I've had a handheld computer since 2000, and that was already fourth or fifth generation. Symbol makes ruggedized handhelds running both PalmOS and WinCE that are already widely used by people doing basically the same kind of thing the census takers are... walking up to the door and running through a checklist. This isn't new technology, this is stuff that's already been out for years.
I suspected that might be the case. When you're able to pick from the best albums of the past 50 years you get a different view of the medium than if you're buying recent music. Albums that only have a couple of decent tracks don't even show up on your radar.
I'm not an "album person". I've bought a couple of full albums on iTunes with their "Complete My Album" offer, but it's not common. Mostly I hear a song I like and I buy that song and a few other tracks from the same album that appeal to me. If I had to buy the whole shebang, I probably wouldn't bother buying anything nearly as often.
Also I suspect that a lot of tracks I think of as filler are an integral part of the album for you... it may be a matter of how you feel about a track that isn't intended to stand on its own: I rarely play two tracks from the same album in sequence, let alone playing through an album in sequence.
Those plates do not tend to be easy for a small, low power electronic device to read and process.
I've worked on trackside sensors in the railroad industry, and we found the most reliable way to recognize a train was by measuring the timing of the wheels as the passed the sensor. Optical scanners would have been a last resort. The union wouldn't let us put transponders on the trains, or we'd have done that.
Having every car transmitting a unique easily-fingerprinted signal? That'd make things so easy. It doesn't matter if you can't decode it, there has to be *some* preamble that tells the receiver that it's getting a signal and what to expect, and unless there's a lot of effort put into anonymizing as well as encrypting the preamble that by itself would give you a lot of info.
Remember, too, these sensors have to be low power and cheap.
For example, Freescale's MPXY8300 has 16k of Flash memory, and 512 *bytes* of RAM, and runs at 8 MHz. What kind of crypto are you going to use on that?
Encryption will allow you to keep them from knowing what your tire pressure is, but you'd need to have the signal anonymized as well to keep someone from fingerprinting it.
I didn't have problems following the story line, I just didn't care about the story line. I didn't have problems understanding the characters, I just found them unconvincing and couldn't bring myself to care about them. For Cryptonomicon if there was a problem with my interest in the subject matter it would only be that I was too interested, so the shortcomings were all the more glaring.
I don't expect them to produce a decent headless Mac, but then a month before the Mini came out I bought a Radeon 9200 for my "Beige G4" because I wasn't expecting them to ever produce a headless Mac again, decent or not.
But I agree, they need a "Mini pro" with a 3.5" drive, a real GPU, audio in, and full power USB ports.
Cryptonomicon was OK but I had to hold my nose and slog through places. Quicksilver I had to force myself to finish. But you're right about the shortage of really good SF writers. My shortlist is getting pretty short... here's the names that come to mind right now: Charlie Stross, David Langford, Greg Egan, Vernor Vinge, Iain M Banks, and Linda Nagata.
If your favorite cool SF writer isn't in this list, I'm sorry! I'm sorry! It's off the top of my head.
I don't think that building an FPGA into the video card is going to be cost effective. The point is that even implemented as an FPGA Philipp's raytracing engine was rendering as fast as a hypothetical 8 GHz P4.
The thing is, a Core 2 Quad also has the performance of a hypothetical 8 GHz P4, and it's much easier to program than an FPGA. An FPGA in a graphics card would get used by maybe one game... one subsidized by the vendor... and otherwise act like nothing more than a load on the power supply.
The point to it being an FPGA is not that an FPGA is the best way to do the job, but that it's not the best way to do the job but the design was STILL terribly effective. If you built an RPU using current GPU technology it would run 6-8 times as fast and have about 100 times the effective transistor budget. The SaarCOR RPU was severly limited by the requirement that it fit in an FPGA. It had no integer unit, it couldn't even compute new pointers when traversing geometry!
His earlier books were great, but somewhere in Cryptonomicon he seems to have lost the plot, literally. I had a lot of trouble actually caring about the characters in Cryptonomicon... and I couldn't really care much about the background or plot either... it all seemed to be an excuse for him to write about the places he'd been as a hacker tourist and try and drum up geek cred... and he didn't seem to understand what bits of geek culture were things his allegedly competent protagonist should care about. The Baroque Cycle? I gave up halfway through the second one. It was like reading the "Swiss Family Robinson" version of the Renaissance. You know how "Swiss Family Robinson" was kind of like teenager's wish-fulfillment version of "Robinson Crusoe"? That's how I felt about Quicksilver... too many protagonists had too many convenient 20th century attitudes and too much 20th century understanding of biology and physics.
If we can convince the public opinion that MS doesn't implement OOXML, it will need to.
No, if we can convince the public opinion that MS doesn't implement OOXML and that it matters, it will need to. But if the public in general actually cared about stuff like that they wouldn't be using Office file formats for anything but working copies of documents anyway.
Look, if Microsoft implemented ODF and did a great job of it, and it was included as an option in Office right now people would still overwhelming be passing documents around in.DOC and.XLS format. And government bodies would just check "supports ODF" and still send out proposals and accept tenders in.DOC and.XLS. But at least they'd mostly also accept ODF, whether they cared about it or not. But most of them won't... again, if they did, they'd already be using software that supported open standards.
Hell, there's a company I do business with that provides reports in.XLS format. The reports don't contain any more information than a simple.CSV would. People have asked them to provide.CSV as an option, and they haven't. The company's servers are all running Linux and they're active users of open source software. If people like that don't care enough to use open formats, why the heck do you think the public opinion will make a difference?
Well, yes, that's just the flip side of the fact that you don't need to do occlusion culling, which you might notice I mentioned. that is to say, I did not suggest splitting up the scene, but rather replicating it.
What I suggested was giving the pipelines local memory and caching the static part of the mesh. Since that's almost all of the mesh, almost all of the time, they will only need to fetch dynamically changing meshes. With some analog for vertex buffer objects you could even cache subelements like vehicles and just fetch location changes.
The implication in the term being that there's no need for communication between each core or thread or process: they each just get handed a rectangular portion of the offscreen screen memory, and they do their job alone, and when they're all done, then the screen can be flipped to show the results.
Well, yes, you could do that. You could also render your scenes by having the pipelines re-parse shader programs from text for every frame. But you don't.
If access to the mesh is a bottleneck, give the pipelines local cache and cache static portions of the mesh. If access to textures is a bottleneck, cache them. You don't need to update the mesh when the camera moves, or when the lights move, because raytracers doesn't need (or want) to precompute occlusion.
Slashdot Mirror
I've had a handheld computer since 2000, and that was already fourth or fifth generation. Symbol makes ruggedized handhelds running both PalmOS and WinCE that are already widely used by people doing basically the same kind of thing the census takers are... walking up to the door and running through a checklist. This isn't new technology, this is stuff that's already been out for years.
Also iTMS is the worst kind of download, DRMed.
No, the worst kind of download is DRMed, time-limited, in a player that's got kernel DRM support.
That's the bottom line: the "end user" is not Microsoft's customer, the hardware manufacturers are.
I suspected that might be the case. When you're able to pick from the best albums of the past 50 years you get a different view of the medium than if you're buying recent music. Albums that only have a couple of decent tracks don't even show up on your radar.
I'm not an "album person". I've bought a couple of full albums on iTunes with their "Complete My Album" offer, but it's not common. Mostly I hear a song I like and I buy that song and a few other tracks from the same album that appeal to me. If I had to buy the whole shebang, I probably wouldn't bother buying anything nearly as often.
Also I suspect that a lot of tracks I think of as filler are an integral part of the album for you... it may be a matter of how you feel about a track that isn't intended to stand on its own: I rarely play two tracks from the same album in sequence, let alone playing through an album in sequence.
What kind of music do you buy?
I suppose if you stick to Classical, Classics, and "Best Of" collections...
It's the fighting that counts. And what would be the consequences if they didn't?
In NYC? Do you need to ask? The same old stuff. "Human sacrifice, dogs and cats, living together... mass hysteria!"
It's always the animals that suffer.
Have you heard about CDs? No DRM, cheaper than downloading track-by-track,
I buy a CD, I spend $15 and get 3 tracks that end up in my shuffle playlist.
I buy individual tracks, If half the tracks I buy turn out to be filler, I spend $6 and get 3 tracks that end up in my shuffle playlist.
I like the Ajaxiness.
This is Slashdot. It bloody well needs to be completely usable without Javascript. What do they think this is, ZDNet?
I think they need to roll back the whole frigging site about a year.
It's got a long way to go to beat "Netizen" or "Podcast".
I assume you're disqualifying "Blogosphere" because it's derived from "Blog".
Well the other half of the battle is getting something that looks good on a TV screen.
So. At least it isn't a hardware problem.
You don't even need to do that. Just burn your tracks to CD-RW and rip them again. You even keep the ID3 tags.
What's wrong with a low end mini-itx box with passive cooling?
Those plates do not tend to be easy for a small, low power electronic device to read and process.
I've worked on trackside sensors in the railroad industry, and we found the most reliable way to recognize a train was by measuring the timing of the wheels as the passed the sensor. Optical scanners would have been a last resort. The union wouldn't let us put transponders on the trains, or we'd have done that.
Having every car transmitting a unique easily-fingerprinted signal? That'd make things so easy. It doesn't matter if you can't decode it, there has to be *some* preamble that tells the receiver that it's getting a signal and what to expect, and unless there's a lot of effort put into anonymizing as well as encrypting the preamble that by itself would give you a lot of info.
Remember, too, these sensors have to be low power and cheap.
For example, Freescale's MPXY8300 has 16k of Flash memory, and 512 *bytes* of RAM, and runs at 8 MHz. What kind of crypto are you going to use on that?
Encryption will allow you to keep them from knowing what your tire pressure is, but you'd need to have the signal anonymized as well to keep someone from fingerprinting it.
I didn't have problems following the story line, I just didn't care about the story line. I didn't have problems understanding the characters, I just found them unconvincing and couldn't bring myself to care about them. For Cryptonomicon if there was a problem with my interest in the subject matter it would only be that I was too interested, so the shortcomings were all the more glaring.
I don't expect them to produce a decent headless Mac, but then a month before the Mini came out I bought a Radeon 9200 for my "Beige G4" because I wasn't expecting them to ever produce a headless Mac again, decent or not.
But I agree, they need a "Mini pro" with a 3.5" drive, a real GPU, audio in, and full power USB ports.
Cryptonomicon was OK but I had to hold my nose and slog through places. Quicksilver I had to force myself to finish. But you're right about the shortage of really good SF writers. My shortlist is getting pretty short... here's the names that come to mind right now: Charlie Stross, David Langford, Greg Egan, Vernor Vinge, Iain M Banks, and Linda Nagata.
If your favorite cool SF writer isn't in this list, I'm sorry! I'm sorry! It's off the top of my head.
I don't think that building an FPGA into the video card is going to be cost effective. The point is that even implemented as an FPGA Philipp's raytracing engine was rendering as fast as a hypothetical 8 GHz P4.
The thing is, a Core 2 Quad also has the performance of a hypothetical 8 GHz P4, and it's much easier to program than an FPGA. An FPGA in a graphics card would get used by maybe one game... one subsidized by the vendor... and otherwise act like nothing more than a load on the power supply.
The point to it being an FPGA is not that an FPGA is the best way to do the job, but that it's not the best way to do the job but the design was STILL terribly effective. If you built an RPU using current GPU technology it would run 6-8 times as fast and have about 100 times the effective transistor budget. The SaarCOR RPU was severly limited by the requirement that it fit in an FPGA. It had no integer unit, it couldn't even compute new pointers when traversing geometry!
His earlier books were great, but somewhere in Cryptonomicon he seems to have lost the plot, literally. I had a lot of trouble actually caring about the characters in Cryptonomicon... and I couldn't really care much about the background or plot either... it all seemed to be an excuse for him to write about the places he'd been as a hacker tourist and try and drum up geek cred... and he didn't seem to understand what bits of geek culture were things his allegedly competent protagonist should care about. The Baroque Cycle? I gave up halfway through the second one. It was like reading the "Swiss Family Robinson" version of the Renaissance. You know how "Swiss Family Robinson" was kind of like teenager's wish-fulfillment version of "Robinson Crusoe"? That's how I felt about Quicksilver... too many protagonists had too many convenient 20th century attitudes and too much 20th century understanding of biology and physics.
If we can convince the public opinion that MS doesn't implement OOXML, it will need to.
.DOC and .XLS format. And government bodies would just check "supports ODF" and still send out proposals and accept tenders in .DOC and .XLS. But at least they'd mostly also accept ODF, whether they cared about it or not. But most of them won't... again, if they did, they'd already be using software that supported open standards.
.XLS format. The reports don't contain any more information than a simple .CSV would. People have asked them to provide .CSV as an option, and they haven't. The company's servers are all running Linux and they're active users of open source software. If people like that don't care enough to use open formats, why the heck do you think the public opinion will make a difference?
No, if we can convince the public opinion that MS doesn't implement OOXML and that it matters, it will need to. But if the public in general actually cared about stuff like that they wouldn't be using Office file formats for anything but working copies of documents anyway.
Look, if Microsoft implemented ODF and did a great job of it, and it was included as an option in Office right now people would still overwhelming be passing documents around in
Hell, there's a company I do business with that provides reports in
Each core needs fast acces to the complete scene.
Well, yes, that's just the flip side of the fact that you don't need to do occlusion culling, which you might notice I mentioned. that is to say, I did not suggest splitting up the scene, but rather replicating it.
What I suggested was giving the pipelines local memory and caching the static part of the mesh. Since that's almost all of the mesh, almost all of the time, they will only need to fetch dynamically changing meshes. With some analog for vertex buffer objects you could even cache subelements like vehicles and just fetch location changes.
I thought it was going to be about Google being evil, not about Google selling computers to the spooks.
The implication in the term being that there's no need for communication between each core or thread or process: they each just get handed a rectangular portion of the offscreen screen memory, and they do their job alone, and when they're all done, then the screen can be flipped to show the results.
Well, yes, you could do that. You could also render your scenes by having the pipelines re-parse shader programs from text for every frame. But you don't.
If access to the mesh is a bottleneck, give the pipelines local cache and cache static portions of the mesh. If access to textures is a bottleneck, cache them. You don't need to update the mesh when the camera moves, or when the lights move, because raytracers doesn't need (or want) to precompute occlusion.
Could not connect to user database: Can't connect to local MySQL server through socket '/var/lib/mysql/mysql.sock' (11)