequently act in such a way that reflects poorly on FooCorp then you'd expect issues. It isn't private communication when you're broadcasting it in public and using your company's name.
Whether false positives or false negatives matter depends on what you're using it for. If you're using it to highlight 10% of the people in a crowd that a police officer should look at, with the expectation that under 1% are actually criminals, then it's fine to have false positives (the police officer's job is to filter those), but it's bad to have false negatives (they let criminals slip through). If the goal is to lock up everyone who triggers an alert, then it's bad to have false positives because you lock up innocent people. This system is intended for the former use case, not the latter.
but accepted only for that specific hostname, never allowed to be treated as any sort of root cert
The root cause of this problem is that there are a number of hosts that people care about that are not part of a global namespace. router.local, for example, may exist on any network and will be a different machine on each one. For any kind of secure connection, you need a way of identifying the endpoint (if you're not securely communicating with a specific endpoint, then you may be securely communicating with the entity running a MITM attack).
It would be relatively easy to integrate some kind of key signing into mDNS / DNS-SD so that anything on the.local domain could advertise a TLS cert, but designing a good UI that lets a user tell the difference between foo.local on their home network and foo.local on the network that they accidentally joined because it has a stronger signal and which wants to steal their passwords is much harder.
Pretty much all CPU vendors were vulnerable to Spectre. Intel helpfully patented the technique that led to Meltdown, so the only non-Intel chip that was vulnerable was a very recent ARM core (released just after the patent expired, such bad timing!). Some of the more recent vulnerabilities rely on particular microarchitectural choices, so are probably limited to Intel. I suspect that other out-of-order cores will have some similar issues, but Intel is in most machines in all of the big cloud data centres so it's the highest-value target.
No we don't. We have 48-59 bits, depending on the particular microarchitecture. You can exhaust 48 bits pretty quickly if each process wants 32 bits of address space and many things want more because they like to have big gaps to make memory safety bugs more likely to crash.
When the clones became popular, the phrase was shortened in general use to "PC".
And, in particular, when the clone vendors wanted to highlight that they ran DOS / Windows and downplay the fact that they were compatible with an old IBM machine. No one likes advertising their competitor's product in their marketing material.
The 2006 models had 32-bit CPUs when they were already obsolete, so let's not do that again. I had a 2008 Core 2 model, which used the same form factor as the 2006 one, and replaced it with an early 2011 model and a late 2013 one (my partner still uses the 2011 model). The 2008 model had a comparatively small battery (though one that was easy to remove), but aside from that the 2011 unibody model was a lot better - less fragile hinges and more robust. The 2013 version removed the optical drive and increased the battery side again. It's lighter and cooler and nothing Apple has released since has made me want to upgrade. I'd love to have something like the 2013 version with 32 (or, ideally, 64) GB of RAM, a 2TB SSD and a Core i9 CPU. I've kept this computer as my main machine for longer than any other computer that I've owned, waiting for something that's a real upgrade.
Do you mean 64-bit x86 applications or 64-bit ARM applications? This chip is using 64-bit ARM cores, so will run 64-bit software natively. Whether it runs 64-bit x86 software is a question of the functionality of the emulator, not of the hardware. I don't believe that Microsoft's x86 emulator supports 64-bit programs yet, but it was announced to be on the roadmap a few months back.
Why does chrome need 4 processes before it displays a home/start page?
It's been a few years since I looked, but as I recall:
One is the parent process which manages the rest and holds all the rights that the program is started with.
One is the credential store. It manages passwords and hands them out only to the correct renderers.
One is the zygote for renderer processes. This does all of its initialisation and then fork()s clones so that each new tab can have a pristine renderer.
One is the owner for plugins (or possibly the zygote for plugins). NAPI plugins run in a separate process with reduced privileges, so that they can't compromise the rest of the tab's state.
For something that deals with as much untrusted data and code as a web browser, I'd want it to be compartmentalised as much as possible.
But one thing the FOSS community sometimes forgets is that creating software costs time and money, lots of it.
No it doesn't. This is something that the FOSS community explicitly acknowledges. The difference between Free and Proprietary software is that FOSS acknowledges that copying software doesn't cost money. The premise of TFA is flawed, because a lot of people are paid to develop FOSS, but they're not paid to produce copies of it. If someone wants a feature added to a FOSS program or library, then they can pay either the original author or someone else to develop it and that's how a huge amount of FOSS is funded. In contrast, in the proprietary off-the-shelf software world, someone pays to have a feature developed in the hope that they can charge money for copies of the software. This business model is problematic, because developing the software is expensive but copying it is trivial and so you need to rely on legal and technical protections to try to prevent people from doing the easy thing so that they will pay you for the expensive thing that you did for free.
Now, it would be nice to have an easier mechanism for 1,000 people to each pay for 1,000th of the cost of developing a new feature...
Slashdot Mirror
I think around 10-20Mb/s is where you start to hit diminishing returns. That's enough to stream HD and not completely kill everything else that you're doing. I remember as a student we decided to pay extra to get the entire 1Mb/s that our cable company offered (their standard package was 512Kb/s) for a shared house. It was a huge difference from the modem (56Kb/s in theory, 33.6Kb/s if I was lucky, 28.8Kb/s most of the time) that I was used to. I stayed on their fastest (and most expensive) plan until it hit 10Mb/s, and then stayed there as that gradually became their middle plan and then their slowest one. They upgraded me to 20, then 30 and 50Mb/s and I didn't really notice much difference except when I ran speed tests (the main thing was that their 30Mb/s plan, when I was throttled to 25% maximum speed for excess use, was still fast enough to stream HD video). I now have FTTP, but I'm on my provider's slowest plan, which is 54Mb/s. At this point, I notice upstream differences a lot more than downstream (copying 2TB to off-site backups took a few months), but that's not a number ISPs like to advertise.
At work, I have a GigE local network that goes to a faster upstream connection such that the GigE link is the bottleneck if I connect to a relatively close Internet peer. I *very* rarely notice the difference between that and the speed of my home connection (though it is nice when I install a game that happens to be on a local CDN node and it's fast enough to warm up my SSD), most of the time the bottleneck isn't my connection.
The point at which you hit diminishing returns is slowly edging up. 10 years ago, 10Mb/s was ample for anything. Now it's right at the slow end of generally useful. It's hard to tell the difference between 50, 100, or 300Mb/s unless you're actively monitoring the network.
A lot of the fraud was solved in the rest of the world by a simple change to the merchant banking rules: merchants may not take the card out of sight of the customer. If you want people to pay at the table in a restaurant, you come around with a wireless card reader. This removes 99% of the opportunities for skimming and it means that if a merchant does take the card away it's so unusual that the customer will likely remember it when they discover fraudulent transactions and can easily report the source. It's weird visiting the US and seeing that it's still standard to allow waiters to take the card away into a back room where they can make a clone and bring back the original.
That's the theory. Unfortunately, one of the flaws in the EMV protocol is that the authentication is unidirectional. The card must authenticate itself to the bank, but the bank doesn't have to authenticate itself to the card. This makes it comparatively easy to MITM the transaction. It's a shame that the US waited over 20 years until the EMV protocol had been thoroughly analysed and numerous flaws identified and then deployed it.
Graphite holds together, but only just. Pencils work because a tiny amount of shear force is enough to cause layers of it to come off (and that's the direction that you'd be fighting if you tried to pull two tubes apart that were stuck in this way). A child can pull a lump of graphite apart.
This has always been the problem with potential space-elevator materials. It's relatively easy to make something that's strong enough over a very short distance, but none of the proposed materials can either be synthesised in a single long chunk (yet?) or can be woven together to form a rope that maintains anything like the same tensile strength.
'Pro' products have become a synonym for Expensive,
I'm typing this on a late 2013 15" Retina MacBook Pro. I just looked up the order and it cost £2,540.21 for a quad-core 2.6GHz Haswell i7, 16GB of RAM, the GeForce 750M with 2GB of RAM, and a 1TB SSD, plus an external DVD drive and a few dongles. This was basically the top of the range for a MBP at the time (November 2013). It's lasted well and it's only the most recent generation that looks as if it's actually faster by a useful amount.
The new generation is, as you say, really expensive. £2,699.00 buys you the 2.6GHz model (6 core, so a decent speed improvement over the five-year-old model), but with only a 512GB SSD and the same amount of RAM. The markup on larger SSDs is insane. I can buy a 1TB NVMe drive for under £300, but Apple charges £360 for the upgrade to 1TB. They were charging about the same amount five years ago. Upgrading the disk and the RAM so that they're the smallest sizes that are more than my five-year-old model has brings the price to £4,139. That's more than I'm willing to pay for such a small improvement.
Sure, his "good" was about fashion
I think that's a mischaracterisation. Steve Jobs cared a lot about usability, but only for his specific use cases. Under Jobs, doing anything that he did on an Apple product was always efficient and streamlined. Doing things that he never did varied from 'works well' to 'isn't possible at all', often falling in the 'kind of works but is really buggy' part of the spectrum.
Since his death, the best-case usability of Apple products has definitely dropped, though I'm not sure where the average has ended up.
I just tried and in the zero-click info box at the top of DDG I get the exchange rate along with a little drop-down for each currency so that I can change either one. For 'highly technical searches', DDG partners with a couple of source-code indexing domain-specific search engines and so typically points me to the documentation for whatever I'm looking for in the zero-click box too.
I switched some years ago. If I don't find the result, I try with !b and !g (forwards to Bing and Google, respectively). Both Bing and Google will give me a load of results when DDG tells me it can't find anything, but I've yet to see an instance of those results actually corresponding to the thing that I'm searching for. I've no idea why Google thinks that giving me pages and pages of results that don't contain my search term is actually helpful.
If they wanted to actually contribute to the community then they would have made an LLVM frontend
They did. LLVM is not currently a great fit for garbage collected languages, though the LLILC team worked with some of the Azure folks to improve it in this regard. LLVM also suffered from longer compile times (important for a JIT). The Roslyn architecture makes it easy to incrementally replace the JIT, because it supports trying to compile individual functions with a new JIT and falling back to the old one if it doesn't support all of the functionality required. The LLILC team made use of this when trying to bootstrap.
I'm in the UK and get a fair number of scam calls. I've had the same telephone number for 20 years. Not sure about Australia, but in the UK, mobile numbers are allocated with a small number of non-geographical area codes, so if you guess a random number in those area codes there's a very high probability that you'll get a real number. You can also easily find out the blocks allocated to different carriers, so I get a lot of scammers claiming to be from the company that I used when I first had the number (I've since ported it to three other mobile operators, and each time the people phoning up claiming to be from the first ones are more and more ludicrous). I get a lot of scammers telling me I was miss-sold PPI (ironic, because I wasn't eligible for PPI as a result of being self employed during that scandal).
When has wealth not been concentrated? It has _always_ been concentrated. I'd argue that it's less concentrated today. Back in the day, some asshole of a King owned everything, including your LIFE.
We are far from the worst period in history, but that's not exactly something to be proud of. I read an article last week that looked at the amount of money made by labour versus the amount made by owning capital in the UK over the 20th century. The labour percentage was up at around 70% a few decades ago and is now closer to 50%.
Wealth will always concentrate unless you are willing to strip humans of freedom and free will.. One guy will always be just a tiny bit better at his business... He'll make just a tad more profit, save an iota more on expenses.. Slowly his fortune will grow.
Most people don't object to someone who is better at their job earning more. They object to the people with huge incomes that result from the things that they own, not from the things that they do.
Incidentally, have you considered the numbers?
Someone did run the numbers for the UK a few years ago. They assumed a £10K/year payment (which is enough to live on outside of London - very comfortably in some parts of the country - and might help reduce the housing pressure in London), set the tax-free allowance to the same as the UBI amount (so you didn't pay tax on the UBI, but you did on every pound earned after that) and shuffled the tax bands around to make it revenue neutral (i.e. they absorbed unemployment benefits and so on, but assumed that the changes in taxes must raise enough to pay for it). As I recall, anyone currently on £20K/year or less would be better off, anyone earning more would be worse off. I'd be paying a noticeably larger tax bill each year, but I'm okay with that in exchange for a social safety net that means that no one starves because they can't find employment.
You think fraud goes away like magic?
Do you think it's easier to spot fraud in a system with thousands of rules and exceptions to the rules that decides who gets paid what and how much, or in a system where everyone receives exactly the same amount?
Which is exactly what you'd expect. You pump more energy into a chaotic system, you will see more variation at both extremes. Imagine you have a slightly off-axis spinning top, travelling along a roughly straight path. Now you spin it faster, do you expect it to veer off to the left? No, you'd expect it to oscillate wildly until it eventually falls over and reaches a stable equilibrium.
Really interesting thread to read!
However, allow me to disagree *slightly*. I dont[sic] believe the issue lies in the MMORPG genre itself (as your wording seemingly suggest[sic]). I believe the issue lies in the constraints of the Living Story's narrative design; (1 of 3)
That sounds like a polite disagreement. Her response was:
Today in being a female game dev:
"Allow me--a person who does not work with you--explain to you how you do your job"
This was a personal attack, a mischaracterisation of the post (at least the first one, I've not seen 2/3 or 3/3). As a game developer talking to a fan, she was the one in the relative position of power and she uses this to belittle someone.
Her Twitter profile says:
Game producer, writer, editor, howling maenad. ArenaNet Narrative team. Obsessed with lionesses. Salty language. I block often. I won't play demure for you.
i.e. she is explicitly associating herself with ArenaNet. There is no statement that her views do not necessarily reflect the views of her company, she is representing them in public by belittling and insulting their customers. She also seems to think 'play demure' means 'interact like a reasonable human being'.
I'd be willing to bet that if she had a gender-neutral avatar and removed the gendered terminology from her posts and profile then people would think she was a male asshat.
The link in TFS shows only the first post that she responded to, but from the quoted section someone politely suggested that she was generalising from her personal experience to an entire industry and she then immediately launched a personal attack in response. If you do that in private, you're an asshat. If you do that after publicly associating your online persona with your company, there's grounds for disciplinary action. Worse, her response explicitly drew attention to her link with her employer.
Twitter makes no difference, if you go around saying 'as an employee of FooCorp, I am an expert in this' via any communication channel, then if you subsequently act in such a way that reflects poorly on FooCorp then you'd expect issues. It isn't private communication when you're broadcasting it in public and using your company's name.
Whether false positives or false negatives matter depends on what you're using it for. If you're using it to highlight 10% of the people in a crowd that a police officer should look at, with the expectation that under 1% are actually criminals, then it's fine to have false positives (the police officer's job is to filter those), but it's bad to have false negatives (they let criminals slip through). If the goal is to lock up everyone who triggers an alert, then it's bad to have false positives because you lock up innocent people. This system is intended for the former use case, not the latter.
but accepted only for that specific hostname, never allowed to be treated as any sort of root cert
The root cause of this problem is that there are a number of hosts that people care about that are not part of a global namespace. router.local, for example, may exist on any network and will be a different machine on each one. For any kind of secure connection, you need a way of identifying the endpoint (if you're not securely communicating with a specific endpoint, then you may be securely communicating with the entity running a MITM attack).
It would be relatively easy to integrate some kind of key signing into mDNS / DNS-SD so that anything on the .local domain could advertise a TLS cert, but designing a good UI that lets a user tell the difference between foo.local on their home network and foo.local on the network that they accidentally joined because it has a stronger signal and which wants to steal their passwords is much harder.
Pretty much all CPU vendors were vulnerable to Spectre. Intel helpfully patented the technique that led to Meltdown, so the only non-Intel chip that was vulnerable was a very recent ARM core (released just after the patent expired, such bad timing!). Some of the more recent vulnerabilities rely on particular microarchitectural choices, so are probably limited to Intel. I suspect that other out-of-order cores will have some similar issues, but Intel is in most machines in all of the big cloud data centres so it's the highest-value target.
We have 64 bits virtual.
No we don't. We have 48-59 bits, depending on the particular microarchitecture. You can exhaust 48 bits pretty quickly if each process wants 32 bits of address space and many things want more because they like to have big gaps to make memory safety bugs more likely to crash.
When the clones became popular, the phrase was shortened in general use to "PC".
And, in particular, when the clone vendors wanted to highlight that they ran DOS / Windows and downplay the fact that they were compatible with an old IBM machine. No one likes advertising their competitor's product in their marketing material.
The 2006 models had 32-bit CPUs when they were already obsolete, so let's not do that again. I had a 2008 Core 2 model, which used the same form factor as the 2006 one, and replaced it with an early 2011 model and a late 2013 one (my partner still uses the 2011 model). The 2008 model had a comparatively small battery (though one that was easy to remove), but aside from that the 2011 unibody model was a lot better - less fragile hinges and more robust. The 2013 version removed the optical drive and increased the battery side again. It's lighter and cooler and nothing Apple has released since has made me want to upgrade. I'd love to have something like the 2013 version with 32 (or, ideally, 64) GB of RAM, a 2TB SSD and a Core i9 CPU. I've kept this computer as my main machine for longer than any other computer that I've owned, waiting for something that's a real upgrade.
Do you mean 64-bit x86 applications or 64-bit ARM applications? This chip is using 64-bit ARM cores, so will run 64-bit software natively. Whether it runs 64-bit x86 software is a question of the functionality of the emulator, not of the hardware. I don't believe that Microsoft's x86 emulator supports 64-bit programs yet, but it was announced to be on the roadmap a few months back.
Why does chrome need 4 processes before it displays a home/start page?
It's been a few years since I looked, but as I recall:
For something that deals with as much untrusted data and code as a web browser, I'd want it to be compartmentalised as much as possible.
But one thing the FOSS community sometimes forgets is that creating software costs time and money, lots of it.
No it doesn't. This is something that the FOSS community explicitly acknowledges. The difference between Free and Proprietary software is that FOSS acknowledges that copying software doesn't cost money. The premise of TFA is flawed, because a lot of people are paid to develop FOSS, but they're not paid to produce copies of it. If someone wants a feature added to a FOSS program or library, then they can pay either the original author or someone else to develop it and that's how a huge amount of FOSS is funded. In contrast, in the proprietary off-the-shelf software world, someone pays to have a feature developed in the hope that they can charge money for copies of the software. This business model is problematic, because developing the software is expensive but copying it is trivial and so you need to rely on legal and technical protections to try to prevent people from doing the easy thing so that they will pay you for the expensive thing that you did for free.
Now, it would be nice to have an easier mechanism for 1,000 people to each pay for 1,000th of the cost of developing a new feature...