If you consider 7 possible durations, and 10 tone pairs, that makes 70 tones. You have then 70^11 combinations, much more manageable than 10^77 (or 140^11 vs 10^154)
> Remember that you have additional data at the beginning of the stream (and possibly at the end) that indicates which compression program/algorithm is used.
Headers, and even dictionaries are not that big compared to the size of a compressed file. Plus you can always fake them, and it would be hard to tell wether you have a faulty archive or an encrypted file.
Maybe, but just think of how many compressed files fly over the network?
How many times have you asked (nicely or not) for people to send you compressed files because you have a slow connection or a small mail quota ? And all those mp3s, mpgs, jpgs ARE compressed data.
So your average data stream already has (or you may hope so) a rather high entropy. And the compression test does not work well.
> That's only one example. In France, there is a law that forbids people to use any kind of encryption.
Wrong, that law was repelled more than a year ago.
Re:SIGGRAPH 2001 observations
on
SIGGRAPH 2001
·
· Score: 1
"- There weren't many teapots in the papers this year, nor were as many bunnies as in previous years."
You had a nice teapot in the festival's hilarious intro sequence, as well as in the fake 'Alien' trailer (After Cube and Sphere... TEAPOT : "In space, noone hears your steam" )
Not quite - topics covered
on
SIGGRAPH 2001
·
· Score: 1
Here's approximatively what's in.
1 General programming
Optimization
DLLs
Dynamic typing
Property & Factory classes
Debugging/Profiling facilities
Stack winding, self modifying code (in asm)
Resource files
Input recording & playback
Text parsing
A tweaking UI
Random numbers
Bloom filter (hashing for early rejection)
3DSMAX exporting
Using Webcams
2 Math
IEEE floating point tricks
Vector & plane tricks
3D segment intersection
Inverse trajectory determination
Camera & flythrough pathes
Fractals
3 AI
Strategies for optimization
Micro-threads for AI (asm stack tricks)
RTS command queuing
Tile-based LOS & search
Influence maps
Strategic assessment techniques
Path finding
Fuzzy logic, NNs
4 Geometry management
VIPM methods
LOD & terrain tiles
Sphere trees, AABBs, Quadrees
Fishtank effect
Print-res rendering
Decals on arbitrary surfaces
Skyboxes
Self-shadowing
"Mario 64" 3rd person control
5 Graphics Display
Cartoon rendering
Dynamic per-pixel lighting
Procedural clouds in hardware
Faster lens flare, shadows
Impostors (replacing geometry with pictures)
Hardware-accelerated procedural texture animation (NVidia texture shaders can do cool things)
6 Audio programming
Design patterns
Voice reuse in a sample-based synthetizer
Software-based DSP effects
Interactive DSP pipelining
Music sequencers
API
Compared to the first book, I think that a lot has moved from the book to the CD which I haven't looked in detail yet.
Re:My "siggraph experience"
on
SIGGRAPH 2001
·
· Score: 1
At the back of my badge :
"No camera or recording devices are permitted at SIGGRAPH 2001. Abuse of this policy could result in the loss of your registration credentials."
And well, aside from the technical sessions & the animation festive, all you have left is... the exhibition floor. Not very exciting IMO.
Re:Siggraph seemed much smaller/calmer
on
SIGGRAPH 2001
·
· Score: 1
Wait until next year. The GeForce3 was released past the paper submission deadline. I fully expect to have loads of applications of the pixel/texture/vertex shaders next year.
As they say : "Speed-up any algorithm by a factor of 100 and everybody will find new applications for it."
Re:My "siggraph experience"
on
SIGGRAPH 2001
·
· Score: 1
SIGGRAPH's policy forbids cameras (not that people really care).
Siggraph archive available online in september
on
SIGGRAPH 2001
·
· Score: 5, Informative
This year there were signs posted here and there in the conference center saying that all the paper presentations, some panels and courses would be put online. They had their video crew recording them.
Hello world is already a pretty complex piece of software...
Think about it : you are actually doing an output to the screen.
What are the chances that the print instruction is completely bug-free ?
Especially if it needs to go through the OS rather than just being a memory copy...
[Win]2000 is pretty new and they are still tuning it up, so give them sometime and they would have it running better than NT.
From what I understood from the demonstration of WinXP we were given, Win2000 was only designed as a kind of 'proof of concept' on the way to WinXP. (Which looks even more bloated by the way.) So I think you can forget about them tuning Win2000.
That becomes just another weapon:
once everybody relies on high-accuracy
data from civilian receivers, just turn
encryption on again and watch the ensuing
chaos.
Just think about the speech recognition problems: if YOU cannot understand what is said, how can your computer ? Much less translate.
One advantage I can see of the translator being wearable is to carry it out. While it may be useful in a business environment - though people there are supposed to be educated enough to speak foreign languages (ok, well maybe not), it could not do anything for you out in the street, where they have to deal with mangled words and thick accents.
Kind of defeat the purpose.
Kind of defeat the purpose.
And in the meantime, while they have your money, they can invest it. Then, when THEY consider the 'experiment' over, they can give you a refund.
More seriously, that's an integral part of some financial institutions revenue. Ever wondered why banks process charges to your accounts way faster than they process deposits ? Even taking the necessary communication between banks (after all the money has to come from somewhere) doesn't satisfactorily explain the two to three (business) days difference I have observed in some cases.
Sure, over a few dollars it doesn't make a difference, but even a couple of days' worth of interest can add up for a bank.
Slashdot Mirror
no, that would be 70^11
If you consider 7 possible durations, and 10 tone pairs, that makes 70 tones. You have then 70^11 combinations, much more manageable than 10^77 (or 140^11 vs 10^154)
For all the Illuminati wannabe among us.
> Remember that you have additional data at the beginning of the stream (and possibly at the end) that indicates which compression program/algorithm is used.
Headers, and even dictionaries are not that big compared to the size of a compressed file. Plus you can always fake them, and it would be hard to tell wether you have a faulty archive or an encrypted file.
Maybe, but just think of how many compressed files fly over the network?
How many times have you asked (nicely or not) for people to send you compressed files because you have a slow connection or a small mail quota ? And all those mp3s, mpgs, jpgs ARE compressed data.
So your average data stream already has (or you may hope so) a rather high entropy. And the compression test does not work well.
> That's only one example. In France, there is a law that forbids people to use any kind of encryption.
Wrong, that law was repelled more than a year ago.
"- There weren't many teapots in the papers this year, nor were as many bunnies as in previous years."
You had a nice teapot in the festival's hilarious intro sequence, as well as in the fake 'Alien' trailer (After Cube and Sphere... TEAPOT : "In space, noone hears your steam" )
Here's approximatively what's in.
1 General programming
Optimization
DLLs
Dynamic typing
Property & Factory classes
Debugging/Profiling facilities
Stack winding, self modifying code (in asm)
Resource files
Input recording & playback
Text parsing
A tweaking UI
Random numbers
Bloom filter (hashing for early rejection)
3DSMAX exporting
Using Webcams
2 Math
IEEE floating point tricks
Vector & plane tricks
3D segment intersection
Inverse trajectory determination
Camera & flythrough pathes
Fractals
3 AI
Strategies for optimization
Micro-threads for AI (asm stack tricks)
RTS command queuing
Tile-based LOS & search
Influence maps
Strategic assessment techniques
Path finding
Fuzzy logic, NNs
4 Geometry management
VIPM methods
LOD & terrain tiles
Sphere trees, AABBs, Quadrees
Fishtank effect
Print-res rendering
Decals on arbitrary surfaces
Skyboxes
Self-shadowing
"Mario 64" 3rd person control
5 Graphics Display
Cartoon rendering
Dynamic per-pixel lighting
Procedural clouds in hardware
Faster lens flare, shadows
Impostors (replacing geometry with pictures)
Hardware-accelerated procedural texture animation (NVidia texture shaders can do cool things)
6 Audio programming
Design patterns
Voice reuse in a sample-based synthetizer
Software-based DSP effects
Interactive DSP pipelining
Music sequencers
API
Compared to the first book, I think that a lot has moved from the book to the CD which I haven't looked in detail yet.
At the back of my badge :
"No camera or recording devices are permitted at SIGGRAPH 2001. Abuse of this policy could result in the loss of your registration credentials."
And well, aside from the technical sessions & the animation festive, all you have left is... the exhibition floor. Not very exciting IMO.
Wait until next year. The GeForce3 was released past the paper submission deadline. I fully expect to have loads of applications of the pixel/texture/vertex shaders next year.
As they say : "Speed-up any algorithm by a factor of 100 and everybody will find new applications for it."
SIGGRAPH's policy forbids cameras (not that people really care).
I picked my copy at the CRM booth.
This year there were signs posted here and there in the conference center saying that all the paper presentations, some panels and courses would be put online. They had their video crew recording them.
The address is http://online.siggraph.org
> Maybe not the DMCA, but the FCC might have something to say.
Yeah ! Sue the radiations !
You can always get another nation to launch the satellite... and pride be damned. Of course, with the recent failures of Ariane 5...
So it's Mel-Bourne again, right ?
If it can get more people reading and less people watching TV. ( Or spending time online, me included )
Hello world is already a pretty complex piece of software...
Think about it : you are actually doing an output to the screen.
What are the chances that the print instruction is completely bug-free ?
Especially if it needs to go through the OS rather than just being a memory copy...
[Win]2000 is pretty new and they are still tuning it up, so give them sometime and they would have it running better than NT.
From what I understood from the demonstration of WinXP we were given, Win2000 was only designed as a kind of 'proof of concept' on the way to WinXP. (Which looks even more bloated by the way.) So I think you can forget about them tuning Win2000.
Cause of accident : Victim was downloading pr0n on the net.
That becomes just another weapon :
once everybody relies on high-accuracy data from civilian receivers, just turn encryption on again and watch the ensuing chaos.
Not exactly. What the vertex+pixel shaders enable you to do is :
- hardware mesh deformation
- hardware bump mapping
Ray-tracing is a completely different scheme than current hardware polygon renderers.
Just think about the speech recognition problems: if YOU cannot understand what is said, how can your computer ? Much less translate.
One advantage I can see of the translator being wearable is to carry it out. While it may be useful in a business environment - though people there are supposed to be educated enough to speak foreign languages (ok, well maybe not), it could not do anything for you out in the street, where they have to deal with mangled words and thick accents.
Kind of defeat the purpose. Kind of defeat the purpose.
I'll soon have patents on the one-double-clic patenting process.
Send royalties to my Swiss account or be prepared to meet my lawyers.
And in the meantime, while they have your money, they can invest it. Then, when THEY consider the 'experiment' over, they can give you a refund.
More seriously, that's an integral part of some financial institutions revenue. Ever wondered why banks process charges to your accounts way faster than they process deposits ? Even taking the necessary communication between banks (after all the money has to come from somewhere) doesn't satisfactorily explain the two to three (business) days difference I have observed in some cases.
Sure, over a few dollars it doesn't make a difference, but even a couple of days' worth of interest can add up for a bank.
You will get judges who know COBOL.