But AFAIK, I have yet to hear about any catastrophic failures of HP/Compaq laptop batteries. If the Dell/Apple/etc. battery fireworks were random catastrophic failures, I would expect failure count for each manufacturer to be roughly proportional to market share... but the failures are centered around specific product runs from select manufacturers, which hints at production problems in some specific Li-ion production runs at specific production plants in more-or-less specific calendar windows.
If each fire was an isolated and evenly distributed event, it would be written off as a product of aging. When catastrophic failures appear to affect specific products, there is reason to investigate whether or not there has been anomalous production defects or QA failure and issue a recall order to avoid major liability if more failures are to be expected.
The battery recalls may sound alarmist but these recalls most likely happened because detailed analysis/dissection of battery samples from the to-be-recalled battery runs revealed production anomalies or unexpected degradation that explain why these batteries failed and hint that many more batteries from these runs will fail catastrophically if given enough time. Companies generally do not order recalls unless they expect massive PR backlash and extremely expensive class-action suits if they do any less.
Who cares if the NSA has found a way to solve AES128 in a timely fashion? Many of the AES candidate algorithms can be extended to arbitrary key lengths and some 4096bits variants are currently out in the wild. Rijindael (today's AES) itself supports 128, 192 and 256bits as standard (required by NSA) and 160/224 as common extras. With more rounds and larger S-tables, larger keys could be supported. The main parameters that really limit key lengths of most AES candidates is throughput and implementation complexity: among other things, AES had to lend itself reasonably well to both software and hardware implementations.
I wish IEEE-1394 managed to gain stronger traction in the external storage area - since FW is dual-simplex, it behaves much better in cases where large bulk transfers are involved when compared to the half-duplex Hi-Speed USB2. I usually get ~25MB/s while moving data between two chained FW boxes but I rarely get more than 15MB/s while doing the same thing with USB boxes even if I mess around to put them on two separate root hubs... and the CPU load is usually higher with USB boxes too.
Now I am slowly switching from FW&USB to eSATA for external storage... I could use an external eSATA-to-eSATA port replicator with hot-swap.
My current ISP had a relatively cheap "unlimited" plan. Just like in this Comcast story, there were no official limits but some of my friends did receive notification letters with their bills telling them to cut back if they routinely exceeded this arbitrary monthly limit. At least two of my friends routinely do over 200GB/month. Now, the unspecified "unlimited" limit has become an official 100GB/month threshold where metered access kicks in and bills for extra bandwidth at over $5/GB. Current subscribers have 45 days from that notification/bill to jump off the ship if they are not happy.
Given that a ~$30k/month OC-48 can accommodate about 3000 250GB/month accounts (assuming the load is distributed evenly around the clock), the bandwidth cost itself would boil down to under $0.10/GB all-inclusive. However, the same OC-48 can only handle about 300 8Mbps accounts going full-tilt simultaneously so I am guessing the real issue is the increased peak bandwidth that needs to be supported to maintain a reasonable Worst-of-Worst QoS and will leave the ISPs with much more under-used OC-xxx capacity than they'd like.
People who want to have truly unlimited download should pay a premium equivalent to their share of an OC-48. 8Mbps = 1/300th of an OC-48's bandwidth so ~$100/month would be the premium that buys you the privilege of downloading up to about 2.6TB/month - the most that can be downloaded on a 8Mbps link over a 30 days period. This way, ISPs would have their base income and heavy-user-loaded OC-XXXes costs covered with extra capacity to cover traffic surges from regular accounts.
For less extreme users (like me), I would like to see customizable plans: 1- $10/month account maintenance fee, 2- select your speed (~$5/Mbps) 3- select either Unlimited ($share-of-OC-48) or your base GBs/month (price breaks could depend on selected speed) 4- pay extra for GB/months beyond prepaid (price breaks could vary depending on selected speed and base GB/month), GBs charges should be limited to no more than twice the OC-48 share equivalent. With those "rules", a true unlimited 8Mbps account would cost about $150/month and I would be able to get a more useful service (trading speed for more GBs/month) for the same $40/month.
I have played MP3s on my 486DX33... 11kHz, 8bits mono was the highest "quality" was ever able to play on it using a DOS-based player so I pretty much had to decode to WAV before playing to make them worth listening to.
Modern hardware can encode an entire album in less time than my 486 could play a single track... it is somewhat amazing that this old piece of junk (by yesterday's standards) served me about five years.
MP3 and MPEG video killed my 486, XviD and MPEG2 (and FFVII-PC) killed my P200MMX, SDRAM's extinction killed my 512MB P3, 1080p h264 and lack of PCIe are killing my P4... but I'll try to stretch it until Nehalem becomes a proven and affordable mainstream part - I can watch 1080p videos if I disable HT and overclock by about 10% to smooth out hiccups, good enough for me until then since I neither have nor plan to acquire much 1080p content (at least not the shiny disc form factor) any time soon.
BioShock's SecuROM server is down = you cannot re-install and use the copy you bought. Re-install Windows because of HD crash or OS corruption = your BioShock's SecuROM install count goes up and you eventually lose the ability to install. WGA servers are down = Vista downgrades to non-genuine mode should you be unlucky enough that it phoned home during an outage. The company goes out of business = you're fu**ed.
Fair compensation for work is... fair. But the restrictions they impose on legit customers and the risk of legit customers being hung out to dry should the company go bankrupt or experience technical difficulties is unacceptable.
I went to check about 30 minutes after your post and it is now up to a rather stupidly high $50k... but there are about a dozen other supposedly unlocked phones on eBay for ~$1k now.
IMO, today's larger caches and cheap RAM actually do make super-compact instructions less relevant. Also, if you consider all the tack-on prefixes, an hypothetical x86-like ISA based on 16bits base opcodes (instead of the current 8) could actually yield more compact code by eliminating the vast majority of prefix codes and work-arounds - at least while dealing with 64bits code. For example: on x86-64, a REX prefix is required to specify 64bits operands or to access extended general registers, which basically grows all 64bits r/r and extended register file (like "mov rax, r10") operations to at least 16bits.
Making instructions more uniform in both length and structure are keys to increasing instruction decoder throughput. x86's screwy ISA makes it difficult to improve, that's why current x86 CPUs can only decode 2-3 instructions per cycle (not to be confused with the 3-4 uOps/cycle dispatch to execution units) and we are unlikely to see them go much beyond that. (Wider dispatch units would still be useful in SMT CPUs - I remember reading that HT will be back in some flavors of either Nehalem or its 32nm follow-up... I hope they won't resurrect the Netburst replay engine this time around.)
The legacy stuff (modes and quirks) still costs transistors and these transistors translate into extra power and extra execution latency... but because the CPU cores (excluding L2 caches) are getting bigger and pipelines have stretched, the legacy overhead has become proportionally less significant.
Still, an hypothetical legacy-free x86-like CPU with a redesigned ISA to simplify instruction decoding could shed at least one pipeline stage along with ~10% of the core transistor count and ~10% of the power for a given performance level.
Since 70-80% of the transistors in today's upper mid-range CPUs go into L2 cache, this hypothetical complete core (including L2) would not be noticeably smaller but it would be slightly more efficient and could take performance/watt up a notch. Then again, minor performance/watt CPU gains (something like 10-15% in my hypothetical case) are quickly eclipsed by the ~40W idling power from HD2600 and GF8600 GPUs. (Yuck! And even entry-level GPUs from older generations are 20-30W idle.)
It would be nice to see a fresh ISA optimized for current and foreseeable desktop needs instead of sticking with stuff designed back when memory was over $10/KB with all sorts of extensions grafted on in variably elegant/seamless ways... but with Itanium headed Itanic (many predicted this based on the ISA's structure and production delays before the name was announced), Apple gone x86 and the few remaining independent ISA vendors slowly abandoning their low-volume in-house designs to jump on Opteron or Itanium, a desktop ISA refresh appears more unlikely than ever.
If we're lucky, mainstream desktop Linux will materialize and lure embedded ISAs into beefing up a bit and go after desktop market share.
Why not Windows 3.0/3.1 while we're at it? No IP stack out-of-the-box means he'd have to go back to dial-up BBSes at least long enough to download one. With neither USB nor large volume support, his download options (usable storage space) would also be severely restricted.
This is assuming Win3.x still runs on relatively recent hardware... but IIRC, the Win3.11 installer crashed on my P3 last time I tried it just for the heck of it.
The only BD advantages are 10GB of unused space, JVM and extra DRM... all of which add (mostly unnecessary/futile) costs/complexity in the player and media distribution chains.
Since both HD-DVD and Blueray streams have maximum bitrates of 18Mbps (nearly twice DVD's 1X spec), HD-DVD's 15GB is already (though barely) sufficient to store a 2h movie at the maximum allowed bitrate. From what I read though, it seems most HD movies (both HD-DVD and Blueray) are encoded at rates in the area of 5-6Mbps so there should be plenty of space left for extras even on HD-DVD - at current typical rates, HD-DVD would be good for 5-7 hours, plenty long enough for any of the LotR extended editions. I personally do not care which one wins as long as I can watch stuff in full HD without flipping discs half-way.
HD-DVD's 15GB capacity is sufficient for its primary purpose: cost-efficient HD movie distribution. Worst case, HD-DVD specs do allow for dual-layer discs should some titles (or disc writers) require extra space. For Joe Sixpack (at least those who do not have a PS3), the format war is likely to remain irrelevant until stand-alone players drop below $200. After this point, things could snowball towards HD-DVD - HD-DVD will almost certainly get there first, possibly this year.
I have six computers in my room alone... so I could print at least a dozen coupons without deleting any cookies and whatnot.
On top of that, I have at least two browsers installed on each PC so this number can double to two dozens.
Some are also multi-boot with two or three OSes, that's half a dozen extras.
So, by simply going around all my current PCs and all my OS installs, I could have at least 30 coupons without any work-arounds. Trying to use cookies to limit the number of units I can print coupons for is pretty daft. They should have simply limited the deal to two units per address so it wouldn't matter if the guy printed 1 or 1000 coupons.
Metal fatigue does not really apply to traces within ICs since they are not subjected to any significant bending, stretching or thermal cycling. With the IHS and soft floor-planning to improve timings and distribute hot-spots across the die for more even heating/cooling, thermally-induced mechanical stresses within the chip are further reduced.
The worst factor is electromigration but it can be kept in check by keeping voltages and temperatures down. I first started reading of overclockers concerned about it around the Coppermine days but AFAIK, it seems Prescott and even Conroe have fared fairly well so far with ~10% over-volting and ~30% overclocking so electromigration has yet to become a major issue down to at least 90nm or even 65nm for the common overclocking targets.
For the time being, cooling upgrades can still save the day/CPU. As usual, the practice is not recommended for production systems - with the possible exception of pre-production stress-testing.
Setting speed limits based on safety alone is a tricky business. In my area, there are many highway stretches with 70km/h speed limits (instead of the usual 100) where almost everyone does 100-120km/h during summer but 50-60km/h in typical winter weather or even down to 30km/h in heavy rain/snow... the safe/typical speeds vary a lot between seasons, weather and road conditions.
Since safe speeds depend on so many factors, many areas (particularly those with wildly variable conditions) could benefit significantly from having dynamic limits. But for now, most limits are set to what is generally considered safe given the particular mix of regularly occurring or expected environmental degradations in the region, usually weighted towards the worst cases.
The other hazard from over-speeding is that it may distract or confuse other drivers - I am personally far more worried about induced indecision than speed... and somewhat ironically, the most confusing situations I have been through involved under-speeding vehicle(s).
When lawyers pick up cases with no-win-no-pay clauses, they usually also aim for a larger share of any settlement or other gains when they win. It is a double/triple/etc.-or-nothing proposition lawyers usually offer for cases they have high confidence in - a lawyer may have a normal rate of $150/h but in no-win-no-pay, they will usually want to collect over $500/h up to the sum of any gained amounts.
It may look like charity to some people but it is simply lawyers gambling to maximize their gains while also making their services available to less fortunate people.
My mom went through something similar too with the contractor that re-did her basement after a flooding but butchered the decontamination.
Since it was a private company at the time (the guy did incorporate his company between that job and the court proceedings) and the guy had no seizable property to his name (all under his wife's), the judge gave my mom the paperwork to request bailiffs at her leisure. Since he wouldn't give my mom her money by claiming he did not have it, my mom decided to pay $2000 to have bailiffs lock up all the guy's credit and banking accounts right away. The guy noticed his accounts were frozen the next day and his wife wrote my mom a ~$60k check right away. (And yes, it cleared.)
I wonder how long the RIAA would defer payment of their fine if the woman in this case did get bailiffs to suspend RIAA banking and credit accounts until payment is delivered... that would certainly be funny - imagine how the RIAA's lawyers would react to their bouncing paychecks!
I used to encode my CDs at 128kbps until I listened to an encoded track with lots of hats. At 128kbps, it sounded absolutely horrible. 160kbps was a substantial improvement but still clearly inferior to the CD version - the hits still sounded indistinct. At 192kbps, the toughest hat passages were finally clear enough to be tolerable so I decided this was a good enough balance between sound quality and size for me and my 1GB player. At 224kbps, the differences start getting quite subtle so I never bothered with bitrates beyond 256kbps - at this point, you start to need a quiet room and good equipment to notice improvements. Since I listen to music mostly during commute or while working on computers, silence is scarce so I favor packing more music over extreme bitrates.
For casual listening, MP3/192kbps is good enough for be... but if I had the luxury of a quiet room and decent AV setup, I probably would upgrade my standard to 256kbps to iron out the vast majority of 192kbps' wrinkles I can still distinguish.
If you want something with DVI-in that could potentially be used to re-encode digital output, you can look at Xilinx's VIODC add-on module for the ML40x development platforms... http://www.xilinx.com/xlnx/xebiz/designResources/i p_product_details.jsp?key=HW-V4SX35-VIDEO-SK-US - it supports nearly all common video sources (other than HDMI) but only up to 720p. Actually, the link is for the ML402+VIODC bundle since the stand-alone VIODC card is (conveniently?) down. Of course, you need to supply your own application-specific firmware.
Ripping from HDMI/DVI/etc. is a little dumb: it needs insane amounts of storage, insane amounts of processing power for (near-)real-time re-encoding to make said storage requirements manageable, the re-encoded content can only be as good as the initial decoding and the re-encoding(s) will add its/their own lot of extra noise/blurriness/artifacts/etc.
Fossil fuels are the 'fruit' of millions of years worth of energy absorbed by plants and animals and this energy source is near its end since we are using it a million times faster than it can renew itself - hundreds of millions of years worth of it has been used in less than 200 years of modern industry... claiming that oil has a net positive energy balance is very short-term thinking... even more so when that particular energy source has less than a century left to it.
Ethanol is the new craze but producing it requires both surface area and a significant energy investment... most of the energy comes from the sun but the machinery that runs the plant also needs power to regulate production parameters and process the algae (a fairly new approach expected to have better yields) until it becomes usable fuel. The problems with ethanol is that it cannot sustain a worldwide switch (I guesstimate we currently need about one square mile per 60k people... so that's ~150k square miles for a full-scale ethanol switch) and on the pollution side, ethanol requires perfectly tuned engines to burn cleanly or it can quickly become worse than biodiesel in the nitrous monoxide department.
The problem with batteries is that they are no match for the range and power currently taken for granted with the combination of internal combustion engines and liquid fuels. While hydrogen does kind of suck in terms of energy/gallon, it looks much better in terms of energy/weight compared to batteries and gasoline.
Another "new" energy source I have issues with is the "renewableness" of geothermal energy - surely there is a limit to how much energy can be extracted from the planet's core before bad/unexpected things start to happen. If we started poaching geothermal energy like we poached fossil fuels, I wonder what sort of surprises we are going to run into... I (almost) can't wait for the first GW-class geothermal power plants to find out.
Rotational inertia will increased if you add more rotating material (drive shaft) in the drive train if all other parts are the same. The extra U-joints add extra failure/maintenance points and since extra parts add mass, you also end up with extra linear inertia as well. The motor-wheel assembly can be more reliable since its only moving parts are bearings. The general concept of motor-wheel has been around for over ten years, I am a little surprised that things have remained this quiet so far.
Since I remember reading about a motor-wheel patent eight or so years ago, we'll probably hear about the real new developments in about ten more years - after those late-'90s patents have expired.
I have been expecting moments like this ever since the DRM fever started spreading... unfortunately, this is too mild to be the big *IT* that I have been waiting/wishing for. What I want to see is for a highly popular media outfit with centralized DRM management mess up the service for all their customers due to some random server hiccup (like somehow losing the root keys for the DRM scheme) and cause instant worldwide outrage.
Seeing how Apple and others have started backpedaling on DRM though, I am starting to doubt we'll see sufficiently high-profile DRM failures to trigger that ideal "DRM-awareness" event.
Given the many ways DRM can turn around and bite both customers (unusable content) and companies (class-action lawsuits when files become mysteriously unplayable) in the ass, the only good DRM is NO DRM.
As the GP said, the applet window sets itself on-top and keeps resetting the focus to itself so even task-manager is difficult to use since it keeps losing focus. For people who set their task bar to auto-hide (like me) and do not run their browser on a secondary display, this could be really annoying. I guess an advertiser could very well upgrade the script to enumerate displays and open instances of the thing for each display device.
Since lithium battery chargers already have to closely and accurately monitor battery voltage, current and temperature to correctly set the charge current to reduce the risk of catastrophic failures, simply regulating the input current to have zero net battery current requires no extra hardware, no extra space, no extra power and nearly no extra design costs - the only extra design costs lie in the charge controller's firmware and related testing. As far as battery life is concerned, zero current is just as good as a disconnect since the tiny ripple voltage the battery might see from the charge controller will displace an insignificant number of ions.
With some luck, most of the battery lifespan/safety issues will be resolved when engineered material (like the titanium "nanoSafe") will start replacing graphite in NiMH and Li-ion batteries. These materials will allow 100X greater sustained charge/discharge currents along with 10k+ deep-cycle endurance. On top of that, these remain stable beyond 200C - unlike current Li-ion that ignite somewhere in the 120-160C range. That technology will be a perfect fit for electric-drive cars.
Slashdot Mirror
Yes, high-capacity batteries can fail.
But AFAIK, I have yet to hear about any catastrophic failures of HP/Compaq laptop batteries. If the Dell/Apple/etc. battery fireworks were random catastrophic failures, I would expect failure count for each manufacturer to be roughly proportional to market share... but the failures are centered around specific product runs from select manufacturers, which hints at production problems in some specific Li-ion production runs at specific production plants in more-or-less specific calendar windows.
If each fire was an isolated and evenly distributed event, it would be written off as a product of aging. When catastrophic failures appear to affect specific products, there is reason to investigate whether or not there has been anomalous production defects or QA failure and issue a recall order to avoid major liability if more failures are to be expected.
The battery recalls may sound alarmist but these recalls most likely happened because detailed analysis/dissection of battery samples from the to-be-recalled battery runs revealed production anomalies or unexpected degradation that explain why these batteries failed and hint that many more batteries from these runs will fail catastrophically if given enough time. Companies generally do not order recalls unless they expect massive PR backlash and extremely expensive class-action suits if they do any less.
Who cares if the NSA has found a way to solve AES128 in a timely fashion? Many of the AES candidate algorithms can be extended to arbitrary key lengths and some 4096bits variants are currently out in the wild. Rijindael (today's AES) itself supports 128, 192 and 256bits as standard (required by NSA) and 160/224 as common extras. With more rounds and larger S-tables, larger keys could be supported. The main parameters that really limit key lengths of most AES candidates is throughput and implementation complexity: among other things, AES had to lend itself reasonably well to both software and hardware implementations.
I wish IEEE-1394 managed to gain stronger traction in the external storage area - since FW is dual-simplex, it behaves much better in cases where large bulk transfers are involved when compared to the half-duplex Hi-Speed USB2. I usually get ~25MB/s while moving data between two chained FW boxes but I rarely get more than 15MB/s while doing the same thing with USB boxes even if I mess around to put them on two separate root hubs... and the CPU load is usually higher with USB boxes too.
Now I am slowly switching from FW&USB to eSATA for external storage... I could use an external eSATA-to-eSATA port replicator with hot-swap.
My current ISP had a relatively cheap "unlimited" plan. Just like in this Comcast story, there were no official limits but some of my friends did receive notification letters with their bills telling them to cut back if they routinely exceeded this arbitrary monthly limit. At least two of my friends routinely do over 200GB/month. Now, the unspecified "unlimited" limit has become an official 100GB/month threshold where metered access kicks in and bills for extra bandwidth at over $5/GB. Current subscribers have 45 days from that notification/bill to jump off the ship if they are not happy.
Given that a ~$30k/month OC-48 can accommodate about 3000 250GB/month accounts (assuming the load is distributed evenly around the clock), the bandwidth cost itself would boil down to under $0.10/GB all-inclusive. However, the same OC-48 can only handle about 300 8Mbps accounts going full-tilt simultaneously so I am guessing the real issue is the increased peak bandwidth that needs to be supported to maintain a reasonable Worst-of-Worst QoS and will leave the ISPs with much more under-used OC-xxx capacity than they'd like.
People who want to have truly unlimited download should pay a premium equivalent to their share of an OC-48. 8Mbps = 1/300th of an OC-48's bandwidth so ~$100/month would be the premium that buys you the privilege of downloading up to about 2.6TB/month - the most that can be downloaded on a 8Mbps link over a 30 days period. This way, ISPs would have their base income and heavy-user-loaded OC-XXXes costs covered with extra capacity to cover traffic surges from regular accounts.
For less extreme users (like me), I would like to see customizable plans: 1- $10/month account maintenance fee, 2- select your speed (~$5/Mbps) 3- select either Unlimited ($share-of-OC-48) or your base GBs/month (price breaks could depend on selected speed) 4- pay extra for GB/months beyond prepaid (price breaks could vary depending on selected speed and base GB/month), GBs charges should be limited to no more than twice the OC-48 share equivalent. With those "rules", a true unlimited 8Mbps account would cost about $150/month and I would be able to get a more useful service (trading speed for more GBs/month) for the same $40/month.
I have played MP3s on my 486DX33... 11kHz, 8bits mono was the highest "quality" was ever able to play on it using a DOS-based player so I pretty much had to decode to WAV before playing to make them worth listening to.
Modern hardware can encode an entire album in less time than my 486 could play a single track... it is somewhat amazing that this old piece of junk (by yesterday's standards) served me about five years.
MP3 and MPEG video killed my 486, XviD and MPEG2 (and FFVII-PC) killed my P200MMX, SDRAM's extinction killed my 512MB P3, 1080p h264 and lack of PCIe are killing my P4... but I'll try to stretch it until Nehalem becomes a proven and affordable mainstream part - I can watch 1080p videos if I disable HT and overclock by about 10% to smooth out hiccups, good enough for me until then since I neither have nor plan to acquire much 1080p content (at least not the shiny disc form factor) any time soon.
BioShock's SecuROM server is down = you cannot re-install and use the copy you bought.
Re-install Windows because of HD crash or OS corruption = your BioShock's SecuROM install count goes up and you eventually lose the ability to install.
WGA servers are down = Vista downgrades to non-genuine mode should you be unlucky enough that it phoned home during an outage.
The company goes out of business = you're fu**ed.
Fair compensation for work is... fair. But the restrictions they impose on legit customers and the risk of legit customers being hung out to dry should the company go bankrupt or experience technical difficulties is unacceptable.
Also keep in mind that the EU PS3 comes without the PS1/PS2 emulation assist hardware found in NA and Japanese models.
It is kind of ironic/funny/moronic that the EU version is so much more expensive even though it costs at least $30 less to produce.
Well, the auction has reached $100k now... I wonder if any of those bids are real.
My bet is that Apple/AT&T will get eBay to take it down next Monday.
I went to check about 30 minutes after your post and it is now up to a rather stupidly high $50k... but there are about a dozen other supposedly unlocked phones on eBay for ~$1k now.
IMO, today's larger caches and cheap RAM actually do make super-compact instructions less relevant. Also, if you consider all the tack-on prefixes, an hypothetical x86-like ISA based on 16bits base opcodes (instead of the current 8) could actually yield more compact code by eliminating the vast majority of prefix codes and work-arounds - at least while dealing with 64bits code. For example: on x86-64, a REX prefix is required to specify 64bits operands or to access extended general registers, which basically grows all 64bits r/r and extended register file (like "mov rax, r10") operations to at least 16bits.
Making instructions more uniform in both length and structure are keys to increasing instruction decoder throughput. x86's screwy ISA makes it difficult to improve, that's why current x86 CPUs can only decode 2-3 instructions per cycle (not to be confused with the 3-4 uOps/cycle dispatch to execution units) and we are unlikely to see them go much beyond that. (Wider dispatch units would still be useful in SMT CPUs - I remember reading that HT will be back in some flavors of either Nehalem or its 32nm follow-up... I hope they won't resurrect the Netburst replay engine this time around.)
The legacy stuff (modes and quirks) still costs transistors and these transistors translate into extra power and extra execution latency... but because the CPU cores (excluding L2 caches) are getting bigger and pipelines have stretched, the legacy overhead has become proportionally less significant.
Still, an hypothetical legacy-free x86-like CPU with a redesigned ISA to simplify instruction decoding could shed at least one pipeline stage along with ~10% of the core transistor count and ~10% of the power for a given performance level.
Since 70-80% of the transistors in today's upper mid-range CPUs go into L2 cache, this hypothetical complete core (including L2) would not be noticeably smaller but it would be slightly more efficient and could take performance/watt up a notch. Then again, minor performance/watt CPU gains (something like 10-15% in my hypothetical case) are quickly eclipsed by the ~40W idling power from HD2600 and GF8600 GPUs. (Yuck! And even entry-level GPUs from older generations are 20-30W idle.)
It would be nice to see a fresh ISA optimized for current and foreseeable desktop needs instead of sticking with stuff designed back when memory was over $10/KB with all sorts of extensions grafted on in variably elegant/seamless ways... but with Itanium headed Itanic (many predicted this based on the ISA's structure and production delays before the name was announced), Apple gone x86 and the few remaining independent ISA vendors slowly abandoning their low-volume in-house designs to jump on Opteron or Itanium, a desktop ISA refresh appears more unlikely than ever.
If we're lucky, mainstream desktop Linux will materialize and lure embedded ISAs into beefing up a bit and go after desktop market share.
Why not Windows 3.0/3.1 while we're at it? No IP stack out-of-the-box means he'd have to go back to dial-up BBSes at least long enough to download one. With neither USB nor large volume support, his download options (usable storage space) would also be severely restricted.
This is assuming Win3.x still runs on relatively recent hardware... but IIRC, the Win3.11 installer crashed on my P3 last time I tried it just for the heck of it.
The only BD advantages are 10GB of unused space, JVM and extra DRM... all of which add (mostly unnecessary/futile) costs/complexity in the player and media distribution chains.
Since both HD-DVD and Blueray streams have maximum bitrates of 18Mbps (nearly twice DVD's 1X spec), HD-DVD's 15GB is already (though barely) sufficient to store a 2h movie at the maximum allowed bitrate. From what I read though, it seems most HD movies (both HD-DVD and Blueray) are encoded at rates in the area of 5-6Mbps so there should be plenty of space left for extras even on HD-DVD - at current typical rates, HD-DVD would be good for 5-7 hours, plenty long enough for any of the LotR extended editions. I personally do not care which one wins as long as I can watch stuff in full HD without flipping discs half-way.
HD-DVD's 15GB capacity is sufficient for its primary purpose: cost-efficient HD movie distribution. Worst case, HD-DVD specs do allow for dual-layer discs should some titles (or disc writers) require extra space. For Joe Sixpack (at least those who do not have a PS3), the format war is likely to remain irrelevant until stand-alone players drop below $200. After this point, things could snowball towards HD-DVD - HD-DVD will almost certainly get there first, possibly this year.
I have six computers in my room alone... so I could print at least a dozen coupons without deleting any cookies and whatnot.
On top of that, I have at least two browsers installed on each PC so this number can double to two dozens.
Some are also multi-boot with two or three OSes, that's half a dozen extras.
So, by simply going around all my current PCs and all my OS installs, I could have at least 30 coupons without any work-arounds. Trying to use cookies to limit the number of units I can print coupons for is pretty daft. They should have simply limited the deal to two units per address so it wouldn't matter if the guy printed 1 or 1000 coupons.
Metal fatigue does not really apply to traces within ICs since they are not subjected to any significant bending, stretching or thermal cycling. With the IHS and soft floor-planning to improve timings and distribute hot-spots across the die for more even heating/cooling, thermally-induced mechanical stresses within the chip are further reduced.
The worst factor is electromigration but it can be kept in check by keeping voltages and temperatures down. I first started reading of overclockers concerned about it around the Coppermine days but AFAIK, it seems Prescott and even Conroe have fared fairly well so far with ~10% over-volting and ~30% overclocking so electromigration has yet to become a major issue down to at least 90nm or even 65nm for the common overclocking targets.
For the time being, cooling upgrades can still save the day/CPU. As usual, the practice is not recommended for production systems - with the possible exception of pre-production stress-testing.
Setting speed limits based on safety alone is a tricky business. In my area, there are many highway stretches with 70km/h speed limits (instead of the usual 100) where almost everyone does 100-120km/h during summer but 50-60km/h in typical winter weather or even down to 30km/h in heavy rain/snow... the safe/typical speeds vary a lot between seasons, weather and road conditions.
Since safe speeds depend on so many factors, many areas (particularly those with wildly variable conditions) could benefit significantly from having dynamic limits. But for now, most limits are set to what is generally considered safe given the particular mix of regularly occurring or expected environmental degradations in the region, usually weighted towards the worst cases.
The other hazard from over-speeding is that it may distract or confuse other drivers - I am personally far more worried about induced indecision than speed... and somewhat ironically, the most confusing situations I have been through involved under-speeding vehicle(s).
When lawyers pick up cases with no-win-no-pay clauses, they usually also aim for a larger share of any settlement or other gains when they win. It is a double/triple/etc.-or-nothing proposition lawyers usually offer for cases they have high confidence in - a lawyer may have a normal rate of $150/h but in no-win-no-pay, they will usually want to collect over $500/h up to the sum of any gained amounts.
It may look like charity to some people but it is simply lawyers gambling to maximize their gains while also making their services available to less fortunate people.
My mom went through something similar too with the contractor that re-did her basement after a flooding but butchered the decontamination.
Since it was a private company at the time (the guy did incorporate his company between that job and the court proceedings) and the guy had no seizable property to his name (all under his wife's), the judge gave my mom the paperwork to request bailiffs at her leisure. Since he wouldn't give my mom her money by claiming he did not have it, my mom decided to pay $2000 to have bailiffs lock up all the guy's credit and banking accounts right away. The guy noticed his accounts were frozen the next day and his wife wrote my mom a ~$60k check right away. (And yes, it cleared.)
I wonder how long the RIAA would defer payment of their fine if the woman in this case did get bailiffs to suspend RIAA banking and credit accounts until payment is delivered... that would certainly be funny - imagine how the RIAA's lawyers would react to their bouncing paychecks!
I used to encode my CDs at 128kbps until I listened to an encoded track with lots of hats. At 128kbps, it sounded absolutely horrible. 160kbps was a substantial improvement but still clearly inferior to the CD version - the hits still sounded indistinct. At 192kbps, the toughest hat passages were finally clear enough to be tolerable so I decided this was a good enough balance between sound quality and size for me and my 1GB player. At 224kbps, the differences start getting quite subtle so I never bothered with bitrates beyond 256kbps - at this point, you start to need a quiet room and good equipment to notice improvements. Since I listen to music mostly during commute or while working on computers, silence is scarce so I favor packing more music over extreme bitrates.
For casual listening, MP3/192kbps is good enough for be... but if I had the luxury of a quiet room and decent AV setup, I probably would upgrade my standard to 256kbps to iron out the vast majority of 192kbps' wrinkles I can still distinguish.
If you want something with DVI-in that could potentially be used to re-encode digital output, you can look at Xilinx's VIODC add-on module for the ML40x development platforms... http://www.xilinx.com/xlnx/xebiz/designResources/i p_product_details.jsp?key=HW-V4SX35-VIDEO-SK-US - it supports nearly all common video sources (other than HDMI) but only up to 720p. Actually, the link is for the ML402+VIODC bundle since the stand-alone VIODC card is (conveniently?) down. Of course, you need to supply your own application-specific firmware.
Ripping from HDMI/DVI/etc. is a little dumb: it needs insane amounts of storage, insane amounts of processing power for (near-)real-time re-encoding to make said storage requirements manageable, the re-encoded content can only be as good as the initial decoding and the re-encoding(s) will add its/their own lot of extra noise/blurriness/artifacts/etc.
Fossil fuels are the 'fruit' of millions of years worth of energy absorbed by plants and animals and this energy source is near its end since we are using it a million times faster than it can renew itself - hundreds of millions of years worth of it has been used in less than 200 years of modern industry... claiming that oil has a net positive energy balance is very short-term thinking... even more so when that particular energy source has less than a century left to it.
Ethanol is the new craze but producing it requires both surface area and a significant energy investment... most of the energy comes from the sun but the machinery that runs the plant also needs power to regulate production parameters and process the algae (a fairly new approach expected to have better yields) until it becomes usable fuel. The problems with ethanol is that it cannot sustain a worldwide switch (I guesstimate we currently need about one square mile per 60k people... so that's ~150k square miles for a full-scale ethanol switch) and on the pollution side, ethanol requires perfectly tuned engines to burn cleanly or it can quickly become worse than biodiesel in the nitrous monoxide department.
The problem with batteries is that they are no match for the range and power currently taken for granted with the combination of internal combustion engines and liquid fuels. While hydrogen does kind of suck in terms of energy/gallon, it looks much better in terms of energy/weight compared to batteries and gasoline.
Another "new" energy source I have issues with is the "renewableness" of geothermal energy - surely there is a limit to how much energy can be extracted from the planet's core before bad/unexpected things start to happen. If we started poaching geothermal energy like we poached fossil fuels, I wonder what sort of surprises we are going to run into... I (almost) can't wait for the first GW-class geothermal power plants to find out.
Rotational inertia will increased if you add more rotating material (drive shaft) in the drive train if all other parts are the same. The extra U-joints add extra failure/maintenance points and since extra parts add mass, you also end up with extra linear inertia as well. The motor-wheel assembly can be more reliable since its only moving parts are bearings. The general concept of motor-wheel has been around for over ten years, I am a little surprised that things have remained this quiet so far.
Since I remember reading about a motor-wheel patent eight or so years ago, we'll probably hear about the real new developments in about ten more years - after those late-'90s patents have expired.
I have been expecting moments like this ever since the DRM fever started spreading... unfortunately, this is too mild to be the big *IT* that I have been waiting/wishing for. What I want to see is for a highly popular media outfit with centralized DRM management mess up the service for all their customers due to some random server hiccup (like somehow losing the root keys for the DRM scheme) and cause instant worldwide outrage.
Seeing how Apple and others have started backpedaling on DRM though, I am starting to doubt we'll see sufficiently high-profile DRM failures to trigger that ideal "DRM-awareness" event.
Given the many ways DRM can turn around and bite both customers (unusable content) and companies (class-action lawsuits when files become mysteriously unplayable) in the ass, the only good DRM is NO DRM.
I tried it, Alt-F4 did not work.
As the GP said, the applet window sets itself on-top and keeps resetting the focus to itself so even task-manager is difficult to use since it keeps losing focus. For people who set their task bar to auto-hide (like me) and do not run their browser on a secondary display, this could be really annoying. I guess an advertiser could very well upgrade the script to enumerate displays and open instances of the thing for each display device.
Aerogel super/ultra-capacitors are expensive.
Since lithium battery chargers already have to closely and accurately monitor battery voltage, current and temperature to correctly set the charge current to reduce the risk of catastrophic failures, simply regulating the input current to have zero net battery current requires no extra hardware, no extra space, no extra power and nearly no extra design costs - the only extra design costs lie in the charge controller's firmware and related testing. As far as battery life is concerned, zero current is just as good as a disconnect since the tiny ripple voltage the battery might see from the charge controller will displace an insignificant number of ions.
With some luck, most of the battery lifespan/safety issues will be resolved when engineered material (like the titanium "nanoSafe") will start replacing graphite in NiMH and Li-ion batteries. These materials will allow 100X greater sustained charge/discharge currents along with 10k+ deep-cycle endurance. On top of that, these remain stable beyond 200C - unlike current Li-ion that ignite somewhere in the 120-160C range. That technology will be a perfect fit for electric-drive cars.