* ~2x improvement in transistor density, for either smaller chip size or increased transistor count * ~30% reduction in transistor switching power * >20%improvement in transistor switching speed or >5x reduction in source-drain leakage power * >10x reduction in gate oxide leakage power As a layman this sounds like a pretty massive improvement. Is this a major breakthrough or is this progress as usual?
While not a major breakthrough per se, it demonstrates intel's commitment to work on architectural evolutions again. They work both to improve the IPC (instructions per cycle) and the transistors quality, and that is a good thing. It's also written in the article that the layer are now deposited one atomic layer at a time. I believe that by demonstrating their skills in this area, they're improving our global knowledge and capabilities in microelectronics. The stupid GHz race has ended and smart people can express themselves again. Now that is a major breakthrough.
I want arrows with different size arrowheads, lines that stay attached to objects as you move them, and the ability to make them curved / bendy or straight. That's it.
Try Xfig then. It's not much maintained anymore, but it's the easiest piece of software I found to draw networks. It supports all what you described above. It supports user libraries and I regularly import Cisco's free icons into it. Its numerous exports (among which EPS, PDF and PNG) make the result cleanly integrable in virtually anything (including Word). Also, it makes efficient use of the mouse's 3 buttons and keyboard so that you don't waste your time clicking 20 times a minute on the same tool to perform the same action like in visio.
Its real weak point is text support. It supports text lines, but not text areas. This means that if you want to change a multi-line comment, you may end up modifying multiple lines.
It's both a result and a cause. The more ice will melt, the darker the area will become and the less it will reflect sun's light, which means that it will heat faster and faster. We would need to find a way to make some foam on the sea so that it starts reflecting light again and maybe we would get it colder again. Easier to say than to do...
I predict that Microsoft will soon rely on Linux existence to change their economic model. They will sell their productivity applications (eg: office) as net services, and people will use cheap boxes with cheap OS to access them. Microsoft will then put money in one Linux distro to help it package the light client as they want, then later will give money to a major web browser to stop maintaining its own for free, and will buy a few games editors to still make revenues from per-seat licenses.
With this, they will not need anymore to care about the hardware compatibility problems, software piracy nor any of their current problems. They will simply provide online services to customers. And they will possibly provide some server versions of their applications to run in large enterprises who want their users to work locally.
In ten years, when PC will boot linux from a linuxbios, to run games, firefox, and microsoft's MSN and free document viewers, people will think it was stupid to embed such a large and complicated thing such as windows on every user's PC.
The DNS addresses alphabet was defined as 37 latin chars. And people complain that to be able to enter this system, they want to extend it. That's as shameful as to define that there are not enough IPv4 addresses for everybody, so let's extend the alphabet of IPv4 addresses to 1000 values for each byte, so that we will be able to have addresses from 000.000.000.000 to 999.999.999.999. Ah, it doesn't fit a byte ? let's extend the byte to 10 bits! After all, it's already what has been done with this stupid UTF8 which breaks every mailer, including those handling it natively !
There's a system which works as a whole and which relies on sensible technical specifications. Changing some of them can have huge impacts on everything. Look at all the junk mail you all receive with unreadable chars. We really don't need to get unreadable addresses.
And yes, I do already use accents and some characters that don't enter the ASCII map in my every day life, but when I use the net, I conform to established standards. What will happen IMHO is that people using such new an unsupported features won't have an easy access to the rest of the net, so they will finally avoid it. On the other hand, spammers and phishers will intensively abuse the new mechanisms. I'm sure we'll see new security options in all common tools to allow or block usage of such domains...
And you'd need a hell of a RAID subsystem to manage handle writing at 14Tbits/sec sustained.
There's nothing to do with RAID here, it's all about telecommunications (hence the second 'T' in 'NTT'). Basically, just aggregation and switching/routing of many many many communication streams (voice, video, data,...).
That means you cannot take the Firefox logo, pull it up in your favorite image editor, and turn it into something else -- like a 16x16 grayscale icon, or putting a Superman costume on the fox, or turning it into a texture map for a game.
I still have a hard time trying to figure out why any user would need to modify a logo. Perhaps only debian users want to customize their logos to be able to say "hey, look, I'm even free to customize my logos" ? I've never seen this happen on other distros, nor Solaris, nor Mac, nor Windows. And if you need to ship a different logo for technical reasons (eg: black & white, or a 16x16 one), I think the display program should be able to perform the adaptation itself, and that if you ask Mozilla kindly enough to get new logos, they might accept to provide alternative ones for specific setups.
Well, I can understand the problem associated with patches reviews, even if I don't understand why they need so many patches to integrate a cross-platform product such as firefox. But there's something else which really makes me nervous :
From TFA :
The Firefox logo is trademarked, so Debian doesn't consider it to be Free and will not include it as part of its distribution.
I believe that everything has been said here ! As long as people with such a childish attitude will play a role in the free software world, Linux will have a hard time getting accepted by end users ! It's amazing how politics can anihilate technical merits ! What's important to these morons is not to provide a good products to their users, it's just to have their names etched (no pun intended) somewhere, associated to "the man with big balls who didn't bend in front of Firefox". Thanks to these "smart" people, there will be yet another browser.
What I find sad is that I recommended debian it to skilled friends who did not want to install bloated distros, and as they followed my advice, they now have notebooks which approximately work, because of all the things debian don't consider to be Free and simply removes (please note the touch of ideology behind the uppercase 'F' in 'Free'). It is so much annoying to them that they often consider switching away from this distro. Sometimes, I wonder if we'll have to wait for "Debian is dying" posts on Slashdot for this stupidity to end... Oh, it seems to be getting closer : http://www.linux-watch.com/news/NS7543606709.html
Many recent motherboard use 12V to control voltage regulators' MOSFETs gates because the higher the voltage, the lower the internal resistance, so the higher the efficiency. 5V is generally too low to achieve good efficiency, but 12V is fine. From 12V, the MB can produce 3.3V and 1.xxx Volt for the CPU. It's easy to also provide 5V on the MB.
Would be interesting to know what the efficiency is of a 380 -> 12/5 DC-DC converter, compared to a traditional 110 AC -> 12/5 DC converter.
It will be better, because your 110 AC (or 230 AC here) -> 12/5 DC in fact does 110 AC -> 110 DC -> 12/5 DC. Every switching power supply units first convert AC to DC, and lose about 1.2V (converted in heat) due to the input diode bridge. Then, the ultra-huge capacitors will not be needed anymore, and the airflow within the PSU will be better and will keep it cooler (thus, improve its power conversion factor). Also, the higher the voltage, the lower the current, so the lower the losses in the wires and transformers.
In fact, I'm not surprized they choose 380V. Most high voltage components (diodes, capacitors, FETs) are designed for 400V. 380V will be below this limit so compatible, while it will optimize the current to the lowest possible levels.
Killer shaves off these extra milliseconds while also prioritizing gaming network packets. ICMP Ping that can be run from a DOS prompt (ICMP Ping) does not actually measure the overhead of the operating system's network stack.
while this is a utterly bogus claim, basically what they say is "don't worry if you don't notice any change, that's expected". Considering that a gamer's machine can easily eat 2-4 Gbps of UDP traffic at 100% CPU, I suspect that 1 Mbps would at most eat 0.025%, which should not be noticeable at all. By the way, I find it funny that this card pretending to increase performance is still limited to a PCI bus, limited to less than 1 Gbps in+out, and with latencies as high as 4 microseconds at 33 MHz.
Have you seen the design ? It's just a card for hardware modders, it seems to even have leds. I suspect that the card will basically do nothing but the people running it will say "look at my NIC, it's beautiful and blazingly fast, have you seen how faster I run ?"...
As long as they swindle those stupid people, I don't really mind...
Every disk gets full after about 1-1.5 month. It's an unbreachable law, true for every disk that sees some use. Capacity is irrelevant, the time is pretty much constant.
That doesn't match my experience. I've had 54 GB (4*18, RAID5) for years to store all my projects. They were always full, I regularly had to remove old extracted tar archives and make distclean in some source directories. A year ago, I've upgraded to 146 GB (3*73, RAID5) and now my space went up to 55 GB and has stabilized there for a year now. So my needs seem to be 55 GB, whatever the time. And I'm pretty sure that I'm not the only one in this situation.
I don't think manufacturers have any business preventing me from running my own code on hardware I purchased
It's now necessarily you they prevent from running your code on the hardware you purchased. In the appliances world, they try to prevent others from running their software on the hardware you paid for. Signing the software is the only solution against rootkits and such things. I agree with you that the purchaser needs to be able to replace everything if he wants, but we must find a solution for him to be alerted if the content changes. That is the real problem that appliance companies are facing. If a manufacturer of wifi routers uses linux, and those routers finally get su much intruded and modified that it results in millions of zombies around the world, spammers will not attack windows anymore. Even worse for mobile phones !
The purchaser must be able to disable protection (eg: remove a jumper in the box), but remote attackers should be prevented from doing so without the owner knowing it.
Software support is another aspect that can be solved quite easily by checking software signatures when the phone rings.
Please stop thinking that the software world is on your PC. Embedded systems are everywhere. Your mobile phone might already be running Linux, and while you might find it funny to install bash on it, you might not find funny at all that someone else installs a packet duplicator to sniff all your conversations.
That's why it is important to provide means to the hardware manufacturer to block modifications, while leaving access to the real owner.
You use XOR to clear a register. XOR CX, CX sets the CX register to 0. It is faster than MOV CX, 0.
wrong ! The instruction is shorter (2 bytes instead of 3 or 5, depending on the word size). But it sets flags on return while MOV CX,0 will keep them intact, so the CPU cannot reorder instructions around a XOR. In the best case (no dependency), they will all need 1 cycle. In the worst case, XOR will stall the pipeline while MOV will not.
ssh-agent is a solution to grant any process on the system full access to a means of authenticating through your private key.
Any process on the system running as you or as root, you mean.
Or as yourself. It will be a real joy for virus writers when they will start to attack unix mail and web clients. We are currently laughing hard at windows with the users' address book sent all over the world by viruses, now you will simply open all your hosts to the rest of the world thanks to your agent and one undetected input relay running in your browser. Kind of smart !
And to reply to the other poster, yes I'm running all days with a dozen of terminals, and quite frankly, typing the same passphrase 30 times a day trains you enough to type if without even thinking about it.
Personnaly, I never understood how talented SSH developers came to the conclusion that they needed to invent such a crappy thing : ssh-agent. And I've seen people use it, the same people who put their private keys on USB sticks to ensure that nobody will steal them, but who are not afraid of collegues having root access on their machine...
ssh-agent is a solution to grant any process on the system full access to a means of authenticating through your private key. No comment ! There's nothing difficult in typing a passphrase each time you connect to a remote site !
I have pity on those admin that are forced to maintain Microsoft solutions against their will
You don't understand : there are many incompetent windows admins which are asked to deploy a web server. What do you think they do ? They click on the Start button, searching for anything which looks like 'web server' and start it, then click on a few boxes and believe they are secure.
Don't imagine that a windows admin will suddenly choose to take weeks to try to discover how linux and apache work, then install it to impress their boss !
This is called "reverse engineering". We had to follow the copper tracks on any PCB to draw the equivalent schematics on paper and understand what it did. You had to use an ohmmeter to guess whether a transistor was NPN or PNP, and you had to think long to figure out what an IC could be. Now, I find google far easier for this. If I had it back then, I would not have searched so long for small information like this. I even remember litteraly spending months manually injecting opcodes in others' machines to find undocumented CPU instructions because I had no documentation. The same is true for DOS and BIOS interrupts, while right now, a quick google search will give you all the functions supported by every version and their bugs.
I know I could have done 100 times more things with todays tools, but sometimes I wonder whether I would have had the motivation, because there's absolutely no challenge.
I think the best solution would be to have Nagle's on by default to address these issues and having a simple system call flush() that forces the transmission of a segment to be used whenever ever you write a small buffer with time-sensitive data.
Not exactly, you'd better need a send() flag to tag data to be sent immediately. No need to slow down your application with another syscall.
To get around the above problems, I came up with the following scheme: Leave Nagle's algorithm enabled, but create a FlushSocket() function that merely disables Nagle on the socket, then calls send() on the socket with a 0-byte buffer, then enables Nagle again.
I tried this in the past and it was not that good because of the added syscalls. In a pure network application, your worst ennemy are syscalls. And by avoiding this trick and carefully grouping your data into large writes, you both reduce the number of syscalls caused by the fewer calls to write(), and reduce syscalls by removing two setsockopt(). It's a win-win, and I'm never considering going back.
Slashdot Mirror
Willy
I want arrows with different size arrowheads, lines that stay attached to objects as you move them, and the ability to make them curved / bendy or straight. That's it.
Try Xfig then. It's not much maintained anymore, but it's the easiest piece of software I found to draw networks. It supports all what you described above. It supports user libraries and I regularly import Cisco's free icons into it. Its numerous exports (among which EPS, PDF and PNG) make the result cleanly integrable in virtually anything (including Word). Also, it makes efficient use of the mouse's 3 buttons and keyboard so that you don't waste your time clicking 20 times a minute on the same tool to perform the same action like in visio.
Its real weak point is text support. It supports text lines, but not text areas. This means that if you want to change a multi-line comment, you may end up modifying multiple lines.
Willy
I mean the result of, RESULT OF! Oh good lord.
It's both a result and a cause. The more ice will melt, the darker the area will become and the less it will reflect sun's light, which means that it will heat faster and faster. We would need to find a way to make some
foam on the sea so that it starts reflecting light again and maybe we would get it colder again. Easier to say
than to do...
I predict that Microsoft will soon rely on Linux existence to change their economic model. They will sell their productivity applications (eg: office) as net services, and people will use cheap boxes with cheap OS to access them. Microsoft will then put money in one Linux distro to help it package the light client as they want, then later will give money to a major web browser to stop maintaining its own for free, and will buy a few games editors to still make revenues from per-seat licenses.
With this, they will not need anymore to care about the hardware compatibility problems, software piracy nor any of their current problems. They will simply provide online services to customers. And they will possibly provide some server versions of their applications to run in large enterprises who want their users to work locally.
In ten years, when PC will boot linux from a linuxbios, to run games, firefox, and microsoft's MSN and free document viewers, people will think it was stupid to embed such a large and complicated thing such as windows on every user's PC.
Willy
I'm really fed up with those stupidities.
The DNS addresses alphabet was defined as 37 latin chars. And people complain that to be able
to enter this system, they want to extend it. That's as shameful as to define that there
are not enough IPv4 addresses for everybody, so let's extend the alphabet of IPv4 addresses to
1000 values for each byte, so that we will be able to have addresses from 000.000.000.000 to
999.999.999.999. Ah, it doesn't fit a byte ? let's extend the byte to 10 bits! After all, it's
already what has been done with this stupid UTF8 which breaks every mailer, including
those handling it natively !
There's a system which works as a whole and which relies on sensible technical specifications.
Changing some of them can have huge impacts on everything. Look at all the junk mail you all
receive with unreadable chars. We really don't need to get unreadable addresses.
And yes, I do already use accents and some characters that don't enter the ASCII map in my
every day life, but when I use the net, I conform to established standards. What will happen
IMHO is that people using such new an unsupported features won't have an easy access to the
rest of the net, so they will finally avoid it. On the other hand, spammers and phishers will
intensively abuse the new mechanisms. I'm sure we'll see new security options in all common
tools to allow or block usage of such domains...
That's becoming very very sad.
How does one encrypt a bit, exactly?
I guess by changing it to any other value?
And you'd need a hell of a RAID subsystem to manage handle writing at 14Tbits/sec sustained.
...).
There's nothing to do with RAID here, it's all about telecommunications (hence the second 'T' in 'NTT'). Basically, just aggregation and switching/routing of many many many communication streams (voice, video, data,
Willy
I still have a hard time trying to figure out why any user would need to modify a logo. Perhaps only debian users want to customize their logos to be able to say "hey, look, I'm even free to customize my logos" ? I've never seen this happen on other distros, nor Solaris, nor Mac, nor Windows. And if you need to ship a different logo for technical reasons (eg: black & white, or a 16x16 one), I think the display program should be able to perform the adaptation itself, and that if you ask Mozilla kindly enough to get new logos, they might accept to provide alternative ones for specific setups.
willy
From TFA :
I believe that everything has been said here ! As long as people with such a childish attitude will play a role in the free software world, Linux will have a hard time getting accepted by end users ! It's amazing how politics can anihilate technical merits ! What's important to these morons is not to provide a good products to their users, it's just to have their names etched (no pun intended) somewhere, associated to "the man with big balls who didn't bend in front of Firefox". Thanks to these "smart" people, there will be yet another browser.
What I find sad is that I recommended debian it to skilled friends who did not want to install bloated distros, and as they followed my advice, they now have notebooks which approximately work, because of all the things debian don't consider to be Free and simply removes (please note the touch of ideology behind the uppercase 'F' in 'Free'). It is so much annoying to them that they often consider switching away from this distro. Sometimes, I wonder if we'll have to wait for "Debian is dying" posts on Slashdot for this stupidity to end... Oh, it seems to be getting closer : http://www.linux-watch.com/news/NS7543606709.html
Willy
Many recent motherboard use 12V to control voltage regulators' MOSFETs gates because the higher the voltage, the lower the internal resistance, so the higher the efficiency. 5V is generally too low to achieve good efficiency, but 12V is fine.
From 12V, the MB can produce 3.3V and 1.xxx Volt for the CPU. It's easy to also provide 5V on the MB.
Tyan had a 4-sockets K8 motherboard for 1U servers. You would have 16 cores/1U, which means 80 cores per 5U. Definitely better !
willy
Would be interesting to know what the efficiency is of a 380 -> 12/5 DC-DC converter, compared to a traditional 110 AC -> 12/5 DC converter.
It will be better, because your 110 AC (or 230 AC here) -> 12/5 DC in fact does 110 AC -> 110 DC -> 12/5 DC. Every switching power supply units first convert AC to DC, and lose about 1.2V (converted in heat) due to the input diode bridge. Then, the ultra-huge capacitors will not be needed anymore, and the airflow within the PSU will be better and will keep it cooler (thus, improve its power conversion factor). Also, the higher the voltage, the lower the current, so the lower the losses in the wires and transformers.
In fact, I'm not surprized they choose 380V. Most high voltage components (diodes, capacitors, FETs) are designed for 400V. 380V will be below this limit so compatible, while it will optimize the current to the lowest possible levels.
Willy
From the FAQ :
Killer shaves off these extra milliseconds while also prioritizing gaming network packets. ICMP Ping that can be run from a DOS prompt (ICMP Ping) does not actually measure the overhead of the operating system's network stack.
while this is a utterly bogus claim, basically what they say is "don't worry if you don't notice any change, that's expected". Considering that a gamer's machine can easily eat 2-4 Gbps of UDP traffic at 100% CPU, I suspect that 1 Mbps would at most eat 0.025%, which should not be noticeable at all. By the way, I find it funny that this card pretending to increase performance is still limited to a PCI bus, limited to less than 1 Gbps in+out, and with latencies as high as 4 microseconds at 33 MHz.
Have you seen the design ? It's just a card for hardware modders, it seems to even have leds. I suspect that the card will basically do nothing but the people running it will say "look at my NIC, it's beautiful and blazingly fast, have you seen how faster I run ?"...
As long as they swindle those stupid people, I don't really mind...
Willy
Every disk gets full after about 1-1.5 month. It's an unbreachable law, true for every disk that sees some use.
Capacity is irrelevant, the time is pretty much constant.
That doesn't match my experience. I've had 54 GB (4*18, RAID5) for years to store all my projects. They were always full, I regularly had to remove old extracted tar archives and make distclean in some source directories. A year ago, I've upgraded to 146 GB (3*73, RAID5) and now my space went up to 55 GB and has stabilized there for a year now. So my needs seem to be 55 GB, whatever the time. And I'm pretty sure that I'm not the only one in this situation.
I don't think manufacturers have any business preventing me from running my own code on hardware I purchased
It's now necessarily you they prevent from running your code on the hardware you purchased. In the appliances world, they try to prevent others from running their software on the hardware you paid for. Signing the software is the only solution against rootkits and such things. I agree with you that the purchaser needs to be able to replace everything if he wants, but we must find a solution for him to be alerted if the content changes. That is the real problem that appliance companies are facing. If a manufacturer of wifi routers uses linux, and those routers finally get su much intruded and modified that it results in millions of zombies around the world, spammers will not attack windows anymore. Even worse for mobile phones !
The purchaser must be able to disable protection (eg: remove a jumper in the box), but remote attackers should be prevented from doing so without the owner knowing it.
Software support is another aspect that can be solved quite easily by checking software signatures when the phone rings.
Please stop thinking that the software world is on your PC. Embedded systems are everywhere. Your mobile phone might already be running Linux, and while you might find it funny to install bash on it, you might not find funny at all that someone else installs a packet duplicator to sniff all your conversations.
That's why it is important to provide means to the hardware manufacturer to block modifications, while leaving access to the real owner.
Willy
You use XOR to clear a register. XOR CX, CX sets the CX register to 0. It is faster than MOV CX, 0.
wrong !
The instruction is shorter (2 bytes instead of 3 or 5, depending on the word size). But it sets flags on return while MOV CX,0 will keep them intact, so the CPU cannot reorder instructions around a XOR. In the best case (no dependency), they will all need 1 cycle. In the worst case, XOR will stall the pipeline while MOV will not.
ssh-agent is a solution to grant any process on the system full access to a means of authenticating through your private key.
Any process on the system running as you or as root, you mean.
Or as yourself. It will be a real joy for virus writers when they will start to attack unix mail and web clients. We are currently laughing hard at windows with the users' address book sent all over the world by viruses, now you will simply open all your hosts to the rest of the world thanks to your agent and one undetected input relay running in your browser. Kind of smart !
And to reply to the other poster, yes I'm running all days with a dozen of terminals, and quite frankly, typing the same
passphrase 30 times a day trains you enough to type if without even thinking about it.
Personnaly, I never understood how talented SSH developers came to the conclusion that they needed to invent such a crappy thing : ssh-agent. And I've seen people use it, the same people who put their private keys on USB sticks to ensure that nobody will steal them, but who are not afraid of collegues having root access on their machine ...
ssh-agent is a solution to grant any process on the system full access to a means of authenticating through your private key. No comment ! There's nothing difficult in typing
a passphrase each time you connect to a remote site !
--
willy
high performance free load balancing solution - http://w.ods.org/haproxy/
780 million years ago, there was nothing close to a monkey on earth. There wasn't even any ammonites IIRC
IIRC ?
No monkeys, no ammonites, but you were there ? Damn ! I'm wondering what you're looking like !
should be applied next, otherwise the bug is still there
I have pity on those admin that are forced to maintain Microsoft solutions against their will
You don't understand : there are many incompetent windows admins which are asked to deploy a web server. What do you think they do ? They click on the Start button, searching for anything which looks like 'web server' and start it, then click on a few boxes and believe they are secure.
Don't imagine that a windows admin will suddenly choose to take weeks to try to discover how linux and apache work, then install it to impress their boss !
He did not say he wanted to kill init, but a process named "1" :
killall -KILL 1
willy
This is called "reverse engineering". We had to follow the copper tracks on any PCB to draw the equivalent schematics on paper and understand what it did. You had to use an ohmmeter to guess whether a transistor was NPN or PNP, and you had to think long to figure out what an IC could be. Now, I find google far easier for this. If I had it back then, I would not have searched so long for small information like this. I even remember litteraly spending months manually injecting opcodes in others' machines to find undocumented CPU instructions because I had no documentation. The same is true for DOS and BIOS interrupts, while right now, a quick google search will give you all the functions supported by every version and their bugs.
I know I could have done 100 times more things with todays tools, but sometimes I wonder whether I would have had the motivation, because there's absolutely no challenge.
Willy
I think the best solution would be to have Nagle's on by default to address these issues and having a simple system call flush() that forces the transmission of a segment to be used whenever ever you write a small buffer with time-sensitive data.
Not exactly, you'd better need a send() flag to tag data to be sent immediately. No need to slow down your application with another syscall.
willy
To get around the above problems, I came up with the following scheme: Leave Nagle's algorithm enabled, but create a FlushSocket() function that merely disables Nagle on the socket, then calls send() on the socket with a 0-byte buffer, then enables Nagle again.
I tried this in the past and it was not that good because of the added syscalls. In a pure network application, your worst ennemy are syscalls. And by avoiding this trick and carefully grouping your data into large writes, you both reduce the number of syscalls caused by the fewer calls to write(), and reduce syscalls by removing two setsockopt(). It's a win-win, and I'm never considering going back.
willy