There was zero wasted time and effort open sourcing it. The work was done either open or closed, they just put the code in a different electronic place. You should be fired from your business.
That's very unusual. Typically, open sourcing a proprietary codebase requires at least a basic audit to ensure that you actually own the code, that it doesn't contain any third-party proprietary data that you're not allowed to release, and so on. Open sourcing code in a commercial environment is not just a matter of copying the code to GitHub, and if that's how you've been doing it then the grandparent isn't the one in this thread who should expect to be fired.
I used Firefox on Android with the self-destructing cookies plugin (best cookie management policy ever: moves cookies away after you leave a site, so the next visit doesn't see them, but keeps them for a bit in case you decide later that you needed them). Unfortunately, their hostile relationship with F-Droid (which could be fixed by simply providing their own F-Droid repo) meant that updating became painful and so I switched to using the new Chrome-based browser that comes with LineageOS.
Why would you see a performance hit? At that level of antialiasing, you're rendering at more than 4K and then downsampling. Just rendering at 4K and not downsampling would use less GPU power and give you a better image.
Nope. Memories are typically powers of two because that lets you make the most efficient use of addressing lines. If you have 32 addressing lines, then you can address 4GiB of memory with each possible value on the wire representing one address. If you have an amount of memory that isn't a power of two, then you are making inefficient use of the addressing lines. This has become less important with serial protocols (you don't need more wires, you just have a small amount of redundant data for the interface).
Good point, though 4K monitors are already pretty cheap: at work we switched to buying them by default two years ago because there wasn't much price difference between them and lower res ones.
I was watching a DVD the other day and noticing the pockmarks on the face of one of the characters. The jump from VHS to DVD quality was very noticeable. Anything beyond that and I stop caring - occasionally there are scenes when it makes a difference, but they're rare. 4K is nice for a monitor, because it makes text rendering a lot crisper (barely any antialiasing needed), but for video I don't care. 8K is probably more pixels than I can resolve unless the display is positioned such that I don't fit the entire thing in my field of vision at once.
That doesn't contradict what the grandparent was saying. I stopped going to the cinema for many reasons, but one was that they'd turn up the base a lot, so that action sequences were painfully loud, but dialog was very hard to understand.
If you discount people who can't afford much by way of luxury goods, the USA is probably closer to 20-25%. The Chinese middle class, for example, is about the same size as the entire population of the USA.
Record implies storing at the rate that the data is produced. Depending on the device, you may have a fairly small amount of buffering. You need to be able to accept a constant stream of data at around 30Mb/s, with a fairly small buffer, for a period of tens of minutes, which means that things like garbage collection in the controller can't block writes enough that this causes back-pressure and data loss.
In other words, you got fired and the government shut down had nothing to do with it
I don't know about Creimer's particular case, but one of the problems with the shutdown was that a lot of government contractors didn't get paid for quite a long time. If the company didn't have enough liquid cash to keep paying employees, they most likely had to fire some until they got paid the money that they were owed. For companies whose main customer is the government, then that's a possible explanation.
A UBI-like system can coexist with a purely capitalist-based economy if UBI is replaced by dividends and workers whose jobs are replaced by robots are compensated by shares in the company that owns the robots to the extent that the dividends would be equivalent to the production from the automation. This kind of system would likely result in a greater productivity gain overall, because the incentive for a worker is to automate themselves out of a job. Currently, if you're doing something that can be done more efficiently by a machine, your incentive is to keep quiet about it because if it happens then you'll be fired. Under this system, your incentive would be to encourage management to replace you because then you'd get the same (or more) income without having to do the work.
But you can't even cut and paste the last I looked.
When was that? Not only can you copy and paste on iOS, you can share the clipboard with your Mac so you can copy on the phone and paste on the computer or vice versa.
Uh, most of those cores are designed in the UK, just down the road from me. A few others are designed in a subsidiary in Austin. Some are designed in Cupertino, under license from a company based in the UK.
How long before there's a GRUB2 module (installed by default on Gentoo) that will recompile the kernel from source at every boot?
It exists already. This was one of the demos for tcc (which didn't optimise much, but compiled really fast): it would compile the kernel from the loader and then boot.
C (and far worse, C++) compilation is incredibly I/O bound because of the insanely archaic include system (and preprocessor too)
Not on any reasonably spec'd machine. After the first time a header is read, it's going to be in the buffer cache and so the next compilation job that needs them gets them out of RAM (unless you have such a tiny amount of RAM that they're evicted already, but even huge projects only have a few hundred MBs of headers and a single compiler invocation will use more RAM than that). Watching our build machines, they never read from the disk during big C/C++ compile jobs.
The bottleneck for C/C++ is that you end up re-parsing the same things over and over again. A typical C/C++ compilation unit is a few hundred thousand lines of code, but only a few hundred to a few thousand of those are unique to the compilation unit, the rest are shared. This is even worse for C++ than C, because template instantiations (often the same ones) are computationally expensive and because C++ implementations typically involve generating the same functions repeatedly and then having the linker throw them away. For example, a debug build of clang is 1.4GB, but generates 11GB of object code along the way, so the linker is throwing away almost 10GB of duplicates.
This is what the C++ module system addresses: it allows you to store ASTs from headers once and avoid re-parsing. With LTO, you can also avoid optimising the same functions multiple times and avoid generating binary code from them.
Who modded this insightful - nobody serious uses AMD.
True, now, but that's a fairly recent development. The Opteron shipped in 2003. It wasn't until 2009 that Intel integrated the memory controller on die and caught up. In that period, all serious server deployments were AMD. Xeons were overpriced and underpowered in comparison.
Actually the PASemi PA6T was very good in terms of performance and power, and it was clearly more advanced than Intel's Core2 Duo: memory controller and PCI-E on die, while Intel's needed an external chipset. It would have been perfect in a laptop
On paper. They never actually shipped any silicon. Switching to them would have been a huge gamble and would have left them in the same situation that they were attempting to avoid previously: paying 100% of the R&D costs and competing with companies that are were only paying a quarter of it.
There's a hardware cost, but there's also a non-trivial software cost -- to maintain two compiler versions, support two versions of applications, etc
They already maintain the compiler, toolchain, kernel, and most of their core frameworks on x86-64, AArch32 and AArch64 and they only just stopped supporting them on x86-32. The biggest issue would be for third-party developers, but Apple hasn't had a problem with this in the past and they already support fat binaries and have infrastructure for building them in their dev tools.
If they just swapped AMD for Intel they don't need to leave x86 land. And if they bought AMD outright, they control the cost of chips, too.
They already have their own in-house CPU and GPU design teams. The only thing that they'd gain from buying AMD would be an x86 license. They'd lose a lot because most PC makers wouldn't want to buy CPUs from Apple, so AMD would lose out on economies of scale.
It's also worth remembering that a big part of the reason for their switch from IBM to Intel was that they were buying all of the particular models of PowerPC chips that IBM was making. When they switched to Intel, they were shipping about 20% of Intel's production: enough that they were the largest single customer and so got preferential treatment but not enough that they were paying the entire R&D costs. For mobile chips, it's worth it because it's a competitive advantage, but for mobile chips it probably isn't.
Point is it's all going video and writers are being fired
It is, at 'Mic, a website aimed at millennials' that runs articles 'on topics like "celebrating beauty" and "strong women."'. Irrespective of the format, the content is no great loss.
You and I have very different definitions of optimal. 2TB SSDs cost a little under double what 10TB hard disks cost. For a consumer NAS, the performance difference vs having a reasonable amount of RAM / L2ARC is negligible over the network (the disks are not the bottleneck), but the cost for the same capacity is at least 10 times (probably more, as you need a lot more SSDs and so need more space, more controllers, and so on).
I'm not sure how they calculated the failure rates. For one model, they had 5/400 fail, for another 13/157. These are both pretty high (though with such small number, could be statistical flukes), but neither is 30%.
Slashdot Mirror
There was zero wasted time and effort open sourcing it. The work was done either open or closed, they just put the code in a different electronic place. You should be fired from your business.
That's very unusual. Typically, open sourcing a proprietary codebase requires at least a basic audit to ensure that you actually own the code, that it doesn't contain any third-party proprietary data that you're not allowed to release, and so on. Open sourcing code in a commercial environment is not just a matter of copying the code to GitHub, and if that's how you've been doing it then the grandparent isn't the one in this thread who should expect to be fired.
I used Firefox on Android with the self-destructing cookies plugin (best cookie management policy ever: moves cookies away after you leave a site, so the next visit doesn't see them, but keeps them for a bit in case you decide later that you needed them). Unfortunately, their hostile relationship with F-Droid (which could be fixed by simply providing their own F-Droid repo) meant that updating became painful and so I switched to using the new Chrome-based browser that comes with LineageOS.
Why would you see a performance hit? At that level of antialiasing, you're rendering at more than 4K and then downsampling. Just rendering at 4K and not downsampling would use less GPU power and give you a better image.
Nope. Memories are typically powers of two because that lets you make the most efficient use of addressing lines. If you have 32 addressing lines, then you can address 4GiB of memory with each possible value on the wire representing one address. If you have an amount of memory that isn't a power of two, then you are making inefficient use of the addressing lines. This has become less important with serial protocols (you don't need more wires, you just have a small amount of redundant data for the interface).
Good point, though 4K monitors are already pretty cheap: at work we switched to buying them by default two years ago because there wasn't much price difference between them and lower res ones.
I was watching a DVD the other day and noticing the pockmarks on the face of one of the characters. The jump from VHS to DVD quality was very noticeable. Anything beyond that and I stop caring - occasionally there are scenes when it makes a difference, but they're rare. 4K is nice for a monitor, because it makes text rendering a lot crisper (barely any antialiasing needed), but for video I don't care. 8K is probably more pixels than I can resolve unless the display is positioned such that I don't fit the entire thing in my field of vision at once.
That doesn't contradict what the grandparent was saying. I stopped going to the cinema for many reasons, but one was that they'd turn up the base a lot, so that action sequences were painfully loud, but dialog was very hard to understand.
If you discount people who can't afford much by way of luxury goods, the USA is probably closer to 20-25%. The Chinese middle class, for example, is about the same size as the entire population of the USA.
Record implies storing at the rate that the data is produced. Depending on the device, you may have a fairly small amount of buffering. You need to be able to accept a constant stream of data at around 30Mb/s, with a fairly small buffer, for a period of tens of minutes, which means that things like garbage collection in the controller can't block writes enough that this causes back-pressure and data loss.
In other words, you got fired and the government shut down had nothing to do with it
I don't know about Creimer's particular case, but one of the problems with the shutdown was that a lot of government contractors didn't get paid for quite a long time. If the company didn't have enough liquid cash to keep paying employees, they most likely had to fire some until they got paid the money that they were owed. For companies whose main customer is the government, then that's a possible explanation.
A UBI-like system can coexist with a purely capitalist-based economy if UBI is replaced by dividends and workers whose jobs are replaced by robots are compensated by shares in the company that owns the robots to the extent that the dividends would be equivalent to the production from the automation. This kind of system would likely result in a greater productivity gain overall, because the incentive for a worker is to automate themselves out of a job. Currently, if you're doing something that can be done more efficiently by a machine, your incentive is to keep quiet about it because if it happens then you'll be fired. Under this system, your incentive would be to encourage management to replace you because then you'd get the same (or more) income without having to do the work.
But you can't even cut and paste the last I looked.
When was that? Not only can you copy and paste on iOS, you can share the clipboard with your Mac so you can copy on the phone and paste on the computer or vice versa.
Uh, most of those cores are designed in the UK, just down the road from me. A few others are designed in a subsidiary in Austin. Some are designed in Cupertino, under license from a company based in the UK.
You can already use APFS for non-root filesystems on any storage media with macOS.
I pay my dishwasher with salt
As the grandparent said, you literally pay it a salary.
If you were worth a shit you'd be running Risc OS, but it, like you, sucks.
And yet descendants the processors that were co-designed with it are now in every smartphone, every tablet, and most embedded devices.
How long before there's a GRUB2 module (installed by default on Gentoo) that will recompile the kernel from source at every boot?
It exists already. This was one of the demos for tcc (which didn't optimise much, but compiled really fast): it would compile the kernel from the loader and then boot.
C (and far worse, C++) compilation is incredibly I/O bound because of the insanely archaic include system (and preprocessor too)
Not on any reasonably spec'd machine. After the first time a header is read, it's going to be in the buffer cache and so the next compilation job that needs them gets them out of RAM (unless you have such a tiny amount of RAM that they're evicted already, but even huge projects only have a few hundred MBs of headers and a single compiler invocation will use more RAM than that). Watching our build machines, they never read from the disk during big C/C++ compile jobs.
The bottleneck for C/C++ is that you end up re-parsing the same things over and over again. A typical C/C++ compilation unit is a few hundred thousand lines of code, but only a few hundred to a few thousand of those are unique to the compilation unit, the rest are shared. This is even worse for C++ than C, because template instantiations (often the same ones) are computationally expensive and because C++ implementations typically involve generating the same functions repeatedly and then having the linker throw them away. For example, a debug build of clang is 1.4GB, but generates 11GB of object code along the way, so the linker is throwing away almost 10GB of duplicates.
This is what the C++ module system addresses: it allows you to store ASTs from headers once and avoid re-parsing. With LTO, you can also avoid optimising the same functions multiple times and avoid generating binary code from them.
Who modded this insightful - nobody serious uses AMD.
True, now, but that's a fairly recent development. The Opteron shipped in 2003. It wasn't until 2009 that Intel integrated the memory controller on die and caught up. In that period, all serious server deployments were AMD. Xeons were overpriced and underpowered in comparison.
Actually the PASemi PA6T was very good in terms of performance and power, and it was clearly more advanced than Intel's Core2 Duo: memory controller and PCI-E on die, while Intel's needed an external chipset. It would have been perfect in a laptop
On paper. They never actually shipped any silicon. Switching to them would have been a huge gamble and would have left them in the same situation that they were attempting to avoid previously: paying 100% of the R&D costs and competing with companies that are were only paying a quarter of it.
There's a hardware cost, but there's also a non-trivial software cost -- to maintain two compiler versions, support two versions of applications, etc
They already maintain the compiler, toolchain, kernel, and most of their core frameworks on x86-64, AArch32 and AArch64 and they only just stopped supporting them on x86-32. The biggest issue would be for third-party developers, but Apple hasn't had a problem with this in the past and they already support fat binaries and have infrastructure for building them in their dev tools.
If they just swapped AMD for Intel they don't need to leave x86 land. And if they bought AMD outright, they control the cost of chips, too.
They already have their own in-house CPU and GPU design teams. The only thing that they'd gain from buying AMD would be an x86 license. They'd lose a lot because most PC makers wouldn't want to buy CPUs from Apple, so AMD would lose out on economies of scale.
It's also worth remembering that a big part of the reason for their switch from IBM to Intel was that they were buying all of the particular models of PowerPC chips that IBM was making. When they switched to Intel, they were shipping about 20% of Intel's production: enough that they were the largest single customer and so got preferential treatment but not enough that they were paying the entire R&D costs. For mobile chips, it's worth it because it's a competitive advantage, but for mobile chips it probably isn't.
It takes longer to produce a good video than to produce a good article, but it's a lot quicker to produce a crappy video.
Point is it's all going video and writers are being fired
It is, at 'Mic, a website aimed at millennials' that runs articles 'on topics like "celebrating beauty" and "strong women."'. Irrespective of the format, the content is no great loss.
You and I have very different definitions of optimal. 2TB SSDs cost a little under double what 10TB hard disks cost. For a consumer NAS, the performance difference vs having a reasonable amount of RAM / L2ARC is negligible over the network (the disks are not the bottleneck), but the cost for the same capacity is at least 10 times (probably more, as you need a lot more SSDs and so need more space, more controllers, and so on).
I'm not sure how they calculated the failure rates. For one model, they had 5/400 fail, for another 13/157. These are both pretty high (though with such small number, could be statistical flukes), but neither is 30%.