The point is that games like Q3A and UT make great CPU/bandwidth benchmarks at low res. look at Tom's Hardware. He uses 640x480 as a CPU benchmark, and then ignores anything below at least 800x600 or higher when looking at graphics. He has a great article a while back (I don't remember the link. sorry) explaining exactly that. Also, if the card/CPU manufacturers give more speed, then that lets the game developers add better physics/AI and more complex graphics. Anyway, no one had mentioned this yet, thought I might.
Open it up in ahex editor; I bet they didn't even encrypt it. change www.aol.com to www.slashdot.com or equiv. Keep trying each instance till it works. My experience with hex editing / dissasembling is limited, but I'll bet you need to keep the number of characters the same. Anyway, worht a try for those "advanced" users.
As far as I can see, the article is on "our" side. The treaty supports security through obscurity, not the article. And I have this sneaking suspicion you are really pissed off by that which the article is describing.
But there's a problem here, isn't there? That argument works no matter where you are in your population curve. It also has one underlying neglected assumption: one urn has ten, one has 1 million. We can't make that asumption, and the model is therefore inherently flawed. If we knew that there were only two possible outcomes (end after 20 Billion people have lived, and end after 20 Trillion people have lived), I would agree with you. But, in reality we don't know that. Not only that, but the model also assumes equal probability. We don't know the probability distribution across all possible outcomes, which is another assumtion required by Bayes' Theorem. Without that distribution known (ie the 50/50 initial probability for the urns), we can't produce a final answer. So, his argument is inherently problematic.
BTW, my qualifications consist of a relatively brief study of Bayes' Theorem in a stat course. If someone has more knowledge on this and would care to offer an explanation, please do. But "two researchers offer a solution and don't explain why my argument is invalid" doesn't cut it. It's the same (almost) argument that our teacher used when explaining when you could/couldn't apply the theorem, so I am pretty sure it's correct.
Are we (they?0 missing something?
on
Watch Camera
·
· Score: 2
Random passerby: excuse me, what time is it?
Wrist Camera wearer: It's... oh wait, I left my watch at home. sorry.
R.P.: what's that on your wrist?
W.C.W.: oh, that's a camera. See, it takes pictures! -click- See, that's you!
R.P.: pervert.
am I missing something, or does it not tell time? I certainly didn't see anything, much less buttons to change time, etc. It would seem a simple enough feature to add.
Some clarification of the original post:
(also a general response)
The L1 and L2 caches are so much higher bandwidth because they run at core speed. Main memory runs at FSB speed (133MHz doubled -- its a DDR board). The board is NOT SMP, so I still don't see the point. I think syncing the cache with the FSB in terms of clock and width would provide the lowest latency, and extra bandwidth doesn't matter anyway. Granted, DMA devices can use SOME bandwidth, but not much. I think even 3GB/s or so would be PLENTY.
This isn't about that at all. This is Micron's new chipset, Mamba. It is the successor to Samaurai, their DDR reference chipset. When they built Samaurai, they found they had not used 40% of the die space, so they added an 8MB DRAM cache to it. The cache is 50% lower latency, with a 9.6GB/s bandwidth; it is completely different from the 760MP buffer, which is strictly a BUFFER not a cache, and only allows some reordering to improve performance.
The FSB can't sustain it anyway... so why 9.6GB? I can understand the lower latencies, those will be useful, but that bandwidth??? even FSB and IO together can't.
Go take a look at the XPP page. It sits on a PCI card, it runs linux (it really does! there's both the weird XPP bit and a 32bit RISC core programmed with gcc), and it delivers amazing processing power. It sounds awfully familiar to the SETI accelerator PCI card hoax! Except this seems to be for real. So, can I get it to do SETI? I think I recall from the last set of comments that they don't treat ports as legit; that's no fun, cuz the thing is intended for DSP/image analysis, which is all SETI is anyway. I think I can get it to distributed.net though; after all, it's also suggested for crypto stuff. I mean heck, it can sit there, talk to the net every so often to get a new key, and with 256MB of RAM and 51.2 GFLOPs available I think it will do just fine at moving me up the ranks in distributed.net, or the great internet mersenne prime search. Too bad its expensive... and no software yet (plz? I don't think this thing should die too soon). Does anyone have pricing info?
OK slashdot users, it's time to try out a whole bunch of things at once. Someone go set up a site, collect money for the application (we need a lawyer to write it up, call it $5000-$10000; it isn't that long or complicated, and I don't know what going rates are, but I'll bet that covers it). Then, get people to pledge money via credit card (Paypal perhaps? don't have to deal with merchant side crap; maybe someone with a merchant account would). Then, if 10 grand gets pledged, hire the lawyer, submit the app. We get to see:
Do geeks believe in this strongly enough to pledge money?
Does the honor system work as a payment method for a one-time service performed for the community at large (like writing a song or book or whatever)?
And lastly, is the patent system really as COMPLETELY fucked as we seem to think it is?
As a side note, I think I would pay money. Heck, get copyleft (OK not them, but someone similar) to print a T-shirt with the patent on it; sell shirts to get money to submit the patent.
A method for converting binary data into a smaller quantity of binary data which conveys the same meaning when put to its original purpose. The process would be carried out initially by a piece of computer software ("compressor") which would apply an algorithm to the original binary data. This algorithm would take advantage of patterns in the data, either by searching for them at the time of the running of the compressor or patterns assumed to be present because of the type of data being processed by the algorithm ("compressed"). Depending upon the situation, and the neccessity for the original data to be preserved exactly as opposed to only in meaning, the algorithm applied could be either reversible completely ("lossless") or could produce a different, but similar, set of original data when reversed ("lossy"). For example, a lossy compression algorithm applied to a picture would, when reversed ("decompressed"), produce a picture similar to the original; that is, to the human eye, it would appear to be the same picture, although potenntially of a lower quality. A lossless algorithm would reproduce the original exactly, giving a higher quality output. This technique could be used for many purposes. These include, but are not limited to, compression of images and sounds associated with Web pages; compression of videos to facilitate their transfer across the Internet, and compression of data on a storage media for purposes of allowing more data to be stored.
It's not yet up to NBC quality...
on
Nobel Prizes
·
· Score: 2
Until it can survive being slashdotted, that is. I have yet to be able to get through. There's something to be said for accessibility, even if the coverage is better (biased in a way I prefer). Whatever problems I may have with NBC, I have a strong suspicion that their site is up and doing just fine (I'll admit I haven't checked).
BTW, any mirrors up yet?
So what was it for, anyway? I gather from the comments it was prozac, but that could be completely wrong. Anyone have something more specific?
A firend of mine built a Spud gun one. Combustion chamber was a 4" by 12" cylinder, with a 4' by 1" barrel on it. Ignition was by means of a gas grill igniter (piezoelectric spark). He usually used hair spray, sometimes propane as the fuel. I don't know what the range was on potatoes, but he did get it used in the school play (Peter Pan) with small bits of styrofoam as the ammo. It easily crossed the stage with styrofoam, which has approximately zero mass (no momentum...) and a lot of air friction. The carrel was interchangeable; When he instead connected up a.68" barrel, and used paintballs, it got truly destructive. With a paintball (which is supposed to break open and just leave paint, right?) he could take ark off pine trees and seriously scar telephone poles (no structural damage, I don't think...) He claimed 600FPS muzzle velocity on the paintballs; I don't know how true that was, without a good electronic measuring device. Damn cool tho.
As best I can tell, we already do. One ATC operator could send a plane onto a collision course with another, and then there would be nothing to tell the planes not to. We implicitly trust the towers. If we do this, we just shift to a system that has DIFFERENT, but probably neither inherently more nor less, problems. The problems change from many potentially dishonest operators to very few operators, but many potential crackers. Navigation has had trusted client problems forever (lighthouses, whatever the land-based radio location system was, could all be messed with). I don't know of any recorded case of a problem. Admitedly, we have many more people who would want to mess things up, but I don't think that this is a new problem.
---
Re:How it can be made to work...
on
Online Voting?
·
· Score: 3
Depends on the number of voters/district:)
Basically, I wrote a (poorly optimized) Java program that did crypto. It couldn't do this, but I think I could have made it do it in about 30s/vote; Now, that is Sun Java 1.2 compiler and VM, which isn't fast, so I would assume about 5 s/vote on my PII-350. A few thousand (hundred thousand?) votes would probably need some sort of clustered thing. Now, if you put in special hardware to do the RSA, like say a 256-bit integer multipier chip, it gets much faster. The chip could be very deeply pipelined and run very fast, because there are lots of independent multiplications to do. But then, that's expensive too. So, I think it owuld be somewhat expensive, but not TOO much more so than what's currently used.
---
How it can be made to work...
on
Online Voting?
·
· Score: 4
OK, in a brief crypto class I took we produced an online voting algorithm. Here's the way it works:
Through normal channels, each person gets their voter registration card. This card has a unique number on it, very long to reduce chance of being able to guess one correctly. Person goes to computer, sets up key pair, connects to central server, and encrypts and signs their number + ID info (name, age, etc). This proves that key pair belongs to person. Then comes the actual voting protocol. Let's assume for simplicity only one thing to vote on (Pat v. whoever).
I create 10 votes for pat and 10 votes for Ross (I am pretty sure it's not him but I don't remember the real guy's name; I'll use Ross). Each vote consists of a GUID, and who the vote is for. The GUID is a long number (128bit; longer if neede to prevent collisions) that is randomly generated. However, each Ross vote is paired with a Pat vote (same GUID). I then Blind the votes with a blinding function -- need the blinding factor to get the vote out. This is actually just multiplication by a large number; DES or equiv doesn't work here. I then sign each vote and encrypt to central office.
Central office gets all the votes, picks one to make valid, and asks for the blinding factors for the rest. It then decrypts, verifies, and unblinds these. It then checks that each is a valid vote and that the GUIDs come in pairs. For the last vote it decrypts, verifies, signs with a DIFFERENT KEY PAIR, encrypts to voter, and returns the vote (both of them). The voter then decrypts the vote, unblinds it, picks one, and sends it in to vote. This last step (submission) is not connected to the others; I could put my vote on a disk and take it to the library to vote if I'm paranoid.
So, this meets all the needed requirements:
One person, one vote: registration + GUID (can't submit vote more than once; central office won't sign more than one.
Anonymous: when I send in my vote, it no longer has my key connected to it, and the central office doesn't know the GUID.
can't be faked: partly in the registration, partly in the crypto.
Third party can't see: its encrypted
Third party can't change: same
I can cverify that I voted and who I voted for: I can send a request "who did this GUID vote for" to office, and it can tell me. If I'm paranoid, I worry about the central office tracking IPs and such, so I don't ask or ask from a library etc.
How about just block any site that attempts to bring up more than 8,000 little windows, and any site any of those link to? I think that ONLY gets porn sites. Admittedly, not quite all, but it seems easy enough. Then, you can modify the software for your home use, so that you can look at porn without the annoying windows:)
A simple modification: Multiply the score (if negative) by the number of JPEGs over 50k. Or something similar. Banners and other "Business Graphics" are usually GIFs or small. Porn, however, is much more often large JPEGs. So a site that only uses a little bad language would get a much higher rating if there were lots of large photos.
I think eBay should act as respectable corporate US citizen. That doesn't mean they will of course. What I mean by this is that they should try to stay in the spirit of the US law, which forbids selling marijuana. The only difference between that and body parts is that most people have a much greater objection to body parts (myself included). Also, most (not all, I believe) countries make sales of body parts illegal, including the US. eBay could say that such auctions are illegal somewhere and so they will be left up, but I think that is irresponsible and that many people would object. So the French Yahoo! (and perhaps the english one too?) should similarly be responsible and act in accordance with the intent of such laws, whether or not hosting the auction is illegal. They should simply pull the auction and be done with it. In general, I agree with your senitments.
Because you are (presumably) a relatively small shop, perhaps you can do better than OTS censorware. I don't know if any censorware products would allow this, but here's some thoughts.
Buy a censorware product, and install it. Have a policy that says something akin to "no porn, indecent material, or illegal activities (no DoS, not no browsing drug sites)" or some similar AUP. Then , implement censorware to block all such sites, using an OTS product. But, have it set up so that there is an easy way (web form?) to submit a link to a site that should be allowed. Periodically got through the submissions manually and remove those sites (if appropriate). While something of a hassle, I think it is a reasonable solution. I imagine that only a small fraction of users each day would submit links, so having someone spend an hour at the end of each day might be sufficient. Or perhaps do that throughout the day along with other admin duties. Granted, it woudn't provide immediate response, but its far better than nothing. I think you have an obligation to provide reasonable censorship at your site. Good Luck!
they should not have to take down the marijuana for US legal reasons. I guess I feel that Ebay as a responsible corporate entity should take it down, as they do with auctions of body parts, etc. However, I don't believe that eba should be required to take it down. Now, if someone in the US bought it, it would be illegal to ship the marijuana to the US (for both parties involved? I don't know). I think that the French case should be handled similarly: The auction should stay, but it should be illegal to ship the item into France. I don't know exactly what the legal reasoning is, but I bet that its that it is illegal to sell such things in France. However, I doubt that the actual pages are illegal. The article on ZDNN that I read didn't explain it very well. Perhaps someone could fill in. BTW: IANAL.
they do. My point was just that we will go to needing better cooling at some point, with Rambus or SDRAM. We went from no cooling to very powerful cooling with chips, RAM will just take longer. I have of course nothing to back this up but my own analysis, but hey...
Yeah. The problem with many transistors on CPUs is power consumption. Put too many on and run them too fast and you can fry the chip. The 1GHz Tbird can do it in 8 seconds. That's what heat sinks are for. I know we haven't needed them on RAM yet, but I'm sure we will eventually, be it RIMMs or DIMMs, they will get fast enough and high enough transistor count to need it one day.
Slashdot Mirror
The point is that games like Q3A and UT make great CPU/bandwidth benchmarks at low res. look at Tom's Hardware. He uses 640x480 as a CPU benchmark, and then ignores anything below at least 800x600 or higher when looking at graphics. He has a great article a while back (I don't remember the link. sorry) explaining exactly that. Also, if the card/CPU manufacturers give more speed, then that lets the game developers add better physics/AI and more complex graphics. Anyway, no one had mentioned this yet, thought I might.
Open it up in ahex editor; I bet they didn't even encrypt it. change www.aol.com to www.slashdot.com or equiv. Keep trying each instance till it works. My experience with hex editing / dissasembling is limited, but I'll bet you need to keep the number of characters the same. Anyway, worht a try for those "advanced" users.
As far as I can see, the article is on "our" side. The treaty supports security through obscurity, not the article. And I have this sneaking suspicion you are really pissed off by that which the article is describing.
then only criminals will know about the exploits.
BTW, my qualifications consist of a relatively brief study of Bayes' Theorem in a stat course. If someone has more knowledge on this and would care to offer an explanation, please do. But "two researchers offer a solution and don't explain why my argument is invalid" doesn't cut it. It's the same (almost) argument that our teacher used when explaining when you could/couldn't apply the theorem, so I am pretty sure it's correct.
Wrist Camera wearer: It's... oh wait, I left my watch at home. sorry.
R.P.: what's that on your wrist?
W.C.W.: oh, that's a camera. See, it takes pictures! -click- See, that's you! R.P.: pervert.
am I missing something, or does it not tell time? I certainly didn't see anything, much less buttons to change time, etc. It would seem a simple enough feature to add.
Some clarification of the original post: (also a general response) The L1 and L2 caches are so much higher bandwidth because they run at core speed. Main memory runs at FSB speed (133MHz doubled -- its a DDR board). The board is NOT SMP, so I still don't see the point. I think syncing the cache with the FSB in terms of clock and width would provide the lowest latency, and extra bandwidth doesn't matter anyway. Granted, DMA devices can use SOME bandwidth, but not much. I think even 3GB/s or so would be PLENTY.
This isn't about that at all. This is Micron's new chipset, Mamba. It is the successor to Samaurai, their DDR reference chipset. When they built Samaurai, they found they had not used 40% of the die space, so they added an 8MB DRAM cache to it. The cache is 50% lower latency, with a 9.6GB/s bandwidth; it is completely different from the 760MP buffer, which is strictly a BUFFER not a cache, and only allows some reordering to improve performance.
The FSB can't sustain it anyway... so why 9.6GB? I can understand the lower latencies, those will be useful, but that bandwidth??? even FSB and IO together can't.
Go take a look at the XPP page. It sits on a PCI card, it runs linux (it really does! there's both the weird XPP bit and a 32bit RISC core programmed with gcc), and it delivers amazing processing power. It sounds awfully familiar to the SETI accelerator PCI card hoax! Except this seems to be for real. So, can I get it to do SETI? I think I recall from the last set of comments that they don't treat ports as legit; that's no fun, cuz the thing is intended for DSP/image analysis, which is all SETI is anyway. I think I can get it to distributed.net though; after all, it's also suggested for crypto stuff. I mean heck, it can sit there, talk to the net every so often to get a new key, and with 256MB of RAM and 51.2 GFLOPs available I think it will do just fine at moving me up the ranks in distributed.net, or the great internet mersenne prime search. Too bad its expensive... and no software yet (plz? I don't think this thing should die too soon). Does anyone have pricing info?
Do geeks believe in this strongly enough to pledge money?
Does the honor system work as a payment method for a one-time service performed for the community at large (like writing a song or book or whatever)?
And lastly, is the patent system really as COMPLETELY fucked as we seem to think it is?
As a side note, I think I would pay money. Heck, get copyleft (OK not them, but someone similar) to print a T-shirt with the patent on it; sell shirts to get money to submit the patent.
no no... he patented a specific algorithm; that is, one exact implementation of my new patent. And I had mine first.
A method for converting binary data into a smaller quantity of binary data which conveys the same meaning when put to its original purpose. The process would be carried out initially by a piece of computer software ("compressor") which would apply an algorithm to the original binary data. This algorithm would take advantage of patterns in the data, either by searching for them at the time of the running of the compressor or patterns assumed to be present because of the type of data being processed by the algorithm ("compressed"). Depending upon the situation, and the neccessity for the original data to be preserved exactly as opposed to only in meaning, the algorithm applied could be either reversible completely ("lossless") or could produce a different, but similar, set of original data when reversed ("lossy"). For example, a lossy compression algorithm applied to a picture would, when reversed ("decompressed"), produce a picture similar to the original; that is, to the human eye, it would appear to be the same picture, although potenntially of a lower quality. A lossless algorithm would reproduce the original exactly, giving a higher quality output. This technique could be used for many purposes. These include, but are not limited to, compression of images and sounds associated with Web pages; compression of videos to facilitate their transfer across the Internet, and compression of data on a storage media for purposes of allowing more data to be stored.
BTW, any mirrors up yet?
So what was it for, anyway? I gather from the comments it was prozac, but that could be completely wrong. Anyone have something more specific?
A firend of mine built a Spud gun one. Combustion chamber was a 4" by 12" cylinder, with a 4' by 1" barrel on it. Ignition was by means of a gas grill igniter (piezoelectric spark). He usually used hair spray, sometimes propane as the fuel. I don't know what the range was on potatoes, but he did get it used in the school play (Peter Pan) with small bits of styrofoam as the ammo. It easily crossed the stage with styrofoam, which has approximately zero mass (no momentum...) and a lot of air friction. The carrel was interchangeable; When he instead connected up a .68" barrel, and used paintballs, it got truly destructive. With a paintball (which is supposed to break open and just leave paint, right?) he could take ark off pine trees and seriously scar telephone poles (no structural damage, I don't think...) He claimed 600FPS muzzle velocity on the paintballs; I don't know how true that was, without a good electronic measuring device. Damn cool tho.
As best I can tell, we already do. One ATC operator could send a plane onto a collision course with another, and then there would be nothing to tell the planes not to. We implicitly trust the towers. If we do this, we just shift to a system that has DIFFERENT, but probably neither inherently more nor less, problems. The problems change from many potentially dishonest operators to very few operators, but many potential crackers. Navigation has had trusted client problems forever (lighthouses, whatever the land-based radio location system was, could all be messed with). I don't know of any recorded case of a problem. Admitedly, we have many more people who would want to mess things up, but I don't think that this is a new problem.
---
Basically, I wrote a (poorly optimized) Java program that did crypto. It couldn't do this, but I think I could have made it do it in about 30s/vote; Now, that is Sun Java 1.2 compiler and VM, which isn't fast, so I would assume about 5 s/vote on my PII-350. A few thousand (hundred thousand?) votes would probably need some sort of clustered thing. Now, if you put in special hardware to do the RSA, like say a 256-bit integer multipier chip, it gets much faster. The chip could be very deeply pipelined and run very fast, because there are lots of independent multiplications to do. But then, that's expensive too. So, I think it owuld be somewhat expensive, but not TOO much more so than what's currently used.
---
Through normal channels, each person gets their voter registration card. This card has a unique number on it, very long to reduce chance of being able to guess one correctly. Person goes to computer, sets up key pair, connects to central server, and encrypts and signs their number + ID info (name, age, etc). This proves that key pair belongs to person. Then comes the actual voting protocol. Let's assume for simplicity only one thing to vote on (Pat v. whoever).
I create 10 votes for pat and 10 votes for Ross (I am pretty sure it's not him but I don't remember the real guy's name; I'll use Ross). Each vote consists of a GUID, and who the vote is for. The GUID is a long number (128bit; longer if neede to prevent collisions) that is randomly generated. However, each Ross vote is paired with a Pat vote (same GUID). I then Blind the votes with a blinding function -- need the blinding factor to get the vote out. This is actually just multiplication by a large number; DES or equiv doesn't work here. I then sign each vote and encrypt to central office.
Central office gets all the votes, picks one to make valid, and asks for the blinding factors for the rest. It then decrypts, verifies, and unblinds these. It then checks that each is a valid vote and that the GUIDs come in pairs. For the last vote it decrypts, verifies, signs with a DIFFERENT KEY PAIR, encrypts to voter, and returns the vote (both of them). The voter then decrypts the vote, unblinds it, picks one, and sends it in to vote. This last step (submission) is not connected to the others; I could put my vote on a disk and take it to the library to vote if I'm paranoid.
So, this meets all the needed requirements:
One person, one vote: registration + GUID (can't submit vote more than once; central office won't sign more than one.
Anonymous: when I send in my vote, it no longer has my key connected to it, and the central office doesn't know the GUID.
can't be faked: partly in the registration, partly in the crypto.
Third party can't see: its encrypted
Third party can't change: same
I can cverify that I voted and who I voted for: I can send a request "who did this GUID vote for" to office, and it can tell me. If I'm paranoid, I worry about the central office tracking IPs and such, so I don't ask or ask from a library etc.
Did I miss anything?
---
How about just block any site that attempts to bring up more than 8,000 little windows, and any site any of those link to? I think that ONLY gets porn sites. Admittedly, not quite all, but it seems easy enough. Then, you can modify the software for your home use, so that you can look at porn without the annoying windows :)
---
A simple modification: Multiply the score (if negative) by the number of JPEGs over 50k. Or something similar. Banners and other "Business Graphics" are usually GIFs or small. Porn, however, is much more often large JPEGs. So a site that only uses a little bad language would get a much higher rating if there were lots of large photos.
---
I think eBay should act as respectable corporate US citizen. That doesn't mean they will of course. What I mean by this is that they should try to stay in the spirit of the US law, which forbids selling marijuana. The only difference between that and body parts is that most people have a much greater objection to body parts (myself included). Also, most (not all, I believe) countries make sales of body parts illegal, including the US. eBay could say that such auctions are illegal somewhere and so they will be left up, but I think that is irresponsible and that many people would object. So the French Yahoo! (and perhaps the english one too?) should similarly be responsible and act in accordance with the intent of such laws, whether or not hosting the auction is illegal. They should simply pull the auction and be done with it. In general, I agree with your senitments.
---
Buy a censorware product, and install it. Have a policy that says something akin to "no porn, indecent material, or illegal activities (no DoS, not no browsing drug sites)" or some similar AUP. Then , implement censorware to block all such sites, using an OTS product. But, have it set up so that there is an easy way (web form?) to submit a link to a site that should be allowed. Periodically got through the submissions manually and remove those sites (if appropriate). While something of a hassle, I think it is a reasonable solution. I imagine that only a small fraction of users each day would submit links, so having someone spend an hour at the end of each day might be sufficient. Or perhaps do that throughout the day along with other admin duties. Granted, it woudn't provide immediate response, but its far better than nothing. I think you have an obligation to provide reasonable censorship at your site. Good Luck!
---
they should not have to take down the marijuana for US legal reasons. I guess I feel that Ebay as a responsible corporate entity should take it down, as they do with auctions of body parts, etc. However, I don't believe that eba should be required to take it down. Now, if someone in the US bought it, it would be illegal to ship the marijuana to the US (for both parties involved? I don't know). I think that the French case should be handled similarly: The auction should stay, but it should be illegal to ship the item into France. I don't know exactly what the legal reasoning is, but I bet that its that it is illegal to sell such things in France. However, I doubt that the actual pages are illegal. The article on ZDNN that I read didn't explain it very well. Perhaps someone could fill in. BTW: IANAL.
---
they do. My point was just that we will go to needing better cooling at some point, with Rambus or SDRAM. We went from no cooling to very powerful cooling with chips, RAM will just take longer. I have of course nothing to back this up but my own analysis, but hey...
---
Yeah. The problem with many transistors on CPUs is power consumption. Put too many on and run them too fast and you can fry the chip. The 1GHz Tbird can do it in 8 seconds. That's what heat sinks are for. I know we haven't needed them on RAM yet, but I'm sure we will eventually, be it RIMMs or DIMMs, they will get fast enough and high enough transistor count to need it one day.
---