Dude... The original poster is deluded. 10 years ago dictation sucked. 5 years ago dictation sucked. (I've got boxes of the stuff that were purchased and stunk so bad no one used them.) I'm talking about dictation that actually works well enough to use.
Modern PDAs do not have the equivalent of a PPro 200. StrongARM 200s are about the same as a Pentium 90 for general purpose work, and right on useless for floating point work.
According to my compilation benchmarks (integer and data pushing, no floating point), a PPro 200 is right about 4 times faster than a strongarm 200. On the other hand the SA200s are little more than twice as fast as a 486-66dx2.:-)
The very latest xscale pdas are proably about twice the speed of the SA-110 200MHz ones, but I don't have any similar hardware to benchmark.
I just unlinked my benchmark page form the web because I hadn't updated it in years, but here is a link. Fun to reminise about all those machines I thought were so fast at the time... metastones
Tiny devices like cell phones and PDAs don't have the CPU power for sexy, high quality voice recognition. They do however have wireless connectivity. So, solve the problem this way...
Install voice recognition servers, network connected boxes with powerful CPUs and the best voice recognition software you can get your hands on. A voice recognition client then just needs to send the voice data up to the server and get the translation back, say 100kbps up and some tiny amount back.
The payback comes because most devices will only use voice recognition for brief periods, so will present a negligible load on the servers. The dictation users will place a higher load on the servers, but even there, I'm guessing there is a lot of pausing involved. I'm also going to guess that some lag is acceptable for dictation. Presumably the person is thinking about what they are saying and proof reading later. This load can be prioritized lower to allow better immediate response for people issuing voice commands on their mobile devices.
Power consumption on the portable device will probably improve. They will have to operate their transmitter (think "talk time" vs. "on time"), but they won't need 5 watts of CPU doing recognition. (Guessing from a mobile G3 PPC, further validated, considering that the CPU spot of my iBook gets far hotter under solid use than a cellphone.)
So, just to pick numbers out of the air, a dual processor, high end commodity hardware voice server might serve 500 pda users giving intermittent commands and 6 simultaneous dictation users.
A company or school could easily justify the hardware cost of this service.
I just followed google news to a story at TIME. Their article surrounded by a kalidescope of ads took more than 10 seconds to load over my DSL line. The article was presented in an uncomfortably skinny space left after the mass of garish ads were plastered around it.
I can't tell you what any of the TIME ads were for. The only impression I carry away is that the article didn't have as much meat as I expected than the web site is annoying.
I'll take a tasteful ultramercial once a day instead of a delay to load a mass of ads on every page.
160 responses on slashdot and virtually none that actually talk about the ultracommercial concept.
I just went to salon and read a premium article. Here is my synopsis...
ultracommercial has a problem on their systems, I got pages of MySQL errors the first time I tried it. Oops.
The second time I tried I got to look at four spiffy pictures of a car with little click spots to get more info.
After the forth picture I was sent to the article I had been reading with a complete version instead of just the front quarter.
All in all, the ad took me less time than it takes me to walk outside and pick up my newspaper, plus my feet didn't get cold.
If a 10 second ad can keep salon and their reporters working I'm all for it. The US needs independent journalists. (Even if they sometimes say things you'd rather not hear. Personally I'm offended by something in Salon every single day. If I wasn't, I wouldn't bother to read it.)
This is exactly what I have. I have a Sony built DirecTV reciever with Tivo built in.
The challenge for Tivo and Replay is to ensure that the TV and set top builders have to license their technology. It can't cost more than $1million to develop the code for a DVR. (Well it could, but it shouldn't.) If Tivo wants more than $1M/year return on their investment they have to legally prevent the TV builders from making and selling their own DVR software.
If Tivo has well written patents they can use litigation, if they have shaky patents they can use the treat of litigation. If they have worthless patents they are out of luck. Maybe the reason to cancel the lawsuit is to avoid a legal precedent on which patents are valid? Maybe each company has amassed a body of prior art on the other's patents? Just guessing.
Good god...how many of these things could be lurking about in automotive airbags...
Well I would hope none. The only thing in your air bag should be the sodium azide and an igniter. The last thing you need in an accident is a bunch of loose capacitors and crap being blown into your face.
I just checked their web site. Looks like their only original series left will be Crossing Over. Sort of a misplaced show. High marks for fiction but no a zero for science.
I should have left the Nagasaki bomb out of the post, but cyclotrons were used for the uranium processing on the Hiroshima bomb. Gaseous diffusion was also used and soon became the preferred method but if you read up on "Calutron"s (modified cyclotrons) you will find their role in the early uranium processing.
this article has a pretty good description, including reasons why the Calutrons feel from favor.
Cyclotrons can be used for uranium enrichment. Most of the uranium used in the Hiroshima (40*WTC911) and Nagaski (20*WTC911) bombs was purified in cyclotrons.
It takes a lot of energy, so you might want to have vast oil reserves that you aren't allowed to export in order to power the cylcotrons.
note: The Nagasaki bomb was not that much smaller, the weather was bad and we missed the population center and hit an industrial area instead.
What you need are dust corners. These are little brass trianglish affairs to go in square corners and turn them into radiused corners. They made sweeping and scrubbing easier back when it was all hand powered.
I am always surprised at the distance distortion that occurs in areas of high population density. I live in the middle of the United States (< 60 miles from the population centroid) and don't think twice about driving 500 miles to my vacation cabin. (Too d*mn hot where I live in the summer. Got to go north.)
I have had friends who live in Manhattan that consider 15 miles to be "far" and 200 miles to be an extraordinary distance to travel.
I wonder if people's definition of far is better correlated to "number of people passed" than "distance"?
Any slashdot readers from the Australian outback want to tell us what they think "far" is?
I haven't used CodeWarrior for years (got a closet full of shirts though), but I believe you are looking at a compiler difference, not the IDE difference.
gcc is not a fast C compiler. It is a portable C compiler and it makes pretty good code, but it is not fast. g++ is a slow C++ compiler.
The Codewarrior compilers were fast compilers that made pretty good code.
Its all in the tradeoffs.
Engineering time
target portability
source language selection
code quality
compilation speed
You can't have it all. Apple has been adding engineering hours to improve speed with gcc, but gcc will always value source language selection and target portability above all else.
Ha! I am immune to the doubling of taxes. ~50% of my income already goes to taxes in one form or another.
The median US family in the US pays about 40% of their income to taxes. nifty graphs. (note: I didn't check out the credibility of the people hosting this report, but it matches what I have read in known reputable sources, so I will link it)
Maybe "taxes would have to be doubled refers to the social security taxes? That would be roughly a 33% increase in total taxes taking the median tax burden to something just over 50%.
Apple aimed wrong with the Newton. It was a good guess and a nice product, but missed the mass market. It turned out that most people were much happier with half the capability, half the size, and half the price. Palm came later, learned from Newton, aimed better, and ate the market.
Tivo on the other hand is a direct hit on what the mass market wants. The late comers to Tivo-ing can aim no better.
So your playing B&W, you are half way through a gesture to create water on your poor people and the game decides to autosave, freezing input in the middle of the gesture. So you try to recover and complete the gesture or at least make it do something sane, but no. You get a fireball or something and incinerate your people. Bad god. Maybe the PC version of BW is better, but the Mac version could inspire one to injure the programmers.
Galeon on the other hand has nice gesture support.
Re:Not now, guys!? Please consider NOT switching.
on
Flirting With Mac OS X
·
· Score: 2
The quicktime ogg library is not yet correct. Apple would never release that.
There seems to be some issue about returning a variable amount of data with each call, but given that variable bit rate mp3s work it seems to me that with an appropriate buffer for reblocking output data oggs should work as well. They may have gotten beyond that, since I last looked closely, but I see on their bug tracker they still have issues.
Still, if the API works out it would make sense for them to use quicktime and save some duplication.
Sorry if this is redundant, but I couldn't find the thread where the rest of us register to get cheap TiBooks from Apple.
Maybe Apple could have a special refurb sale for switching slashdotters? (Don't get your hopes up, refurbed TiBooks go for about $200 off retail. Damn those computers that keep their value.)
So sign up in a reply to this so Apple knows how many machines to set aside for the Switch-a-Nerd program.
Re:Not now, guys!? Please consider NOT switching.
on
Flirting With Mac OS X
·
· Score: 5, Insightful
And from the other viewport...
You are locked into the hardware. - whatever I buy for hardware I will be locked in to for 3 years. (My buying cycle time.) I don't need anything other than what I have now to be happy for three years.
Microsoft controls Mac adoption. - microsoft controls my ability to pay my american express bill online. Thats it. AppleWorks does everything I need in an office suite. Freehand takes care of my compositing. I will kiss whoever integrates top notch vector layers with gimp.
Open Office.org's health is good for everyone. Sounds good to me, but switching to the mac doesn't mean you shell out $300 for Office. That's just silly.
The IPod doesn't support Ogg True, nor does any other mp3 player. Oops, that was a silly contradiction, you know what I meant. The rumor I hear and believe is that iTunes 4 goes to a plugin architecture for sound formats and ogg support is in. I hopefully assume that means ogg in the ipod at the same time. Sometime next year. I can wait to re-rip my aluminized polycarbonate music ownership tokens until then.
You waited for: good, free GUI desktops... And I'm still waiting. Its getting better, but I no longer have the time to work on making the tools, I need something that works. (My linux machine that runs the 8ball robot ate 10 hours of my time last night and still isn't right. USB driver problems. Using the USB camera because kernel updates broke the bttv card support. Maybe I'll go back to the bttv and see how that goes. It doesn't take but a couple of those to pay the software bill to get a machine that just works.)
Believe it or not, Slashdot and linux are wedded communities change and evolve. People's interests change.
I also don't work for Sun, I don't work for RedHat either, but do work for Debian. My choice of helping out with linux works for everybody though.
I wonder how you meausured CPU use. top is tempting, but should never be used for that. It is far too easily skewed by processes that respond to timed interrupts or that yield the processor voluntarily. You should make a benchmark mode where your program does not do the updates for some period and compare its progress to the version with updates.
Wait a ding dong minute! You said terminal updates and simple C++ program. You are printing on standard output to a terminal aren't you?!? Well DUH! So you compute a couple thousand instructions, find a prime and then scroll the terminal window which involves copying a half million pixels. No Quartz Extreme for you, the CPU is doing the scroll.
If this is the case I officially proclaim your test silly on the grounds that...
You can not possible read those numbers as they blast past
You are using a very general purpose tool, terminal, for a specific high performance operation.
What you really mean to show is how many primes/second are being found. A blasting terminal can convey that, but not with any accuracy and at a great CPU cost. Better to draw a scrolling graph at a fraction of the CPU cost and screen space, plus providing clear temporal trends and specific quantitative data. I'd also have a 'biggest one found yet' number that updates every second.
For more normal tasks...
I run OS X on a 300MHz G3 iBook and a 500MHz G3 iBook. Both do a fine job of document prep, email, and web surfing. The 500 does a fine job in Project Builder, the 300 is a little RAM light for that.
I was testing my opengl network visualization program on the 500MHz G3. 20fps (it is throttled at that, no sense going faster, its not like my T1 lines are going to gib me if I'm a reaction time later:-), full screen, 15% cpu utilization. Seems like plenty of power for me. (Ok, if you put a translucent window on top of the opengl view it goes to 1fps, 100% cpu, but that is probably because this unit can't run the Quartz Extreme.)
The Kirkwood United Methodist Church in Missouri had to take their rather massive steeple down due to rot. The church was unable to commit the resources for a replacement. AT&T had a replacement built and installed in exchange for antenna rights.
I'd link a picture, but the web site also rotted away. Trust me, the steeple was an integral architectural element and the building looked silly without it.
And for those worried about emissions, no need. AT&T phones barely work in the building. I suppose the antennas don't radiate down very well. Of course we are probably cooking the Christian Scientists and the Catholics next door...
Slashdot Mirror
Dude... The original poster is deluded. 10 years ago dictation sucked. 5 years ago dictation sucked. (I've got boxes of the stuff that were purchased and stunk so bad no one used them.) I'm talking about dictation that actually works well enough to use.
:-)
Modern PDAs do not have the equivalent of a PPro 200. StrongARM 200s are about the same as a Pentium 90 for general purpose work, and right on useless for floating point work.
According to my compilation benchmarks (integer and data pushing, no floating point), a PPro 200 is right about 4 times faster than a strongarm 200. On the other hand the SA200s are little more than twice as fast as a 486-66dx2.
The very latest xscale pdas are proably about twice the speed of the SA-110 200MHz ones, but I don't have any similar hardware to benchmark.
I just unlinked my benchmark page form the web because I hadn't updated it in years, but here is a link. Fun to reminise about all those machines I thought were so fast at the time...
metastones
Tiny devices like cell phones and PDAs don't have the CPU power for sexy, high quality voice recognition. They do however have wireless connectivity. So, solve the problem this way...
Install voice recognition servers, network connected boxes with powerful CPUs and the best voice recognition software you can get your hands on. A voice recognition client then just needs to send the voice data up to the server and get the translation back, say 100kbps up and some tiny amount back.
The payback comes because most devices will only use voice recognition for brief periods, so will present a negligible load on the servers. The dictation users will place a higher load on the servers, but even there, I'm guessing there is a lot of pausing involved. I'm also going to guess that some lag is acceptable for dictation. Presumably the person is thinking about what they are saying and proof reading later. This load can be prioritized lower to allow better immediate response for people issuing voice commands on their mobile devices.
Power consumption on the portable device will probably improve. They will have to operate their transmitter (think "talk time" vs. "on time"), but they won't need 5 watts of CPU doing recognition. (Guessing from a mobile G3 PPC, further validated, considering that the CPU spot of my iBook gets far hotter under solid use than a cellphone.)
So, just to pick numbers out of the air, a dual processor, high end commodity hardware voice server might serve 500 pda users giving intermittent commands and 6 simultaneous dictation users.
A company or school could easily justify the hardware cost of this service.
Now, someone go out and build one.
I just followed google news to a story at TIME. Their article surrounded by a kalidescope of ads took more than 10 seconds to load over my DSL line. The article was presented in an uncomfortably skinny space left after the mass of garish ads were plastered around it.
I can't tell you what any of the TIME ads were for. The only impression I carry away is that the article didn't have as much meat as I expected than the web site is annoying.
I'll take a tasteful ultramercial once a day instead of a delay to load a mass of ads on every page.
I just went to salon and read a premium article. Here is my synopsis...
If a 10 second ad can keep salon and their reporters working I'm all for it. The US needs independent journalists. (Even if they sometimes say things you'd rather not hear. Personally I'm offended by something in Salon every single day. If I wasn't, I wouldn't bother to read it.)
This is exactly what I have. I have a Sony built DirecTV reciever with Tivo built in.
The challenge for Tivo and Replay is to ensure that the TV and set top builders have to license their technology. It can't cost more than $1million to develop the code for a DVR. (Well it could, but it shouldn't.) If Tivo wants more than $1M/year return on their investment they have to legally prevent the TV builders from making and selling their own DVR software.
If Tivo has well written patents they can use litigation, if they have shaky patents they can use the treat of litigation. If they have worthless patents they are out of luck. Maybe the reason to cancel the lawsuit is to avoid a legal precedent on which patents are valid? Maybe each company has amassed a body of prior art on the other's patents? Just guessing.
I think if my life depended on my "partner" Microsoft I'd be worried that the PC in Tablet PC was short for Potassium Cyanide.
Good god...how many of these things could be lurking about in automotive airbags...
Well I would hope none. The only thing in your air bag should be the sodium azide and an igniter. The last thing you need in an accident is a bunch of loose capacitors and crap being blown into your face.
I just checked their web site. Looks like their only original series left will be Crossing Over. Sort of a misplaced show. High marks for fiction but no a zero for science.
They should change their name to the fi channel
I should have left the Nagasaki bomb out of the post, but cyclotrons were used for the uranium processing on the Hiroshima bomb. Gaseous diffusion was also used and soon became the preferred method but if you read up on "Calutron"s (modified cyclotrons) you will find their role in the early uranium processing.
this article has a pretty good description, including reasons why the Calutrons feel from favor.
Sorry about the confusion, I don't mean to say we should have killed more. I mean that we planned to kill more.
Cyclotrons can be used for uranium enrichment. Most of the uranium used in the Hiroshima (40*WTC911) and Nagaski (20*WTC911) bombs was purified in cyclotrons.
It takes a lot of energy, so you might want to have vast oil reserves that you aren't allowed to export in order to power the cylcotrons.
note: The Nagasaki bomb was not that much smaller, the weather was bad and we missed the population center and hit an industrial area instead.
What you need are dust corners. These are little brass trianglish affairs to go in square corners and turn them into radiused corners. They made sweeping and scrubbing easier back when it was all hand powered.
I am always surprised at the distance distortion that occurs in areas of high population density. I live in the middle of the United States (< 60 miles from the population centroid) and don't think twice about driving 500 miles to my vacation cabin. (Too d*mn hot where I live in the summer. Got to go north.)
I have had friends who live in Manhattan that consider 15 miles to be "far" and 200 miles to be an extraordinary distance to travel.
I wonder if people's definition of far is better correlated to "number of people passed" than "distance"?
Any slashdot readers from the Australian outback want to tell us what they think "far" is?
gcc is not a fast C compiler. It is a portable C compiler and it makes pretty good code, but it is not fast. g++ is a slow C++ compiler.
The Codewarrior compilers were fast compilers that made pretty good code.
Its all in the tradeoffs.
You can't have it all. Apple has been adding engineering hours to improve speed with gcc, but gcc will always value source language selection and target portability above all else.
Been there, done that. Nothing feels quite so "good" as paying tax on your charitable contributions because you gave away too much of your income. :-(
Ha! I am immune to the doubling of taxes. ~50% of my income already goes to taxes in one form or another.
The median US family in the US pays about 40% of their income to taxes. nifty graphs. (note: I didn't check out the credibility of the people hosting this report, but it matches what I have read in known reputable sources, so I will link it)
Maybe "taxes would have to be doubled refers to the social security taxes? That would be roughly a 33% increase in total taxes taking the median tax burden to something just over 50%.
Apple aimed wrong with the Newton. It was a good guess and a nice product, but missed the mass market. It turned out that most people were much happier with half the capability, half the size, and half the price. Palm came later, learned from Newton, aimed better, and ate the market.
Tivo on the other hand is a direct hit on what the mass market wants. The late comers to Tivo-ing can aim no better.
So does that trolling hook hurt when you bite it?
(Those rig workers are going to need to be careful with that 1 megawatt transmitter in the 2GHz range.)
So your playing B&W, you are half way through a gesture to create water on your poor people and the game decides to autosave, freezing input in the middle of the gesture. So you try to recover and complete the gesture or at least make it do something sane, but no. You get a fireball or something and incinerate your people. Bad god. Maybe the PC version of BW is better, but the Mac version could inspire one to injure the programmers.
Galeon on the other hand has nice gesture support.
The quicktime ogg library is not yet correct. Apple would never release that.
There seems to be some issue about returning a variable amount of data with each call, but given that variable bit rate mp3s work it seems to me that with an appropriate buffer for reblocking output data oggs should work as well. They may have gotten beyond that, since I last looked closely, but I see on their bug tracker they still have issues.
Still, if the API works out it would make sense for them to use quicktime and save some duplication.
Sorry if this is redundant, but I couldn't find the thread where the rest of us register to get cheap TiBooks from Apple.
Maybe Apple could have a special refurb sale for switching slashdotters? (Don't get your hopes up, refurbed TiBooks go for about $200 off retail. Damn those computers that keep their value.)
So sign up in a reply to this so Apple knows how many machines to set aside for the Switch-a-Nerd program.
I also don't work for Sun, I don't work for RedHat either, but do work for Debian. My choice of helping out with linux works for everybody though.
Wait a ding dong minute! You said terminal updates and simple C++ program. You are printing on standard output to a terminal aren't you?!? Well DUH! So you compute a couple thousand instructions, find a prime and then scroll the terminal window which involves copying a half million pixels. No Quartz Extreme for you, the CPU is doing the scroll.
If this is the case I officially proclaim your test silly on the grounds that...
For more normal tasks...
I run OS X on a 300MHz G3 iBook and a 500MHz G3 iBook. Both do a fine job of document prep, email, and web surfing. The 500 does a fine job in Project Builder, the 300 is a little RAM light for that.
I was testing my opengl network visualization program on the 500MHz G3. 20fps (it is throttled at that, no sense going faster, its not like my T1 lines are going to gib me if I'm a reaction time later
The Kirkwood United Methodist Church in Missouri had to take their rather massive steeple down due to rot. The church was unable to commit the resources for a replacement. AT&T had a replacement built and installed in exchange for antenna rights.
I'd link a picture, but the web site also rotted away. Trust me, the steeple was an integral architectural element and the building looked silly without it.
And for those worried about emissions, no need. AT&T phones barely work in the building. I suppose the antennas don't radiate down very well. Of course we are probably cooking the Christian Scientists and the Catholics next door...
Please let hitsquad@debian.org know when you get that billion dollars. The bequest acceleration team will take care of the rest.