Seriously though, you should have a listen to Mike Batt's "Tarot Suite". It's one of my favourite albums simply because it's a bit unusual. (For those in Sydney who listened to MMM-FM, the theme tune/jingle they used was taken from the album).
Well, the Beattles should jump in here. For awhile they had the longest silence on any major recording album...but I really am not sure if you can compare the quality of the silences. I mean, a digitally mastered silence has the potential of being leap years ahead of the lower quality vinyl silences of yesterday.
Are you refering to the "silent" dog whistle put on for Paul McCartney's dog? (The name escapes me, (Martha?) but I think it was used for a title of a track). The whistle was hardly silent on the vinyl (at least back in the days when my hearing went up to 25khz), but I wonder if it was filtered out for digital recordings?
Sorry, this post is merely to cancel the moderation I did (on one of the posts in this thread). There definitely seems to be a bug in the moderation system as this is the second time it's occured for me.
I chose "interesting" (however the particular posting may have been marked insightful) but after hitting moderate, the post's score was decremented.
That's a pity. The 4:2:0 => 4:2:2 convertor in the PowerVR chip hardware would save quite a bit of CPU time. I was also wondering if some of the motion-comp could be done using the rendering hardware.
not enough is known about those two
and from what i understand (considering i wrote the video driver) using yuv 4:2:0 instead of 4:2:2 wouldnt speed it up that much as the loop he uses to pack the yuv frames packs them straight into 4:2:2
IIRC, the 4:2:0 conversion processing in CLX1 (the PowerVR chip) DMAs data from separate Y Cr Cb blocks and reformats it to 4:2:2. This would save bandwidth and SH4 time (even if the packing routine did do the reformatting).
As for motion comp, I was thinking that perhaps you could use the 3D hardware (with its texture filtering etc) to do copying and (sub-pixel) moving of blocks to manufacture the "I" (?) frames.
it uses 4:2:2 => rgb but so far thats about it. and i've played 44khz stereo mp3 on it with 600kbit and vcd res (little over 320x240) and it played just fine.
That's a pity. The 4:2:0 => 4:2:2 convertor in the PowerVR chip hardware would save quite a bit of CPU time. I was also wondering if some of the motion-comp could be done using the rendering hardware.
If you had bothered to actually read that site, then you'd know that you have to reencode the videos at lower resolution (320x240 seems to be recommended), very low bitrate (they recommend 500kbit/s) and 22KHz mono audio to ensure that the DC can decode it quickly enough.
I couldn't see any specific mention of a Dreamcast decoder project on that site, so I'm wondering if it's making any use of the graphics hardware at all (e.g. 4:2:0 => RGB conversion). Does anyone know?
Apologies to the poster... I selected "Interesting" when moderating this but the system seems to have assigned "overrated". I don't know who to report this bug(?)to.
If I remember correctly, the Dreamcast sports a first gen Kyro graphics chipset....
Sorry, but Dreamcast uses a PowerVR series 2 chip (the CLX2) while Kyro I & II are series 3 chips. There are, of course, a large number of similarities, but there are also quite a number of significant differences.
For example, Series 2 used vector quantisation texture compression (generally 2bpp although it can go lower!) while series 3 has S3TC/DXTC. Series 3 added several new features such as 8 layer multi-texturing (although you could do emulate this on series 2 because it had a scratch-pad "accumulation" buffer).
CLX2 also included some "special" functions that were requested by Sega that aren't in any of the PC chips.
In college I was able to drive a 1" steel ball through a 12" brick wall with a 3 foot railgun. (Teflon tube with large coils spaced at a semi-logarythmic scale along the length with a simple computer control.)
Pardon for butting in, but isn't what you are describing a coil gun?
(That is where a magnetic projectile is drawn into a sequence of coils and each is, in turn, switched off as the object approaches the centre of the coil?)
You wouldn't need any control for a rail gun as it passes a current through the projectile (which sits between charged rails). The current thus creates a magnet. The project and rails are then inside a larger field magnetic field which repels the projectile out the 'barrel'.
Yes, but then you'd be launching the probe directly towards Earth
Err... the (instaneous) velocity of the ISS is perpendicular to the radius of orbit (as would be the drag but in the opposite direction)) and so surely you wouldn't be aiming that way!
Except there's that funny problem about space. If you launch a spacecraft off of the ISS with a magnetic launcher, then the ISS itself
is moved in the opposite direction (based on the relative mass of the station to the vehicle).
I might be missing something here, but it seems to me that you can kill two birds with one stone by aiming and timing such a launch correctly.
Clearly, due to the tiny amount of drag, over time the oribit of the station decays and so will occasionally need a push. If you launched the probe so that it was in the "opposite" direction to the velocity of the ISS, it would boost the orbit of ISS (effectively becoming a rocket motor). Also given the (assumed) differences in mass between the ISS and the probe, the speed of the probe should be much greater. You simply wait till the orbit position of the ISS is such that the probe will be launched in the appropriate direction.
"Yeah, Hugo Weaving (Elrond/Matrix's Agent Smith) wasn't really made up enough to get away from his character in 'The Matrix'...same impassionate voice too.
I trust you would not have prefered him reusing his character from "Priscilla. Queen of the Desert":-)
.. in the prime number generation code used, for example, in the RSA Public Key encryption system.
In that you need to manufacture very large prime numbers, the process of which usually involves randomly choosing a large odd number and (after running a few trial divisions for trivial factors) using a probabalistic prime number test.
Each execution of the test will return either a "definitely not a prime" message or a "maybe prime". The chance of it returning a "maybe prime" when it's actually a composite is someting like 1 in 4. You basically run the test many times with different seeds until either it says it's a composite (and you choose a new random number) or you have a high confidence that your number is in fact prime.
Of course the result might be wrong, but you're more likely to die by the sun exploding after just winning the lottery so I wouldn't worry it.
The 2% failure sounds pretty damned good provided the re-runs are independent.
"Well, since the SH4 can multiply 4x4 matrices directly why not talking about the powerful 512-bit CPU of the Dreamcast:-)"
Actually that takes about 4 cycles, so the FP unit is not quite that powerful.
If we are going to talk about big numbers, (bus envy anyone?), the PowerVR CLX2 chip in the DC has data paths which are 1024 bits wide. It makes for some very, very high fill rates when you have some level of opaque overdraw or volume shadows. 8*)
On second thoughts, it might be difficult to use three different textures in with this interface unless you can write to an intermediate storage as done in the multitexturing facilities of today's graphics hardware.
First, IANAPL but I have written a few patents in the graphics field.
My understanding too was that the mask image (i.e. equivalent of the alpha) could be a multi-channel ARGB signal wherein each channel of the "source" and "destination" could be individually blended by their respective channels. This is an extension over standard alpha blending.
As for the possible existence of "Prior Art", at least with respect to the first claim in that patent, does anyone have a date for the SGI's
"Graphics Library Programming Guide Version 2.0"? I have a copy here but it doesn't have any dates. Note that this is the GL before it was "opened". I think this document dates from the early 1990s but an exact date would be useful.
The reason I ask is that it contained a lot of the facilities that are in OpenGL, and if you look at the section 15.2 "Blending" you will see that it is possible to do a channel by channel blend by selecting calling
blendfunction(BF_SC, BF_MSC)
along with a texturing operation. I believe this may encapsulate all the functionality described in the 5379129 patent. Since that patent was filed in 1992, there is a good chance that the SGI documentation may consititute full prior art.
If you had actually read the Kama Sutra, then no doubt you would know that it suggests that women should be experts in the field of secret writing, i.e., cryptography, (according to David Khan's book)
More information on Babbage (and a bit on Turing).
on
The Difference Engine
·
· Score: 1
"You know, I learned some of these names in school, I'd heard of Alan Turing but until recently I knew very little about him, ditto for Babbage."
I've just been reading "The Code Breakers" by James Kahn, which although a bit dated in some parts, is a fascinating read of Cryptography VS Cryptanalysis throughout the ages.
Babbage was also into cryptology - both the creation of ciphers and the cracking of others. For example, the famous "Playfair" cipher was actually invented by him but named after one of his friends who was trying to promote its use. IIRC, he also used to have fun deciphering private messages in the newspaper personal columns (often involving love affairs in the prudish Victorian era).
"lahalito, dilto, Montino, Mahalito, and the all mighty Tiltowait"
OMG! Until I read that I thought I'd forgotten it, but it all comes flooding back.
I spent many, many, many hours in "The Bard's Tale" on the apple macintosh - first playing the game and then hacking the database to modify the maps. I even wrote my own disassembler to assist in the process. I wonder how much lower my GPA was because of it. Doesn't matter, I suppose, I still graduated and got a job doing 3D graphics hardware/software.
I still have the maps, all carefully hand drawn, lying around somewhere. However, I used to hate how you could be teleported to a different point on the grid without realising it, which would screw up the cartography!
Try using ghostscript/GSView which will display the postscript file directly. (A quick search on Google gave the following link which should be useful
GhostScript
Comment 2: I don't think this is ever really likely to be part of 'consumer' system. FPGA's are great for
1) prototyping circuits that will later be implemented as an ASIC where the cost of "respinning" a chip is extremely high or 2) Situations where the system is only produced in very small numbers.
The main problems with FPGAs are that they are
1)Expensive!.
2) Relatively small in terms of the gates they can implement and
3) The clock speed that can be achieved is probably about an order of magnitude lower than an equivalent ASIC.
For many situations a multi-CPU system may be a much better option, and I certainly think that they'd be impractical for a mass produced system.
Slashdot Mirror
Seriously though, you should have a listen to Mike Batt's "Tarot Suite". It's one of my favourite albums simply because it's a bit unusual. (For those in Sydney who listened to MMM-FM, the theme tune/jingle they used was taken from the album).
Are you refering to the "silent" dog whistle put on for Paul McCartney's dog? (The name escapes me, (Martha?) but I think it was used for a title of a track). The whistle was hardly silent on the vinyl (at least back in the days when my hearing went up to 25khz), but I wonder if it was filtered out for digital recordings?
Simon
Sorry, this post is merely to cancel the moderation I did (on one of the posts in this thread). There definitely seems to be a bug in the moderation system as this is the second time it's occured for me.
I chose "interesting" (however the particular posting may have been marked insightful) but after hitting moderate, the post's score was decremented.
Simon
PS: Who do you report these sort of bugs to?
I think that is one of the most frightening things I've read in a long time!!!!
Simon
IIRC, the 4:2:0 conversion processing in CLX1 (the PowerVR chip) DMAs data from separate Y Cr Cb blocks and reformats it to 4:2:2. This would save bandwidth and SH4 time (even if the packing routine did do the reformatting).
As for motion comp, I was thinking that perhaps you could use the 3D hardware (with its texture filtering etc) to do copying and (sub-pixel) moving of blocks to manufacture the "I" (?) frames.
Simon
That's a pity. The 4:2:0 => 4:2:2 convertor in the PowerVR chip hardware would save quite a bit of CPU time. I was also wondering if some of the motion-comp could be done using the rendering hardware.
Simon
I couldn't see any specific mention of a Dreamcast decoder project on that site, so I'm wondering if it's making any use of the graphics hardware at all (e.g. 4:2:0 => RGB conversion). Does anyone know?
Simon
Well that's a heck of a lot nicer than what he did to Alderaan! There certainly were no backups there.
(Speaking of which, Terry Pratchett has a fun spin on this concept in "The Thief of Time")
Simon
Is that a (Phillips?) CDP1802? I think mine's gathering dust somewhere. It was too damn painful to program.
Apologies to the poster... I selected "Interesting" when moderating this but the system seems to have assigned "overrated". I don't know who to report this bug(?)to.
Sorry, but Dreamcast uses a PowerVR series 2 chip (the CLX2) while Kyro I & II are series 3 chips. There are, of course, a large number of similarities, but there are also quite a number of significant differences.
For example, Series 2 used vector quantisation texture compression (generally 2bpp although it can go lower!) while series 3 has S3TC/DXTC. Series 3 added several new features such as 8 layer multi-texturing (although you could do emulate this on series 2 because it had a scratch-pad "accumulation" buffer).
CLX2 also included some "special" functions that were requested by Sega that aren't in any of the PC chips.
Cheers
Simon
Pardon for butting in, but isn't what you are describing a coil gun?
(That is where a magnetic projectile is drawn into a sequence of coils and each is, in turn, switched off as the object approaches the centre of the coil?)
You wouldn't need any control for a rail gun as it passes a current through the projectile (which sits between charged rails). The current thus creates a magnet. The project and rails are then inside a larger field magnetic field which repels the projectile out the 'barrel'.
Simon
Err... the (instaneous) velocity of the ISS is perpendicular to the radius of orbit (as would be the drag but in the opposite direction)) and so surely you wouldn't be aiming that way!
Simon
I might be missing something here, but it seems to me that you can kill two birds with one stone by aiming and timing such a launch correctly.
Clearly, due to the tiny amount of drag, over time the oribit of the station decays and so will occasionally need a push. If you launched the probe so that it was in the "opposite" direction to the velocity of the ISS, it would boost the orbit of ISS (effectively becoming a rocket motor). Also given the (assumed) differences in mass between the ISS and the probe, the speed of the probe should be much greater. You simply wait till the orbit position of the ISS is such that the probe will be launched in the appropriate direction.
Simon
/me waits to be told what an idiot I am
"Yeah, Hugo Weaving (Elrond/Matrix's Agent Smith) wasn't really made up enough to get away from his character in 'The Matrix'...same impassionate voice too.
:-)
I trust you would not have prefered him reusing his character from "Priscilla. Queen of the Desert"
.. in the prime number generation code used, for example, in the RSA Public Key encryption system.
In that you need to manufacture very large prime numbers, the process of which usually involves randomly choosing a large odd number and (after running a few trial divisions for trivial factors) using a probabalistic prime number test.
Each execution of the test will return either a "definitely not a prime" message or a "maybe prime". The chance of it returning a "maybe prime" when it's actually a composite is someting like 1 in 4. You basically run the test many times with different seeds until either it says it's a composite (and you choose a new random number) or you have a high confidence that your number is in fact prime.
Of course the result might be wrong, but you're more likely to die by the sun exploding after just winning the lottery so I wouldn't worry it.
The 2% failure sounds pretty damned good provided the re-runs are independent.
Simon
"Well, since the SH4 can multiply 4x4 matrices directly why not talking about the powerful 512-bit CPU of the Dreamcast :-)"
Actually that takes about 4 cycles, so the FP unit is not quite that powerful.
If we are going to talk about big numbers, (bus envy anyone?), the PowerVR CLX2 chip in the DC has data paths which are 1024 bits wide. It makes for some very, very high fill rates when you have some level of opaque overdraw or volume shadows. 8*)
Simon
On second thoughts, it might be difficult to use three different textures in with this interface unless you can write to an intermediate storage as done in the multitexturing facilities of today's graphics hardware.
Simon
First, IANAPL but I have written a few patents in the graphics field.
My understanding too was that the mask image (i.e. equivalent of the alpha) could be a multi-channel ARGB signal wherein each channel of the "source" and "destination" could be individually blended by their respective channels. This is an extension over standard alpha blending.
As for the possible existence of "Prior Art", at least with respect to the first claim in that patent, does anyone have a date for the SGI's
"Graphics Library Programming Guide Version 2.0"? I have a copy here but it doesn't have any dates. Note that this is the GL before it was "opened". I think this document dates from the early 1990s but an exact date would be useful.
The reason I ask is that it contained a lot of the facilities that are in OpenGL, and if you look at the section 15.2 "Blending" you will see that it is possible to do a channel by channel blend by selecting calling
blendfunction(BF_SC, BF_MSC)
along with a texturing operation. I believe this may encapsulate all the functionality described in the 5379129 patent. Since that patent was filed in 1992, there is a good chance that the SGI documentation may consititute full prior art.
Simon
If you had actually read the Kama Sutra, then no doubt you would know that it suggests that women should be experts in the field of secret writing, i.e., cryptography, (according to David Khan's book)
"You know, I learned some of these names in school, I'd heard of Alan Turing but until recently I knew very little about him, ditto for Babbage."
I've just been reading "The Code Breakers" by James Kahn, which although a bit dated in some parts, is a fascinating read of Cryptography VS Cryptanalysis throughout the ages.
Babbage was also into cryptology - both the creation of ciphers and the cracking of others. For example, the famous "Playfair" cipher was actually invented by him but named after one of his friends who was trying to promote its use. IIRC, he also used to have fun deciphering private messages in the newspaper personal columns (often involving love affairs in the prudish Victorian era).
Simon
"lahalito, dilto, Montino, Mahalito, and the all mighty Tiltowait"
OMG! Until I read that I thought I'd forgotten it, but it all comes flooding back.
I spent many, many, many hours in "The Bard's Tale" on the apple macintosh - first playing the game and then hacking the database to modify the maps. I even wrote my own disassembler to assist in the process. I wonder how much lower my GPA was because of it. Doesn't matter, I suppose, I still graduated and got a job doing 3D graphics hardware/software.
I still have the maps, all carefully hand drawn, lying around somewhere. However, I used to hate how you could be teleported to a different point on the grid without realising it, which would screw up the cartography!
Simon
IANAPL, but, IIRC, I heard that the method used to derive the ARM instruction set has been patented and this then effectively protects their IP.
I can't verify this because I'm far too lazy to go searching through the USPTO server to look for the patent.
Simon
Comment 1 RE:Linked directly to Postscript?
Try using ghostscript/GSView which will display the postscript file directly. (A quick search on Google gave the following link which should be useful
GhostScript
Comment 2: I don't think this is ever really likely to be part of 'consumer' system. FPGA's are great for
1) prototyping circuits that will later be implemented as an ASIC where the cost of "respinning" a chip is extremely high or
2) Situations where the system is only produced in very small numbers.
The main problems with FPGAs are that they are
1)Expensive!.
2) Relatively small in terms of the gates they can implement and
3) The clock speed that can be achieved is probably about an order of magnitude lower than an equivalent ASIC.
For many situations a multi-CPU system may be a much better option, and I certainly think that they'd be impractical for a mass produced system.
Simon
Perhaps it would have been the perfect time to film the scene where the Nazgul are washed away in the torrent near Rivendel?