Keep in mind that in many states, union membership is required in order to get the job.
I've never understood this about the USA, it seems to completely miss the point of unions. Here, there are often two or three unions that are competing for members, so you get the benefits of collective bargaining and the benefits of competition. Collective bargaining via a monopoly that has no incentive to represent your interests is much the same position you're in with no union at all...
The core count isn't the interesting thing about this chip. The cores themselves are pretty boring off-the-shelf parts too. I was at the ISCA presentation about this last week and it's actually pretty interesting. I'd recommend reading the paper (linked to from the press release) rather than the press release, because the press release is up to MIT's press department's usual standards (i.e. completely content-free and focussing on totally the wrong thing). The cool stuff is in the interconnect, which uses the bounded latency of the longest path multiplied by single-cycle one-hop delivery times to define an ordering, allowing you to implement a sequentially consistent view of memory relatively cheaply.
Since I'm here, I'll also throw out a plug for the work we presented at ISCA, The CHERI capability model: Revisiting RISC in an age of risk. We've now open sourced (as a code dump, public VCS coming soon) our (64-bit) MIPS softcore, which is the basis for the experimentation in CHERI. It boots FreeBSD and there are a few sitting around the place that we can ssh into and run. This is pretty nice for experimentation, because it takes about 2 hours to produce and boot a new revision of the CPU.
The other part, readings, are not necessarily available to the general public. If you want to get any recent consensus in a field, you need access to e.g. JSTOR, which is not freely available to the public
In computer science, most academics put their papers online (sometimes preprints if the journal doesn't allow distribution) on their own web sites for free. If you find the paper in the ACM Digital Library, and then search the web for the paper title, you can probably find it. In physics, arxiv.org does a good job of centralising preprints for people to search.
The problem with this approach is that you need a fairly good grounding in the field to be able to understand the papers. You can probably get it by reading books, but it's a bit hit-and miss. The point of a university education is not to teach you the field, it's to give you a guided tour of your ignorance. This then lets you identify the things that you should know and learn them.
Uh, what? Quartz is the name of Apple's display system. XQuartz is an implementation of X atop Quartz. Oh, and it's not an Apple product, so was not named by Apple.
They don't, but the do some clever tricks with launchd so that the first time you try to run a program that tries to connect to an X server they'll pop up a thing asking if you want to install XQuartz. Oh, and XQuartz never shipped with OS X, Apple's X11.app did, which was a fork and rebranding of XQuartz and usually an old version by the time it shipped. The reason that they stopped bundling it was that all of the people who actually cared about X11 were installing XQuartz instead.
There are four different protocols that permit double-buffering on X. These days, pretty much everyone has converged on using XRender and manually doing the buffer management. For most of the history of X, different vendors supported different double buffering APIs and writing code that would detect which one was available and using it was painful. Double buffering isn't part of the core specification, because the RAM requirements for two copies of the frame buffer were too big for a lot of early implementations.
Except for the fact that it has a limited set of extensions that can be supported and a load of command numbers are used for 'core protocol' stuff that no one has used for over a decade. It has no concept of security (you can easily steal input from another application, for example).
the library isn't that horrible unless your a newbie to dev
XCB is pretty nice, but xlib is a clusterfuck. It hides interfaces that need to be used asynchronously for good performance behind synchronous API calls. It's impossible to write an application that performs well over a network and does a nontrivial amount of drawing with xlib. It is with XCB, but it requires carefully designing your toolkit for asynchronous drawing, and all modern X toolkits have too much xlib heritage to easily adapt to using XCB as it's intended to be used, rather than as a lighter-weight xlib.
The problem with network transparency in X11 is that it's done at completely the wrong place. With competing systems of the same era, such as NeWS, there was some code running in the display server for display updates. This meant that, for example, you could handle the visual feedback for a button press in the server, while transmitting the 'this button has been pressed' event to the client. In X11, you press the button but the server just sends a 'mouse clicked at x,y' message to the client, so you need a network round trip just to update the button. If you want to animate the button press, then you need to wait for network round trips to get the 'redraw finished' events. Wayland isn't a step backwards in this regard, but it's also not a step forwards.
In a modern X client, you don't really use much of the server's drawing functionality. You do store some images for compositing and will use XRender to composite them, but that's about it. The line drawing stuff can't handle antialiased lines, the text drawing stuff (aside from XRender) can't handle fonts provided by the client easily, so all you're really using the display server for is getting some texture memory and compositing it. With Wayland, you just get an OpenGL context and do the same thing. To be honest, if you're targeting X11 that's also what you should be doing for modern hardware: the rest of your drawing code most likely uses OpenGL (or something higher-level with an OpenGL back end), or just generates pixmaps, so doing the per-window compositing in OpenGL is a lot easier than doing it in a completely different API.
The only reason China 'controls' rare earth production is that they spent a few years dumping them on the market at less than the cost of extraction, pushing the other mines out of business (or, at least, into mothballing). Now that the prices are sane again, other countries are bringing their mines back online.
AMD and Apple picked it up because it's the only replacement for DVI which is capped at 1600x1200 at 60Hz or 1200p at 60Hz
Not to mention that it's royalty free. You have to pay a license for every HDMI port, you can stick any number of DisplayPort ports on a machine without paying a royalty for implementing the standard.
TFA is about monitors. The pixel density on my laptop is about double that of the 28" display on my desk. This is really noticeable for text rendering, where it's clear and crisp on the laptop screen, but looks a little blurry around the edges due to the sub-pixel AA on the bigger display. I'd love to replace the one on my desk with a 4K display once the prices become a bit less silly. And, yes, my laptop can drive a 4K display.
The Z-series and POWER are not quite separate chips. They're separate instruction decoders but they're largely the same pipelines after that. There are some tweaks, but within a generation they share more design than either does with the previous generation of the same processor.
This logic doesn't really make sense to me. Sure, 20 adds up over a few million devices, but 20isn't the important number, the profit margin is. If you're making 40 profit on each device, then 20 is a big deal. You can double your profit by cutting it. If you're making $20 profit on each device, then that 20 will make almost no difference to your bottom line. It's only an issue if you can save 20 on a lot of components in a single device.
Hopefully everyone in this thread is joking, but it's worth noting that it's not quite that clear cut. The smallest assignment that an ISP can hand out is a/64, so you can really only have 2^64 sites. IPv6 has 2^128 addresses, but a lot of the design works around having sparse routing tables. You really want each/64 to correspond to a broadcast domain, and you don't want to fragment the routing tables too much to get to the/64, so you've actually got a lot fewer addresses. A/64 per human is not enough to assign one IP per atom in the person, but it likely is enough for every device that a person may reasonably want to own and give an IP to, even if that person has a lot of injected sensor nodes.
Offer a discount for services that are willing to use v6-only. If you're hosting some back-end services in Azure, you can use v6. If it's something for only accessing within your organisation then you can possibly use v6, depending on your local connectivity. If it's something for the public then you can probably make a certain percentage of the servers v6-only and send customers with working v6 there (does Windows still set up v6 tunnels by default?).
If companies start to pay less for v6-only hosting than for v4-only or dual stack, then they're going to start pushing their customers towards the v6 servers, or making certain features v6-only to penalise ISPs that don't provide v6 connectivity by making their customers complain. That's what's going to trigger mass movement to v6.
If only your browser sent a header telling the server what your preferred language was. Oh, wait, it does, and Google still thinks that I want to go to their Japanese page when I'm in Japan. One of the many reasons I switched to DuckDuckGo a few years ago...
I've had a load of LinkedIn spam sent to mailing lists recently. The problem with this lawsuit is that it's the wrong people - LinkedIn is sending spam, they should be prosecuted, not sued by people on whose behalf they claimed to send it.
I don't understand the hate. I installed Firefox 29 and the UI seemed clean and easy to customise if you like more clutter (possibly useful if you're on a tablet, or like wasting screen real-estate with buttons that are less convenient than keyboard shortcuts, but a waste of space for most users).
A lot of Chrome's memory use comes from using multiple processes to isolate components of the web browser so that a compromise in one tab does not compromise the entire browser. Does Firefox yet do anything similar?
I switched from Chrome to Firefox on Android because as far as I can tell (after trying a load of them) Firefox is the only Android browser that doesn't have a completely fucked-up cookie management policy. It's made even nicer by the self destructing cookies addon, which lets me whitelist sites with cookies I want to keep (i.e. ones where I log in), but destroys tracking cookies and HTML5 local storage as soon as I close a tab or navigate away from a page (keeping them for a little bit in some storage that isn't visible to the page, so I can hit an undo button if I realise it's just deleted something I want it to keep).
The thing that seems to be overlooked in all of the NSA illegality discussion is quite how much the recent revelations point to their incompetence. For example, consider Heartbleed. The NSA claims that they had no knowledge of it. If you assume that they're telling the truth, then this means that they either failed to identify OpenSSL as a critical piece of software to review (odd, given how much US and other government infrastructure uses it), or they did review it and still failed to find the vulnerability. Given how obvious the vulnerability was to anyone who looked at the code (the only reason it survived so long was that there wasn't adequate code review), this implies that the people that are doing code review are incompetent. Alternatively, we can assume that they're lying and they did know about it in advance. Given the amount of government and civilian infrastructure that depends on OpenSSL, and the probability that foreign governments (and organised crime syndicates) have identified that OpenSSL is critical infrastructure and are busy fuzzing it and reviewing the code, that's somewhat problematic. So, the options are:
Their strategic analysts are incompetent for failing to identify that OpenSSL is a critical piece of software.
Their core reviewers are incompetent for failing to spot the vulnerability.
Their threat analysts are incompetent for failing to identify that defending against Heartbleed and similar attacks is more important than holding them for attack.
The idea behind lobbying is that politicians are not experts in everything and so it makes sense for them to listen to domain experts before making their decision. The first problem is that it's hard for someone who isn't an expert to differentiate between an expert and a vested interested (or an expert providing impartial advice and one providing advice promoting self interest). The second problem is that money found its way into the system and so now it's just about vested interests, experts need not apply...
Slashdot Mirror
Keep in mind that in many states, union membership is required in order to get the job.
I've never understood this about the USA, it seems to completely miss the point of unions. Here, there are often two or three unions that are competing for members, so you get the benefits of collective bargaining and the benefits of competition. Collective bargaining via a monopoly that has no incentive to represent your interests is much the same position you're in with no union at all...
The core count isn't the interesting thing about this chip. The cores themselves are pretty boring off-the-shelf parts too. I was at the ISCA presentation about this last week and it's actually pretty interesting. I'd recommend reading the paper (linked to from the press release) rather than the press release, because the press release is up to MIT's press department's usual standards (i.e. completely content-free and focussing on totally the wrong thing). The cool stuff is in the interconnect, which uses the bounded latency of the longest path multiplied by single-cycle one-hop delivery times to define an ordering, allowing you to implement a sequentially consistent view of memory relatively cheaply.
Since I'm here, I'll also throw out a plug for the work we presented at ISCA, The CHERI capability model: Revisiting RISC in an age of risk . We've now open sourced (as a code dump, public VCS coming soon) our (64-bit) MIPS softcore, which is the basis for the experimentation in CHERI. It boots FreeBSD and there are a few sitting around the place that we can ssh into and run. This is pretty nice for experimentation, because it takes about 2 hours to produce and boot a new revision of the CPU.
Half the IT profs I have come across
Where on earth have you managed to find a university that has professors in IT?
The other part, readings, are not necessarily available to the general public. If you want to get any recent consensus in a field, you need access to e.g. JSTOR, which is not freely available to the public
In computer science, most academics put their papers online (sometimes preprints if the journal doesn't allow distribution) on their own web sites for free. If you find the paper in the ACM Digital Library, and then search the web for the paper title, you can probably find it. In physics, arxiv.org does a good job of centralising preprints for people to search.
The problem with this approach is that you need a fairly good grounding in the field to be able to understand the papers. You can probably get it by reading books, but it's a bit hit-and miss. The point of a university education is not to teach you the field, it's to give you a guided tour of your ignorance. This then lets you identify the things that you should know and learn them.
Read the rest of my post. Apple's X11.app shipped on the DVD. This is not quite the same as XQuartz.
Uh, what? Quartz is the name of Apple's display system. XQuartz is an implementation of X atop Quartz. Oh, and it's not an Apple product, so was not named by Apple.
They don't, but the do some clever tricks with launchd so that the first time you try to run a program that tries to connect to an X server they'll pop up a thing asking if you want to install XQuartz. Oh, and XQuartz never shipped with OS X, Apple's X11.app did, which was a fork and rebranding of XQuartz and usually an old version by the time it shipped. The reason that they stopped bundling it was that all of the people who actually cared about X11 were installing XQuartz instead.
There are four different protocols that permit double-buffering on X. These days, pretty much everyone has converged on using XRender and manually doing the buffer management. For most of the history of X, different vendors supported different double buffering APIs and writing code that would detect which one was available and using it was painful. Double buffering isn't part of the core specification, because the RAM requirements for two copies of the frame buffer were too big for a lot of early implementations.
And outside of QT, everyone (including motif/xlib) re-invent C++ badly with home-rolled type systems which often involve pushing strings around
Funny, I always felt that the signals and slots mechanism in Qt was reinventing Objective-C badly...
The protocol is fine,
Except for the fact that it has a limited set of extensions that can be supported and a load of command numbers are used for 'core protocol' stuff that no one has used for over a decade. It has no concept of security (you can easily steal input from another application, for example).
the library isn't that horrible unless your a newbie to dev
XCB is pretty nice, but xlib is a clusterfuck. It hides interfaces that need to be used asynchronously for good performance behind synchronous API calls. It's impossible to write an application that performs well over a network and does a nontrivial amount of drawing with xlib. It is with XCB, but it requires carefully designing your toolkit for asynchronous drawing, and all modern X toolkits have too much xlib heritage to easily adapt to using XCB as it's intended to be used, rather than as a lighter-weight xlib.
The problem with network transparency in X11 is that it's done at completely the wrong place. With competing systems of the same era, such as NeWS, there was some code running in the display server for display updates. This meant that, for example, you could handle the visual feedback for a button press in the server, while transmitting the 'this button has been pressed' event to the client. In X11, you press the button but the server just sends a 'mouse clicked at x,y' message to the client, so you need a network round trip just to update the button. If you want to animate the button press, then you need to wait for network round trips to get the 'redraw finished' events. Wayland isn't a step backwards in this regard, but it's also not a step forwards.
In a modern X client, you don't really use much of the server's drawing functionality. You do store some images for compositing and will use XRender to composite them, but that's about it. The line drawing stuff can't handle antialiased lines, the text drawing stuff (aside from XRender) can't handle fonts provided by the client easily, so all you're really using the display server for is getting some texture memory and compositing it. With Wayland, you just get an OpenGL context and do the same thing. To be honest, if you're targeting X11 that's also what you should be doing for modern hardware: the rest of your drawing code most likely uses OpenGL (or something higher-level with an OpenGL back end), or just generates pixmaps, so doing the per-window compositing in OpenGL is a lot easier than doing it in a completely different API.
The only reason China 'controls' rare earth production is that they spent a few years dumping them on the market at less than the cost of extraction, pushing the other mines out of business (or, at least, into mothballing). Now that the prices are sane again, other countries are bringing their mines back online.
AMD and Apple picked it up because it's the only replacement for DVI which is capped at 1600x1200 at 60Hz or 1200p at 60Hz
Not to mention that it's royalty free. You have to pay a license for every HDMI port, you can stick any number of DisplayPort ports on a machine without paying a royalty for implementing the standard.
TFA is about monitors. The pixel density on my laptop is about double that of the 28" display on my desk. This is really noticeable for text rendering, where it's clear and crisp on the laptop screen, but looks a little blurry around the edges due to the sub-pixel AA on the bigger display. I'd love to replace the one on my desk with a 4K display once the prices become a bit less silly. And, yes, my laptop can drive a 4K display.
The Z-series and POWER are not quite separate chips. They're separate instruction decoders but they're largely the same pipelines after that. There are some tweaks, but within a generation they share more design than either does with the previous generation of the same processor.
This logic doesn't really make sense to me. Sure, 20 adds up over a few million devices, but 20isn't the important number, the profit margin is. If you're making 40 profit on each device, then 20 is a big deal. You can double your profit by cutting it. If you're making $20 profit on each device, then that 20 will make almost no difference to your bottom line. It's only an issue if you can save 20 on a lot of components in a single device.
Hopefully everyone in this thread is joking, but it's worth noting that it's not quite that clear cut. The smallest assignment that an ISP can hand out is a /64, so you can really only have 2^64 sites. IPv6 has 2^128 addresses, but a lot of the design works around having sparse routing tables. You really want each /64 to correspond to a broadcast domain, and you don't want to fragment the routing tables too much to get to the /64, so you've actually got a lot fewer addresses. A /64 per human is not enough to assign one IP per atom in the person, but it likely is enough for every device that a person may reasonably want to own and give an IP to, even if that person has a lot of injected sensor nodes.
Offer a discount for services that are willing to use v6-only. If you're hosting some back-end services in Azure, you can use v6. If it's something for only accessing within your organisation then you can possibly use v6, depending on your local connectivity. If it's something for the public then you can probably make a certain percentage of the servers v6-only and send customers with working v6 there (does Windows still set up v6 tunnels by default?).
If companies start to pay less for v6-only hosting than for v4-only or dual stack, then they're going to start pushing their customers towards the v6 servers, or making certain features v6-only to penalise ISPs that don't provide v6 connectivity by making their customers complain. That's what's going to trigger mass movement to v6.
If only your browser sent a header telling the server what your preferred language was. Oh, wait, it does, and Google still thinks that I want to go to their Japanese page when I'm in Japan. One of the many reasons I switched to DuckDuckGo a few years ago...
I've had a load of LinkedIn spam sent to mailing lists recently. The problem with this lawsuit is that it's the wrong people - LinkedIn is sending spam, they should be prosecuted, not sued by people on whose behalf they claimed to send it.
I don't understand the hate. I installed Firefox 29 and the UI seemed clean and easy to customise if you like more clutter (possibly useful if you're on a tablet, or like wasting screen real-estate with buttons that are less convenient than keyboard shortcuts, but a waste of space for most users).
A lot of Chrome's memory use comes from using multiple processes to isolate components of the web browser so that a compromise in one tab does not compromise the entire browser. Does Firefox yet do anything similar?
I switched from Chrome to Firefox on Android because as far as I can tell (after trying a load of them) Firefox is the only Android browser that doesn't have a completely fucked-up cookie management policy. It's made even nicer by the self destructing cookies addon, which lets me whitelist sites with cookies I want to keep (i.e. ones where I log in), but destroys tracking cookies and HTML5 local storage as soon as I close a tab or navigate away from a page (keeping them for a little bit in some storage that isn't visible to the page, so I can hit an undo button if I realise it's just deleted something I want it to keep).
The idea behind lobbying is that politicians are not experts in everything and so it makes sense for them to listen to domain experts before making their decision. The first problem is that it's hard for someone who isn't an expert to differentiate between an expert and a vested interested (or an expert providing impartial advice and one providing advice promoting self interest). The second problem is that money found its way into the system and so now it's just about vested interests, experts need not apply...