Using 100% shared memory for video causes small lags and stuters all over the place, except in the latest generation of integrated video. Even doing simple day to do things like scrolling, or selecting a menu has tended to cause this. Main memory bandwidth is the biggest bottleneck to the CPU, and the latency added during the moments when video memory needs updates is what causes the lag.
Given two identical machines, except one with shared memory and one with independent video memory, the first will be perceived as significantly slower, because typical users don't judge machines by their floating point performance or any number crunching benchmark, but by how responsive the interface feels.
The latest and only existing generation of integrated video from ATI and Nvidia are a different story though, especially when paired with an AMD processor with it's built in memory controller. Their not silky smooth like when having separate video memory, but the problem is much less noticeable.
Apple has for a long time offered what is ATI's low end for the PC. Most PCs don't ship with as nice graphics as ATI's low end though, and Macs have come with independent video memory, unlike the many PC's that ship with integrated video through Dell and the like.
The latest generation of integrated video is much better though, and I can see the latest offerings from ATI, Nvidia and Intel being sufficient for most non-gamers, as long as they have at at least 32MB of independent memory. I know ATI's chipset supports this, but I'm not sure about the others. Apple has gone with Intel integrated video for their new Mac Mini, but I don't know if it has independent memory.
They are about offering more about bang than the other guy for your buck. The midrange $150-$200 range is where you get the most for your money, and each time one competitor offers a better value, the other can't afford to sit back for too long. The midrange GPU segment is one incredibly efficient market and the that is why there are these frequent releases. Each company is fighting to stay ahead.
One reason for this is that most midrange buyers are enthusiasts, and judging by the # of comments for a product on newegg, one can see that as soon as a better value is offered by a new chip, sales quickly shift towards it. The Nvidia 6800 GS was selling like hotcakes for just the tiny stopgap period it was put out, just to best the ATI x800GTO until the 7600 GT showed up.
I'm shopping for a card for a friend now, and have noticed that the midrange is good, but for high resolution play at 1600x1200 or 1920x1200, the midrange is barely cutting it now, so it becomes important to get the most bang for your buck, especially if you have an LCD with native high res and want to maintain quality. The new 7600 GT is about 15% faster than the 6800 GS, even w/ a 128 bit memory bus, and definitely hits a sweet spot at $190. It should run most popular titles comfortably at 1920x1200 and has next generation shader 3.0, unlike ATI's offerings below $200.
Unfortunately for ATI, they haven't offered the best midrange value since their 9xxx line. ATI took Nvidia's crown a while back but Nvidia has had it back for some time now.
A data bus that is. All electronic devices should plug into the home's "net" as easily as they jack into power, through a unified connector, or through the power line itself. A gateway appliance could allow permitted components within the home's net talk over the internet, but the home net wouldn't have to be super fast, just reliable.
I know similar ideas exist, like X10, but X10 is not very sophisticated. The whole business of getting compatible appliances to talk to each other automatically is also a sticky mess as I've seen it laid out, even if they understand XML and can exchange metadata. For example, how does the TiVo know if the cable box it is talking to is the one it is sitting under, and not the one upstairs? I suppose you could list all the cable boxes present at the home in the TiVo menu, but there are a couple problems with this. Not every device that will use the home net will have a screen, and those that do will present different interfaces for each user to figure out.
One solution would be to have a handheld device that "marries" one device to another. Light switch over here, meet light fixture from over there. This would also add some security to the system, and could avoid excess broadcasts from relying soley on a discovery protocol. An XML based discovery protocol could be optional, with a simplified but limited binary protocol for "dumb" devices like switches and fixtures. Ideally, devices should be required to talk first to the handheld before being recognized on the network.
Now most people will likely question the utility of such a setup, brushing it aside as an unneeded complexity, but it can make things easier through little usability enhancements that come with information sharing.
Examples - -a home phone can show missed calls or who left messages, to your computer or cell phone -climate control systems can adjust output based on occupancy, by adjusting the thermostat directly
-without wiring a sensor back to the original zone controller -your A/V components can recognize each other
-auto-select inputs on the TV and receiver if you play a DVD
-output the right resolution and format
-DVR + cable box can work together to schedule recordings -flash the lights if the phone rings and the radio is cranked -page your phone when the wash is done, same for the oven timer -start your coffee maker from bed, be notified when it is done
It seems every major appliance is going to have at least a microcontoller in it in the long term, so we might as well provide a generic way to interface to those appliances so they can be controlled and provide status information. The infrastructure for a device to integrate information from different devices and then to use that intelligence to programatically control itself and others, should ultimately reduce the effort we ourselves put into programmatic tasks (like changing a/v inputs) and allow forms of control that are now impractical because they require expensive custom infrastructure. Integration between devices will also mean better usability and better decision making, as in the case of a heating system that updates the thermostat to a lower temperature setting rather than cutting power to the valve for an unnoccupied zone. It is only a matter of time.
But to build such a system is a bit of a chicken and the egg problem. You aren't going to get a good selection of products that use this until there is a market for it, and vice versa. The commercial rewards are a long shot at present, since there are no standards.
Maybe someone with EE experience could create an open source platform for this, one that works over power lines for simplicity's sake. First off create reference schematics for simple X10 like outlets, switches, and lighting controllers, while adding the metadata these devices use to identify
I don't think IP Cameras are there yet
on
Cisco Aquires SyPixx
·
· Score: 4, Interesting
Last I looked into IP cameras weren't quite there yet, especially in terms of price/performance ratio. I started out looking for a camera with a pretty typical use case in mind - outdoor surveillance for an office building. Someone was stealing dumpsters worth a few thousand each from a new heating and air conditioning business, and it was happening at night. There is a lot of mob and union influence in the HVAC industry, and they didn't like the new guy in town's hiring practices. The cops seemed to care less so he wanted to catch them red handed.
The requirements came down to being weatherproof, and also having low light capability for night surveillance. As I began pricing things out, I found IP cameras could be had at low starting prices, around $200, but that those models were useless for real surveillance apps. Here are the pitfalls I found.
A) Most IP cameras below $400-$500 lack an auto-iris, but rather simulate one in software. If you can't mechanically restrict how much light is getting to the CCD sensor, you have to sacrifice sensitivity to the point where night time images won't be useful.
B) Many IP cameras use cheap CCD chips. In the CCTV industry people look for SONY Super-HAD and Ex-View CCD chips because of their night time sensitivity. Try finding something IP based with one of these CCDs and see what it costs you. An analog b&w SONY Super-HAD night camera can be had for $115, and a color daytime model only $185. IP Camera? About $1000. Want color and a good night picture? You need a model that uses solenoid to remove the IR cut filter when it gets dark, otherwise the night picture will be no good. Good luck finding an IP version with this at a reasonable price. The cost for a color analog camera with a mechanical day/night filter is $235.
C) Weatherproof models command a much bigger premium than their analog counterparts.
D) Network bandwidth may be an issue for large setups, as full frames are sent via mjpeg. Court precedent says that to be admissable, digital video footage must be stored as complete frames, so count out any of the mpeg codecs.
Now also figure this, whether you use an analog or IP camera you will still need a computer to store all your footage. $50 is what a 4 channel BTTV based CCTV capture board will cost you, and they are much less on ebay. In terms of software, ZoneMinder is open source and will stream compressed video across the internet while recording high quality frames locally. It supports any format ffmpeg supports, even flash video, and does things like auto-cycling and motion detection recording w/ user definable sensitivity areas.
For a 4 camera setup an IP camera solution will cost nearly 3x to 4x as much as analog. So I have judged them as being useful only for large corporate customers with deep pockets. Anyone here using ip cameras, especially for outdoor surveillance? What do you use and what did it cost?
...for just a few ip packets. Uhh, I hope the way Verizon runs their cellular business is not an indication of how they'll treat us under tiered internet, but who am I kidding?
Anyone who has compared developing applications for Verizon phones vs. Sprint/Nextel vs. Cingular knows that Verizon is simply not an option unless you have $$$ and enough clout to negotiate access. No feature that Verizon thinks they can get an extra fee for is left unlocked. DRM is built in and all applications are signed so as to grant just the permissions that have been paid for.
Compare this to Cingular and international gsm providers, who have no DRM and allow access to the phone hardware (bluetooth, gps, ringtones, other content)and the network via java. You own the hardware, you pay for network access, and use it as you will. No getting billed for every single permutation of features like with this Tivo app.
Verizon considers each application a billable "feature" in and of itself, while more open providers bill for network access and leave applications to open hardware and software.
The later architecture allows anyone to get in on the game, while the former restricts access to those that pay up. You can bet that development companies who pony up for access will need to make a return asap, and so will be pushed towards making applications that maximize return quickly. This will only lead to fewer experimental ideas attempted, and fewer niche applications being developed.
If \.'ers want to support more open cell standards I'd suggest looking into Cingular, who at first advertised themselves years ago as "the company the support self expression" - of course no one got it. I hear their network has gotten much wider since the AT&T merger so they are worth a shot.
Telcos and cable companies have been given exclusive rights to wire our neighborhoods through common carrier status, so they should have to play by our rules. If anyone were allowed to bring physical lines into our homes, the market would quickly weed out these attempts to nickle and dime us with tiers. Of course this isn't possible, so we need other ways to foster competition so that bandwidth is brought to us as cheaply and reliably as possible.
To do this, we should mandate that Verizon, SBC, and other cable/phone providers be forced to allow third parties to offer bandwidth from the third party's own backbone, just as is done with DSL. This would also be similar to what is done in energy sales, where many companies offer gas & electricity, but one entity is charged with collection and distribution.
Forget network neutrality, just allow third parties to supply bandwidth while setting their own rates and being charged a reasonable fee for distribution. This would remove temptation to exaggerate the bandwidth crunch, because when a third party X is able to offer more per dollar (up to the limits of what the residential area net supports), people will ask why the incumbents can't do the same.
We already have two distributors, cable and phone companies, so there is competitive incentive to keep the residential lines up to date as Verizon is doing with fiber right now, but it might not hurt to allow a third player into common carrier status either, as a generic provider of infrastructure for bandwidth, without tie-ins to these old business models.
If past battles over DSL are any indication though, the telcos want third parties out. This to me shows that they don't really appreciate the monopoly they have been given, and that politicians need to get a clue stick and beat these fools down. If the few that have the privilege of building infrastructure start fixing prices (with or without tiers), we won't have competitive alternatives. Nothing competes with wired service, so it will not simply be a matter of not buying if your are unhappy with the service.
I wonder what will happen to third party DSL providers once the switch to fiber is done. A verizon rep already came by a few weeks ago to say that the copper line from the pole to my house will soon be replaced with a fiber hookup, even though I don't get net access from them. How are the rules for third party providers written w/ respect to fiber?
I think the best way to address tiers is to allow it, as long as we can choose from alternative open-ended bandwidth sources that come through the same lines of distribution. This would put a much needed fire under the incumbents asses to offer us the best deal possible. After all, we granted them these lines of distribution. Also again, government mandated neutrality should probably be avoided, as this will only choke the market in a different manner.
There are a couple applications that PS3 will absolutely blaze at as a media center type machine, like decoding new formats and transcoding HD content. The PS3's 7 CELL processors will function as a much faster than realtime encoder for mpeg4 and any format in the forseeable future. DVR like functions and transcoding for portable device should become a much less painful affair.
Also picture the CELL behind sound processing for home theater. Audiophiles who are toying w/ DSPs know what I am talking about, they can be used to compensate for less than ideal room conditions, like excessive echoing from hard floors/walls and weird speaker placement, among other things. If PS3 integrated a microphone into a remote, it could automate many of these corrections. The same tech can also process sound in real time to make it sound plausible that it is coming from say, underwater, behind a thick steel plate, or inside a cave.
As far as games go, while programmers may have trouble parallelizing specific operations, there are many parallelizable events within a game environment. It will only be a matter of time till those CELLs are used for creative effects no other system can match. PS2 only had few titles like Kingdom Hearts that used it's emotion engine fully, but I have a hunch this time it will be different.
The CELL will also allow for some new control methods, as we've never had anything capable of analyzing raw sensor data at the rate PS3 will be able too. Couple video analysis with an accelerometer/gyro based controller and the possiblities really expand. How about an active EEG helmet so the PS3 can sense your fear?
Applications that used to cost an order of magnitude more because of the expensive, low volume DSPs used, as in RADAR, will be possible, because the sensors themselves don't represent much of the cost. Speaking of radio waves, I see no reason why the CELL couldn't function as the basis for a software radio, allowing it to analyze and emulate many wireless standards.
I wouldn't underestimate CELL, it can do many things that used to require specialized hardware to be built, with just software upgrades.
Famed investor Peter Lynch says to start worrying when companies "diworsify" as he calls it. When companies find themselves unable to gain additional marketshare in the industries they already compete in, they tend to go around buying into other industries at inflated prices. Often they buy into industries that require different know how to run effectively, and many botch the job once things have played out in a few years. Think of all the internet startups that were overvalued, bought up, and mismanaged. The same thing happens in other fields as well.
There is incentive on the part of executives to diversify, as managers can then get promoted, whereas there was little room to grow before. In the short term the stock goes up and executive salaries also rise, but in the long term, mismanaged divisions only weigh a company down, offsetting profits from the healthy divisions and hurting long term investors.
There is a rising market for nuclear reactors, so this might turn out to be good thing for Toshiba, but I'd do more research before plopping down some coin for Toshiba stock.
Well it goes beyong user privacy. The government is requesting this information to help revive COPA (the Child Online Protection Act), which google may not want or agree with. Many net users disagreed with the bill when it first passed, so this wouldn't be all that surprising.
Also, if there are no ip addresses revealed I don't see the utility of the data. Even if ip addresses were released, wouldn't the government have to also subopena ISPs for the associated personal data - at least if they wanted to find out if there was lots of browsing from houses where minors live? So what exactly are they trying to conclude? Any statisticians willing to speculate?
If they are looking for queries for child porn how will they differentiate just what queries were for child porn? Many porn titles use the word young and teen, would they exclude those or use them to build up their numbers? Also since child porn is already illegal, how would a wider reaching COPA style law changing anything?
This administration has also said that it is gearing up to go after parts the porn industry directly, as they feel certain companies have pushed the boundaries too far on indecency. I thought these things were state/community matters, but I guess not. From the sound of things, I believe they will be announcing indictments against deviant pornagraphy involving consenting adults, so I can't say I trust their intentions.
This administration has also opposed the creation of the.xxx domain, in fear that it would give porn some kind of legitimacy. Of course if the goal is to protect children,.xxx sure would make filtering a lot easier. Even requiring sites to flag content not suitable for minors would help filtering, but they want COPA style censorship. Their argument will be that filters are inneffective, either because they can be bypassed by a saavy teenager or because they simply don't catch everything.
So... Why not require websites to self-flag, combined with passworded filtering options that operate at the ISP end?
COPA style censorship requires age verification by the site, which usually involves identification via credit card. It also lets some federal office pick where to draw the line (power + low accountability is no good) and would chill many adults from viewing material they want. This is because credit card records aren't always that private, and viewing porn requires incredible discretion because of social mores. The new registered airline traveler program requires disclosure of credit card records, and the Total Information Awarness program that this administration supported called for indexing everyones'. Spouses may also see credit card records, along with bank employees. COPA would certainly make adults think twice before viewing porn online, and I think this is part of a hidden agenda.
The court ruled when striking down COPA that protecting minors can be done w/o such overreaching. I don't see the effort to revive COPA as anything more than these these puritan officials trying to nanny us. There are ways for parental control to be achieved w/o sacrificing the freedom of adults.
I applaud google for refusing the subopena, and think this will pan out well for the shareholders, as maintaining user trust will only help in the long term.
So we turn to market forces and technology. In terms of market forces, I have consistently criticized both the recording industry and now Hollywood for being too slow to adapt to the digital world. For years, I was told that you simply "can't compete with free." I believe the advent of online digital music sales has shown that to be a falsehood. Music is still available for free through illegal downloading services, yet Apple's iTunes music service has sold hundreds of millions of downloads. At the same time, it is a fact that rampant piracy of songs on the internet persists. The market doesn't take care of everything.
No the market doesn't take care of everything, but how will blocking analog to digital converters for video thwart piracy? Such technology isn't foolproof, and one technically knowledgeable person is all that it will take for illegal p2p distribution to take place. Analog signals aren't about to change and old equipment will remain around as well. On the other hand barring A2D equipment will obliterate fair use. Joe Blow who pays for HBO and wants to timeshift something to watch later on his ipod, he's the one who will be nickel and dimed for something he already pays for. This bill is there to stuff money into the pockets of the industry fat cats, plain and simple.
So we turn to technology. We are now in a debate about whether copyright owners have a right to limit uses of the content they own the rights to; if so, what limitations are appropriate and how we ensure that those limitations are respected. One answer is found in the iPod and iTunes music store. Music purchased through the music store is subject to limitations which iTunes and the iPod attempt to ensure aren't violated.
So you want the equivalent of iTunes style DRM for analog signals, I see. Why can't that work? People seem happy with iTunes, why should protecting analog signals be any different? It's different because iTunes like protection is optional, and doesn't effect users and markets outside the scope of protected content. The difference is akin to that between positive and negative rights, and there is good reason why the founding fathers preferred the later. Currently we are free to convert and manipulate content as we choose unless there is DRM, but what this legislation proposes is that we should not be able to convert and manipulate content as we choose by default. Obviously there are many uses for converting and manipulating content other than piracy, and this legislation casts far too wide a net. For an experienced congressman to miss this - one who sounds well intentioned enough - is surprising.
Perhaps he doesn't sufficiently understand potential of technology outside a narrow sphere, due to lack of exposure. The founding fathers also didn't want congress meeting too often, lest they become busybodies, out of touch with their constituency. They wanted our leaders to be out and about doing civilian things, so they could better understand the will of the people. Todays professional politicians are far too isolated and "handled" by staff, when they should should be out experiencing America and getting intimate with its problems.
------ For reference - From Wikipedia - Negative right
A negative right is a right, either moral or decreed by law, to not be subject to an action of another human being (usually abuse or coercion). Negative rights are sometimes contrasted with positive rights, which are rights to be provided with something by the positive action of another. The former proscribe action, while the latter prescribe action.
One example of a negative right is the First Amendment of the United States Constitution, which acknowledges it unlawful for the government to restrict a person's speech. A law requiring another person to provide him with a microphone would codify a positive right, as would a law requiring that a person deserves police support to protect and enforce free speech.
A few people in this thread have enthusiastically recommended watchguard, and it looked like the clear winner with its appliance like simplicity, at least until I read your post. What do you think about that one, just for comparison's sake?
This is just a delicate way of saying that Novell has vested too much in R&D. So sacrifice R&D to follow technologies that are already showning wide adoption. Novell has taken the lead in introducing now popular technologies like directory services, but has had trouble keeping marketshare. Why is that? Did R&D prevent prevent Novell's customers from getting something their competitors had? What is that exactly?
It sounds to me like Novell is going the way of HP, but I hope they continue to make R&D enough of a priority.
This thing won't be a viable robotics platform until extra inputs and outputs added. Not to mention it has to stay tethered to a PC. This is screaming for a circuit board with a $10 microcontroller on it, one that can provide for analog/digital inputs & outputs, as well as some program space so it doesn't have to be tethered. One of those gumstix computers would work, but I'd prefer something cheaper, though gumstix does wireless, and that could be handy.
Another problem is that applications are limited due to roomba's form factor. It might be fun equip it with a camera and a wireless gumstix module, and then have a server do processing/control - oh for, you know, things every geek ought to have, like a personal sentry or a reconnaissance drone. I bet it could make its way pretty stealthily through the floor of an office building, if the low profile were maintained, and so that it only moved when no one was looking. Neat, but I just don't see it physically actuating to do any task besides vacuuming and pushing on objects.
I have a portable pulse oximeter sitting right next to me. It is pricey and is about 2.5" x 1.5" x 1.5". It clamps lightly around one's finger and has a numerical LED display for oxygen level and beats per minute. It's as accurate as a bedside hospital unit from what I have read. Adding one of these though would really drive up costs. Here is a pic of the unit I am talking about. $675, ouch.
Incorporating them would also require a major redesign. They clamp around an inserted finger, and this would make them harder to clean and maintain, and also make them more prone to breakage.
The non-invasive principle of operation of these is pretty neat, and might interest slashdoters. They work by shooting dual wavelengths of light through the finger, namely infra-red and a visible red color. On the other side of the finger, a sensor relays readings to a signal processor, which distinguishes between flesh, bone, and what-not based on the absorption differential between the two wavelengths, so it can isolate out variables between different kinds of fingers. The result is incredibly precise, and the LED on the front flashes in precise sync with one's pulse. I'm guessing the signal processor is a major cost, so maybe in time these will come down in price.
It's probably just policy, maybe they got burned in the past. If you think you were singled out, consider why, and if you are being honest with yourself and can think of no good reason, ahh well, your boss is just paranoid then.
Are building these worth these worth it? What is the final cost/hours spent/usability/style/performance of the finished product? Compared to say a cheaper $600 finished projector?
There could very well be people getting into trouble who did nothing wrong. I service lots of residential machines and their loaded not just with spyware, but trojans and viruses that make their way into these machines through remote and browser exploits. Some these machines need complete re-installation even though I clean up all local machine and user specific startup entries.
These I suspect have been root-kitted to act as zombies or proxies. These people have no idea what kinds of traffic is running through their machines and connection. It sounds as if such people are getting sued in some instances, but probably don't the know well enough to realize what is happening.
It doesn't seem to me that a list of bittorrent peers associated with a copyrighted file proves guilt. The environment is too insecure to guarantee who the actual source is. It seems to me the RIAA should have to prove a couple things:
1) That they downloaded the file with the copyrighted name and verified that the content is actually the copyrighted material.
2) That the activity from the IP address of the peer being charged actually represents the activity of a particular machine's owner. They would probably need to confiscate the machine for this - is this feesible? Just charging the owner of a connection seems unreliable, many machines can sit on a home or business network. Can one be held responsible for hijacked traffic running through their pipe?
Where this is headed it seems is a battle over regulating net communication. The RIAA will begin to push technical mandates through congress to make the internet more "secure," which will be difficult at best without implementing lots of centralized control and monitoring. How long till we have sign our packets with keys? Then how long till "sponsored" packets become free, while others cost?
A recent slashdot story featuring Doc Searl's opinion piece, Saving the Net from the pipeholders" sum's up this position very well. It's kind of long, but but offers an insightful view of what's ahead, and is worth reading for anyone with interest in the future net as a decentralized, unprejudiced peer to peer medium.
Why would it be disallowed? I remember DISH or some other sat provider fighting to offer a la carte, but the content producers wouldn't allow it. Why does the FCC have a role in this?
...electronically flippable license plate. This cameras work off OCR and can probably be fooled that way. A similar network is being thought up for NYC, as part of a proposal to charge tolls for using the most crowded streets at certain times, specifically around mid-town. Apparently other cities have been pretty successful w/ such a system. The New York Times ran an article on this over the last month or so.
I wonder how long it will keep records? Or would such a system look for patterns of behavior, like circling in a neighborhood known for prostitution? Seems all fine and good to track a criminal on the run, but can guarantees be made against abuse?
Is there a FPS game out there that has done a good job of keeping hacks out? I play Return to Castle Wolfenstein, based on the Quake 3 engine, and there are plenty of players who can see through walls, and others that seem to have a supernatural sense of aim. I hear even the punkbuster enabled retail version has been hacked.
It seems trusted computing may help this some in the long term, but I find it hard to trust that I'm not wasting my time fighting an impossible opponent at times.
Slashdot Mirror
Using 100% shared memory for video causes small lags and stuters all over the place, except in the latest generation of integrated video. Even doing simple day to do things like scrolling, or selecting a menu has tended to cause this. Main memory bandwidth is the biggest bottleneck to the CPU, and the latency added during the moments when video memory needs updates is what causes the lag.
Given two identical machines, except one with shared memory and one with independent video memory, the first will be perceived as significantly slower, because typical users don't judge machines by their floating point performance or any number crunching benchmark, but by how responsive the interface feels.
The latest and only existing generation of integrated video from ATI and Nvidia are a different story though, especially when paired with an AMD processor with it's built in memory controller. Their not silky smooth like when having separate video memory, but the problem is much less noticeable.
Apple has for a long time offered what is ATI's low end for the PC. Most PCs don't ship with as nice graphics as ATI's low end though, and Macs have come with independent video memory, unlike the many PC's that ship with integrated video through Dell and the like.
The latest generation of integrated video is much better though, and I can see the latest offerings from ATI, Nvidia and Intel being sufficient for most non-gamers, as long as they have at at least 32MB of independent memory. I know ATI's chipset supports this, but I'm not sure about the others. Apple has gone with Intel integrated video for their new Mac Mini, but I don't know if it has independent memory.
They are about offering more about bang than the other guy for your buck. The midrange $150-$200 range is where you get the most for your money, and each time one competitor offers a better value, the other can't afford to sit back for too long. The midrange GPU segment is one incredibly efficient market and the that is why there are these frequent releases. Each company is fighting to stay ahead.
One reason for this is that most midrange buyers are enthusiasts, and judging by the # of comments for a product on newegg, one can see that as soon as a better value is offered by a new chip, sales quickly shift towards it. The Nvidia 6800 GS was selling like hotcakes for just the tiny stopgap period it was put out, just to best the ATI x800GTO until the 7600 GT showed up.
I'm shopping for a card for a friend now, and have noticed that the midrange is good, but for high resolution play at 1600x1200 or 1920x1200, the midrange is barely cutting it now, so it becomes important to get the most bang for your buck, especially if you have an LCD with native high res and want to maintain quality. The new 7600 GT is about 15% faster than the 6800 GS, even w/ a 128 bit memory bus, and definitely hits a sweet spot at $190. It should run most popular titles comfortably at 1920x1200 and has next generation shader 3.0, unlike ATI's offerings below $200.
Unfortunately for ATI, they haven't offered the best midrange value since their 9xxx line. ATI took Nvidia's crown a while back but Nvidia has had it back for some time now.
A data bus that is. All electronic devices should plug into the home's "net" as easily as they jack into power, through a unified connector, or through the power line itself. A gateway appliance could allow permitted components within the home's net talk over the internet, but the home net wouldn't have to be super fast, just reliable.
I know similar ideas exist, like X10, but X10 is not very sophisticated. The whole business of getting compatible appliances to talk to each other automatically is also a sticky mess as I've seen it laid out, even if they understand XML and can exchange metadata. For example, how does the TiVo know if the cable box it is talking to is the one it is sitting under, and not the one upstairs? I suppose you could list all the cable boxes present at the home in the TiVo menu, but there are a couple problems with this. Not every device that will use the home net will have a screen, and those that do will present different interfaces for each user to figure out.
One solution would be to have a handheld device that "marries" one device to another. Light switch over here, meet light fixture from over there. This would also add some security to the system, and could avoid excess broadcasts from relying soley on a discovery protocol. An XML based discovery protocol could be optional, with a simplified but limited binary protocol for "dumb" devices like switches and fixtures. Ideally, devices should be required to talk first to the handheld before being recognized on the network.
Now most people will likely question the utility of such a setup, brushing it aside as an unneeded complexity, but it can make things easier through little usability enhancements that come with information sharing.
Examples -
-a home phone can show missed calls or who left messages, to your computer or cell phone
-climate control systems can adjust output based on occupancy, by adjusting the thermostat directly
-without wiring a sensor back to the original zone controller
-your A/V components can recognize each other
-auto-select inputs on the TV and receiver if you play a DVD
-output the right resolution and format
-DVR + cable box can work together to schedule recordings
-flash the lights if the phone rings and the radio is cranked
-page your phone when the wash is done, same for the oven timer
-start your coffee maker from bed, be notified when it is done
It seems every major appliance is going to have at least a microcontoller in it in the long term, so we might as well provide a generic way to interface to those appliances so they can be controlled and provide status information. The infrastructure for a device to integrate information from different devices and then to use that intelligence to programatically control itself and others, should ultimately reduce the effort we ourselves put into programmatic tasks (like changing a/v inputs) and allow forms of control that are now impractical because they require expensive custom infrastructure. Integration between devices will also mean better usability and better decision making, as in the case of a heating system that updates the thermostat to a lower temperature setting rather than cutting power to the valve for an unnoccupied zone. It is only a matter of time.
But to build such a system is a bit of a chicken and the egg problem. You aren't going to get a good selection of products that use this until there is a market for it, and vice versa. The commercial rewards are a long shot at present, since there are no standards.
Maybe someone with EE experience could create an open source platform for this, one that works over power lines for simplicity's sake. First off create reference schematics for simple X10 like outlets, switches, and lighting controllers, while adding the metadata these devices use to identify
Last I looked into IP cameras weren't quite there yet, especially in terms of price/performance ratio. I started out looking for a camera with a pretty typical use case in mind - outdoor surveillance for an office building. Someone was stealing dumpsters worth a few thousand each from a new heating and air conditioning business, and it was happening at night. There is a lot of mob and union influence in the HVAC industry, and they didn't like the new guy in town's hiring practices. The cops seemed to care less so he wanted to catch them red handed.
The requirements came down to being weatherproof, and also having low light capability for night surveillance. As I began pricing things out, I found IP cameras could be had at low starting prices, around $200, but that those models were useless for real surveillance apps. Here are the pitfalls I found.
A) Most IP cameras below $400-$500 lack an auto-iris, but rather simulate one in software. If you can't mechanically restrict how much light is getting to the CCD sensor, you have to sacrifice sensitivity to the point where night time images won't be useful.
B) Many IP cameras use cheap CCD chips. In the CCTV industry people look for SONY Super-HAD and Ex-View CCD chips because of their night time sensitivity. Try finding something IP based with one of these CCDs and see what it costs you. An analog b&w SONY Super-HAD night camera can be had for $115, and a color daytime model only $185. IP Camera? About $1000. Want color and a good night picture? You need a model that uses solenoid to remove the IR cut filter when it gets dark, otherwise the night picture will be no good. Good luck finding an IP version with this at a reasonable price. The cost for a color analog camera with a mechanical day/night filter is $235.
C) Weatherproof models command a much bigger premium than their analog counterparts.
D) Network bandwidth may be an issue for large setups, as full frames are sent via mjpeg. Court precedent says that to be admissable, digital video footage must be stored as complete frames, so count out any of the mpeg codecs.
Now also figure this, whether you use an analog or IP camera you will still need a computer to store all your footage. $50 is what a 4 channel BTTV based CCTV capture board will cost you, and they are much less on ebay. In terms of software, ZoneMinder is open source and will stream compressed video across the internet while recording high quality frames locally. It supports any format ffmpeg supports, even flash video, and does things like auto-cycling and motion detection recording w/ user definable sensitivity areas.
For a 4 camera setup an IP camera solution will cost nearly 3x to 4x as much as analog. So I have judged them as being useful only for large corporate customers with deep pockets. Anyone here using ip cameras, especially for outdoor surveillance? What do you use and what did it cost?
I mis-typed it.
...for just a few ip packets. Uhh, I hope the way Verizon runs their cellular business is not an indication of how they'll treat us under tiered internet, but who am I kidding?
Anyone who has compared developing applications for Verizon phones vs. Sprint/Nextel vs. Cingular knows that Verizon is simply not an option unless you have $$$ and enough clout to negotiate access. No feature that Verizon thinks they can get an extra fee for is left unlocked. DRM is built in and all applications are signed so as to grant just the permissions that have been paid for.
Compare this to Cingular and international gsm providers, who have no DRM and allow access to the phone hardware (bluetooth, gps, ringtones, other content)and the network via java. You own the hardware, you pay for network access, and use it as you will. No getting billed for every single permutation of features like with this Tivo app.
Verizon considers each application a billable "feature" in and of itself, while more open providers bill for network access and leave applications to open hardware and software.
The later architecture allows anyone to get in on the game, while the former restricts access to those that pay up. You can bet that development companies who pony up for access will need to make a return asap, and so will be pushed towards making applications that maximize return quickly. This will only lead to fewer experimental ideas attempted, and fewer niche applications being developed.
If \.'ers want to support more open cell standards I'd suggest looking into Cingular, who at first advertised themselves years ago as "the company the support self expression" - of course no one got it. I hear their network has gotten much wider since the AT&T merger so they are worth a shot.
Telcos and cable companies have been given exclusive rights to wire our neighborhoods through common carrier status, so they should have to play by our rules. If anyone were allowed to bring physical lines into our homes, the market would quickly weed out these attempts to nickle and dime us with tiers. Of course this isn't possible, so we need other ways to foster competition so that bandwidth is brought to us as cheaply and reliably as possible.
To do this, we should mandate that Verizon, SBC, and other cable/phone providers be forced to allow third parties to offer bandwidth from the third party's own backbone, just as is done with DSL. This would also be similar to what is done in energy sales, where many companies offer gas & electricity, but one entity is charged with collection and distribution.
Forget network neutrality, just allow third parties to supply bandwidth while setting their own rates and being charged a reasonable fee for distribution. This would remove temptation to exaggerate the bandwidth crunch, because when a third party X is able to offer more per dollar (up to the limits of what the residential area net supports), people will ask why the incumbents can't do the same.
We already have two distributors, cable and phone companies, so there is competitive incentive to keep the residential lines up to date as Verizon is doing with fiber right now, but it might not hurt to allow a third player into common carrier status either, as a generic provider of infrastructure for bandwidth, without tie-ins to these old business models.
If past battles over DSL are any indication though, the telcos want third parties out. This to me shows that they don't really appreciate the monopoly they have been given, and that politicians need to get a clue stick and beat these fools down. If the few that have the privilege of building infrastructure start fixing prices (with or without tiers), we won't have competitive alternatives. Nothing competes with wired service, so it will not simply be a matter of not buying if your are unhappy with the service.
I wonder what will happen to third party DSL providers once the switch to fiber is done. A verizon rep already came by a few weeks ago to say that the copper line from the pole to my house will soon be replaced with a fiber hookup, even though I don't get net access from them. How are the rules for third party providers written w/ respect to fiber?
I think the best way to address tiers is to allow it, as long as we can choose from alternative open-ended bandwidth sources that come through the same lines of distribution. This would put a much needed fire under the incumbents asses to offer us the best deal possible. After all, we granted them these lines of distribution. Also again, government mandated neutrality should probably be avoided, as this will only choke the market in a different manner.
There are a couple applications that PS3 will absolutely blaze at as a media center type machine, like decoding new formats and transcoding HD content. The PS3's 7 CELL processors will function as a much faster than realtime encoder for mpeg4 and any format in the forseeable future. DVR like functions and transcoding for portable device should become a much less painful affair.
Also picture the CELL behind sound processing for home theater. Audiophiles who are toying w/ DSPs know what I am talking about, they can be used to compensate for less than ideal room conditions, like excessive echoing from hard floors/walls and weird speaker placement, among other things. If PS3 integrated a microphone into a remote, it could automate many of these corrections. The same tech can also process sound in real time to make it sound plausible that it is coming from say, underwater, behind a thick steel plate, or inside a cave.
As far as games go, while programmers may have trouble parallelizing specific operations, there are many parallelizable events within a game environment. It will only be a matter of time till those CELLs are used for creative effects no other system can match. PS2 only had few titles like Kingdom Hearts that used it's emotion engine fully, but I have a hunch this time it will be different.
The CELL will also allow for some new control methods, as we've never had anything capable of analyzing raw sensor data at the rate PS3 will be able too. Couple video analysis with an accelerometer/gyro based controller and the possiblities really expand. How about an active EEG helmet so the PS3 can sense your fear?
Applications that used to cost an order of magnitude more because of the expensive, low volume DSPs used, as in RADAR, will be possible, because the sensors themselves don't represent much of the cost. Speaking of radio waves, I see no reason why the CELL couldn't function as the basis for a software radio, allowing it to analyze and emulate many wireless standards.
I wouldn't underestimate CELL, it can do many things that used to require specialized hardware to be built, with just software upgrades.
Famed investor Peter Lynch says to start worrying when companies "diworsify" as he calls it. When companies find themselves unable to gain additional marketshare in the industries they already compete in, they tend to go around buying into other industries at inflated prices. Often they buy into industries that require different know how to run effectively, and many botch the job once things have played out in a few years. Think of all the internet startups that were overvalued, bought up, and mismanaged. The same thing happens in other fields as well.
There is incentive on the part of executives to diversify, as managers can then get promoted, whereas there was little room to grow before. In the short term the stock goes up and executive salaries also rise, but in the long term, mismanaged divisions only weigh a company down, offsetting profits from the healthy divisions and hurting long term investors.
There is a rising market for nuclear reactors, so this might turn out to be good thing for Toshiba, but I'd do more research before plopping down some coin for Toshiba stock.
Well it goes beyong user privacy. The government is requesting this information to help revive COPA (the Child Online Protection Act), which google may not want or agree with. Many net users disagreed with the bill when it first passed, so this wouldn't be all that surprising.
.xxx domain, in fear that it would give porn some kind of legitimacy. Of course if the goal is to protect children, .xxx sure would make filtering a lot easier. Even requiring sites to flag content not suitable for minors would help filtering, but they want COPA style censorship. Their argument will be that filters are inneffective, either because they can be bypassed by a saavy teenager or because they simply don't catch everything.
Also, if there are no ip addresses revealed I don't see the utility of the data. Even if ip addresses were released, wouldn't the government have to also subopena ISPs for the associated personal data - at least if they wanted to find out if there was lots of browsing from houses where minors live? So what exactly are they trying to conclude? Any statisticians willing to speculate?
If they are looking for queries for child porn how will they differentiate just what queries were for child porn? Many porn titles use the word young and teen, would they exclude those or use them to build up their numbers? Also since child porn is already illegal, how would a wider reaching COPA style law changing anything?
This administration has also said that it is gearing up to go after parts the porn industry directly, as they feel certain companies have pushed the boundaries too far on indecency. I thought these things were state/community matters, but I guess not. From the sound of things, I believe they will be announcing indictments against deviant pornagraphy involving consenting adults, so I can't say I trust their intentions.
This administration has also opposed the creation of the
So... Why not require websites to self-flag, combined with passworded filtering options that operate at the ISP end?
COPA style censorship requires age verification by the site, which usually involves identification via credit card. It also lets some federal office pick where to draw the line (power + low accountability is no good) and would chill many adults from viewing material they want. This is because credit card records aren't always that private, and viewing porn requires incredible discretion because of social mores. The new registered airline traveler program requires disclosure of credit card records, and the Total Information Awarness program that this administration supported called for indexing everyones'. Spouses may also see credit card records, along with bank employees. COPA would certainly make adults think twice before viewing porn online, and I think this is part of a hidden agenda.
The court ruled when striking down COPA that protecting minors can be done w/o such overreaching. I don't see the effort to revive COPA as anything more than these these puritan officials trying to nanny us. There are ways for parental control to be achieved w/o sacrificing the freedom of adults.
I applaud google for refusing the subopena, and think this will pan out well for the shareholders, as maintaining user trust will only help in the long term.
No the market doesn't take care of everything, but how will blocking analog to digital converters for video thwart piracy? Such technology isn't foolproof, and one technically knowledgeable person is all that it will take for illegal p2p distribution to take place. Analog signals aren't about to change and old equipment will remain around as well. On the other hand barring A2D equipment will obliterate fair use. Joe Blow who pays for HBO and wants to timeshift something to watch later on his ipod, he's the one who will be nickel and dimed for something he already pays for. This bill is there to stuff money into the pockets of the industry fat cats, plain and simple.
So you want the equivalent of iTunes style DRM for analog signals, I see. Why can't that work? People seem happy with iTunes, why should protecting analog signals be any different? It's different because iTunes like protection is optional, and doesn't effect users and markets outside the scope of protected content. The difference is akin to that between positive and negative rights, and there is good reason why the founding fathers preferred the later. Currently we are free to convert and manipulate content as we choose unless there is DRM, but what this legislation proposes is that we should not be able to convert and manipulate content as we choose by default. Obviously there are many uses for converting and manipulating content other than piracy, and this legislation casts far too wide a net. For an experienced congressman to miss this - one who sounds well intentioned enough - is surprising.
Perhaps he doesn't sufficiently understand potential of technology outside a narrow sphere, due to lack of exposure. The founding fathers also didn't want congress meeting too often, lest they become busybodies, out of touch with their constituency. They wanted our leaders to be out and about doing civilian things, so they could better understand the will of the people. Todays professional politicians are far too isolated and "handled" by staff, when they should should be out experiencing America and getting intimate with its problems.
------ For reference - From Wikipedia - Negative right
A negative right is a right, either moral or decreed by law, to not be subject to an action of another human being (usually abuse or coercion). Negative rights are sometimes contrasted with positive rights, which are rights to be provided with something by the positive action of another. The former proscribe action, while the latter prescribe action.
One example of a negative right is the First Amendment of the United States Constitution, which acknowledges it unlawful for the government to restrict a person's speech. A law requiring another person to provide him with a microphone would codify a positive right, as would a law requiring that a person deserves police support to protect and enforce free speech.
A few people in this thread have enthusiastically recommended watchguard, and it looked like the clear winner with its appliance like simplicity, at least until I read your post. What do you think about that one, just for comparison's sake?
This is just a delicate way of saying that Novell has vested too much in R&D. So sacrifice R&D to follow technologies that are already showning wide adoption. Novell has taken the lead in introducing now popular technologies like directory services, but has had trouble keeping marketshare. Why is that? Did R&D prevent prevent Novell's customers from getting something their competitors had? What is that exactly?
It sounds to me like Novell is going the way of HP, but I hope they continue to make R&D enough of a priority.
This thing won't be a viable robotics platform until extra inputs and outputs added. Not to mention it has to stay tethered to a PC. This is screaming for a circuit board with a $10 microcontroller on it, one that can provide for analog/digital inputs & outputs, as well as some program space so it doesn't have to be tethered. One of those gumstix computers would work, but I'd prefer something cheaper, though gumstix does wireless, and that could be handy.
Another problem is that applications are limited due to roomba's form factor. It might be fun equip it with a camera and a wireless gumstix module, and then have a server do processing/control - oh for, you know, things every geek ought to have, like a personal sentry or a reconnaissance drone. I bet it could make its way pretty stealthily through the floor of an office building, if the low profile were maintained, and so that it only moved when no one was looking. Neat, but I just don't see it physically actuating to do any task besides vacuuming and pushing on objects.
I have a portable pulse oximeter sitting right next to me. It is pricey and is about 2.5" x 1.5" x 1.5". It clamps lightly around one's finger and has a numerical LED display for oxygen level and beats per minute. It's as accurate as a bedside hospital unit from what I have read. Adding one of these though would really drive up costs. Here is a pic of the unit I am talking about. $675, ouch.
Incorporating them would also require a major redesign. They clamp around an inserted finger, and this would make them harder to clean and maintain, and also make them more prone to breakage.
The non-invasive principle of operation of these is pretty neat, and might interest slashdoters. They work by shooting dual wavelengths of light through the finger, namely infra-red and a visible red color. On the other side of the finger, a sensor relays readings to a signal processor, which distinguishes between flesh, bone, and what-not based on the absorption differential between the two wavelengths, so it can isolate out variables between different kinds of fingers. The result is incredibly precise, and the LED on the front flashes in precise sync with one's pulse. I'm guessing the signal processor is a major cost, so maybe in time these will come down in price.
Informative post gets buried.
It's probably just policy, maybe they got burned in the past. If you think you were singled out, consider why, and if you are being honest with yourself and can think of no good reason, ahh well, your boss is just paranoid then.
Are building these worth these worth it? What is the final cost/hours spent/usability/style/performance of the finished product? Compared to say a cheaper $600 finished projector?
There could very well be people getting into trouble who did nothing wrong. I service lots of residential machines and their loaded not just with spyware, but trojans and viruses that make their way into these machines through remote and browser exploits. Some these machines need complete re-installation even though I clean up all local machine and user specific startup entries.
These I suspect have been root-kitted to act as zombies or proxies. These people have no idea what kinds of traffic is running through their machines and connection. It sounds as if such people are getting sued in some instances, but probably don't the know well enough to realize what is happening.
It doesn't seem to me that a list of bittorrent peers associated with a copyrighted file proves guilt. The environment is too insecure to guarantee who the actual source is. It seems to me the RIAA should have to prove a couple things:
1) That they downloaded the file with the copyrighted name and verified that the content is actually the copyrighted material.
2) That the activity from the IP address of the peer being charged actually represents the activity of a particular machine's owner. They would probably need to confiscate the machine for this - is this feesible? Just charging the owner of a connection seems unreliable, many machines can sit on a home or business network. Can one be held responsible for hijacked traffic running through their pipe?
Where this is headed it seems is a battle over regulating net communication. The RIAA will begin to push technical mandates through congress to make the internet more "secure," which will be difficult at best without implementing lots of centralized control and monitoring. How long till we have sign our packets with keys? Then how long till "sponsored" packets become free, while others cost?
A recent slashdot story featuring Doc Searl's opinion piece, Saving the Net from the pipeholders" sum's up this position very well. It's kind of long, but but offers an insightful view of what's ahead, and is worth reading for anyone with interest in the future net as a decentralized, unprejudiced peer to peer medium.
Why would it be disallowed? I remember DISH or some other sat provider fighting to offer a la carte, but the content producers wouldn't allow it. Why does the FCC have a role in this?
Leave it up to the Japanese to come up with this one.
...electronically flippable license plate. This cameras work off OCR and can probably be fooled that way. A similar network is being thought up for NYC, as part of a proposal to charge tolls for using the most crowded streets at certain times, specifically around mid-town. Apparently other cities have been pretty successful w/ such a system. The New York Times ran an article on this over the last month or so.
I wonder how long it will keep records? Or would such a system look for patterns of behavior, like circling in a neighborhood known for prostitution? Seems all fine and good to track a criminal on the run, but can guarantees be made against abuse?
Is there a FPS game out there that has done a good job of keeping hacks out? I play Return to Castle Wolfenstein, based on the Quake 3 engine, and there are plenty of players who can see through walls, and others that seem to have a supernatural sense of aim. I hear even the punkbuster enabled retail version has been hacked.
It seems trusted computing may help this some in the long term, but I find it hard to trust that I'm not wasting my time fighting an impossible opponent at times.