No, this is different, here you have a dominant graphics card company threatening to cut off the air supply of a web site because they advertise or even mention a competing product.
If Coke had threatened your station to stop advertising Pepsi or to not run a movie because it had Pepsi product placement, Coke may have have had a law suit on their hands. This kind of crap is anticompetitive.
There are plenty of links on this, not just the Hard OCP article. There are at least five sites making the same kinds of claims, some of them have evidence and threaten to post even more. Also I don't see any denials from nVidia, it appears that they promised a policy statement on this kind of abuse and didn't deliver in the hopes that the story would grow cold. The apology Kyle linked to was not for making invalid claims but for getting folks at nVidia into trouble, it never said anything about those claims being false and infact Kyle is still obviously displeased with nVidia's subsequent actions. If you actually read the original contents of that link it's pretty damning.
There's just too much evidence here indicating a pattern of abuse. I like nVidia's products but I want to see them win on merit, not through uncompetitive practices, there's already been too much of this in the computer industry. If they were more dominant this kind of thing would be illegal, infact it may be anyway. They are clearly asking that some sites remove advertising collateral (logos & reviews) from a competitor (3Dfx) in exchange for free review hardware. Is this legal?
If you read some of the ham fisted crap that the nVidia PR group has pulled (in the more detailed online accounts) you end up questioning their sanity never mind their tactics. It's a litany of how to piss independent web masters off. Those are the very people who are going to be reviewing the product, that's not smart.
Now it's really hitting the fan, and rightly so. PR are paid to prevent this kind of thing not CAUSE it. What a bunch of maroons.
This has nothing to do with the game client. It's in the graphics driver.
There's a lot of evidence to suggest that when open source clients get out there the hacking runs out of control, beyond a few variable tweaks it's hard to modify a client without the code but with the code you can make wholesale changes and this happened with quake.
It takes much more work to make an open source client difficult to hack, and bot's written into these clients make them deadly, that is almost entirely to do with them being open source.
Technical solutions to this are tricky. It's below the level of even the game client in this case. With this cheat you could make it less useful by the game client doing extra work and visibility culling player characters. For hacked clients not even transmitting player positions when very non visible is problematic because in addition to the extra work you kind of need a server to arbitrate. The need to fix this undermines some of the advantage of having a fast hardware zbuffer.
It's interesting to note that this doesn't even require a special driver. A technically savvy developer could implement a graphics library wrapper which intercepted graphics calls to implement the equivalent functionality of these driver modifications. This kind of utility is already available and when done correctly only has a small performance impact on lot's of modern hardware.
The transparency doesn't need to be depth buffered at all.
Just enable blending with say 33% alpha and turn off the zbuffer and you're in business. You don't depth sort. You have your choice of blends. It won't really matter too much what order the tris draw in and will make NO difference for some blend functions.
NO, it looks like the XBox is independent enough to escape the ravages of conflicting corporate interests which would drag it down.
It will support OpenGL, that has already been announced. Why the heck are you commenting on this when you don't even no the most elementary and well publicised facts about the XBox? I know folks here have various axes to grind over the API issues but get the FACTS right. XBox will support OpenGL period. How many times does it have to be said before it sinks in?
Why the heck is a post with blatant factual inaccuracies moderated up?
nVidia and Microsoft have stated that nVidia will support OpenGL on the XBox, INCLUDING support for all the advanced graphics features. DirectX is NOT the equivalent of OpenGL. D3D is the 3D component that competes with OpenGL. You can write code which calls DirectX AND OpenGL and Carmack has, he DOESN'T use the D3D part of DirectX.
DirectX will NEVER exist on Linux, Microsoft's whole strategy is to own the API's and deploy them exclusively on Win32 to lock the applications to that platform.
Carmack is NOT "in team with M$", at most he's advising them. He actually advised them on D3D a long time ago, OK, the advice was "throw it away", but he had lengthy meetings with M$ at the height of the D3D vs OpenGL wars years ago. Guess what his advice on XBox will be today? I'd guess he'll be advising them to support all the great features using OpenGL, including advice on OpenGL extensions. That isn't selling out, it's consistent with what his position has been for years. Infact Carmack has said from the outset that if D3D were better he'd use it but way back in his famous.plan he said that the pain and time taken for D3D to gain parity with OpenGL wouldn't be worth it. So, even if he were to start using D3D (and he isn't AFAIK) he still wouldn't be selling out, he owes no allegiance to an API, API's are tools not religious icons, the only reason he disliked D3D was it's technical flaws and obfuscation. The cross platform non proprietary nature of OpenGL is a more recently recognized bonus for the games industry (but OpenGL's primary objective initially) and one I expect everyone except Microsoft loves.
This is not the death of OpenGL. In fact with D3D copying so much of OpenGL it would almost be a semantic argument if D3D weren't such a darned mess. Even with your nightmare scenario (which isn't going to transpire) porting D3D games to OpenGL should actually be pretty easy and Loki have already demonstrated this.
The dire warnings seem overstated considering what is already accepted practice. They just pull the suspects emails in question prior to searching. Omnivore sounds like it was open to abuse and if that was deployed it should never have been, it's like wire tapping a small town to get evidence on one individual. Carnivore sounds like a right minded attempt to restrict scanning to the suspects account.
So what's new? They still need a court order and they could always tap the suspects phone any time as things stand. This just let's them tap an account than might be moving on a dial in from different locations. The whole system has always been build on trust and controlled by the fact that any abuse of the system won't pass muster as evidence in court anyway.
So, if a Judge let them deploy Omnivore it sounds like there's a need for some legislation to prevent this sort of dragnet approach in future but the Carnivore system is exactly the kind of thing I'd expect the FBI to be getting up to, why is everyone so surprised? The intention of developing Carnivore as a discriminating filter seems to be a move in the right direction IF it only traps and searches the email of the suspect, and that's the whole point of the newer system.
The whole problem is that they don't see the true intrinsic value of manned flight. Instead of lofty goals we get expensive boondoggles in the shuttle which would be better spent on unmanned missions or a manned mission with a worthwhile goal. Shouldn't the Sojourner landing have told NASA something? Right now we're building the most expensive tin can in history to orbit the Earth (IF the Russians don't blow up a critical part on the pad). It's a joke to suggest that this has proportionate scientific value.
When China start manned missions and India put a man on the moon it will really show the world what incredible under accomplishment NASA has been guilty of these last few decades. Hopefully it will prompt some action.
The Challenger disaster and the accompanying Disneyland philosophy that no risk is acceptable for manned flight is equally foolish. Tourists die climbing Everest each year, thousands die in auto accidents, hundreds die in plane crashes fulfilling the most mundane objectives. Historically huge risks have been undertaken by celebrated pioneers of new frontiers. Unfortunately that isn't acceptable for NASA management, they can easily find the few brave souls to shoulder the burden, but instead they are mired in ineffective safety procedures because they have forgotten what it is like to excel. NASA has given up on the dream. Here's a plan to cut costs AND save lives; half the budget for manned space, cut back on the excessive safety and associated bureaucracy and spend the savings on road safety awareness in high schools or on medical equipment at a few hundred hospitals. I only make this point to illustrate how completely imbalanced the pioneering spirit has become at NASA, and it isn't for the want of brave astronauts, it's done for the political careers of their masters. The upshot is that the very people who would volunteer for worthy manned missions get to orbit the Earth in (safe) tin cans instead of doing some real exploration. I think I know what choice they would make given the option.
Well like you say you'd be bus limited. You still get way more bandwidth over AGP4x with fast writes. You couldn't compete with hardware (on the card) t&l. I expect some stuff would be faster, you could reject back faces and cull to frustum per primitive so if you had a lame app with display lists you might win, but you'd have to store the display list geometry on the PCI card, you would just get hosed trying to send it to the cards over PCI then get it back transformed.
You wouldn't need a driver per processor but all told it's only uninteresting for the stated problem because of the bandwidth requirement. This is exactly the kind of problem this card is there to solve provided you have a greater compute to i/o bandwidth ratio.
This is cool, for the right class of problem. I doubt you'd see this in a mainstream server, remember this will be running specialized cross compiled code which just reads & writes to system memory. There would be no real system level interface except back on the native x86 processor once it got the results. Anything running on it would be specially ported although it looks like it can support interface cards of it's own accord. It sounds like the ultimate seti@home processing system once someone ports the client:-). One thing that isn't clear, it runs ELF binaries but I assume you'd have to upload the executable to it's memory first it wouldn't see the native file system, is this correct? Would it be able to read & write to a memory mapped file? One other issue, memory endianess would be reversed with this processor vs x86 right? So your memory communications would have to twiddle all the bytes based on data type of you wanted to do anything with the results on an x86 or other linux native application right?
Does anyone here have any experience using this system?
Rambus have filed hundreds of patents last year. If you look at the rate they are filing (it's accelerating) and given that they make NO product themselves they are in a very strong position. They will never have to pay anyone royalties because they have no need to license others ideas, just invent incrementally on existing ideas patented or not. Anyone hoping to challenge on a few patents is sure to run foul of their portfolio in some other area. This is a really powerful play by Rambus. I don't think people on slashdot really understand the magnitude of what's going on here or the strength of the Rambus position. Rambus is likely to win big time, they have a slew of patents in the works which nobody even knows about yet and a strategy to license them aggressively. I doubt they really give much on a damn about which memory technology wins. Patents are almost always written to be general enough to cover many implementations, hence the SDRAM claims, and Rambus has raced ahead of everyone in terms of focusing on one area of herdware design and filing as many ideas as they can think of. A patent dispute can start with one claim and then snowball afterwards as the stakes increase and more ideas are disputed, the normal defense is to counter sue and then horse trade IP. Unfortunately for the rest of the world, Rambus use no IP from anyone else, they are just an ideas company. So Hitachi puts up this lame defense that Rambus should be forced to hand over their IP. The next Rambus move was probably to threaten them with breaching 50 other patents they happen to have at which point Hitachi caved. This last part is just idle speculation but probably close to the mark. Even if they won on the SDRAM issue they couldn't escape all the patents filed by Rambus.
Your prediction is self defeating. If they don't have the influence over the SDRAM vendors then they don't have the influence to force the RDRAM issue. The absolute best they could manage would be to try and pressure Hitachi & Toshiba to agree to higher volume RDRAM shipments, and that will only serve to reduce the price of RDRAM, not exactly a nightmare scenario. If they do have the influence with other vendors then they don't need to wield it because DDR will be more lucrative for the anyway.
Even if we use your figure of 10% to Intel (I thought it was 20%) that's still HUGE, at current market cap 10% that's just shy of $1.1B (that's B as in billion). You have to remember that in terms of the stock actually controlled by Rambus and their overall assets this is an even more significant number. In addition it would look pretty devastating on their books when they actually have to write it down. When they originally cut the deal they obviously thought they had nothing to lose, they'd have given their right stone for a majority market share but as they near the end of the Faustian pact things will be looking very different to them.
Latencies are only larger for memory on the same channel, and the simpler layouts are not unimportant if you want to try and suport additional channels. The 2 sticks are a non issue, the sticks would obviously be smaller. You're also talking about DDR memory which is faster than RDRAM, it's the comparrisons with 133 SDRAM which have been unfair. You'll have a point when DDR mobos are on the shelves but then Rambus might be pleased about that.
On the issue of Rambus' memory IP, if the advantages were unimportant, the memory manufacturers would eschew them and escape the royalties. Your rhetoric on the standards committee is lifted straight from the Hitachi countersuit against Rambus which has now been dropped like a stone. In fact you didn't get it quite right (maybe you got Dr Tom's contorted version of the facts), because one of the things that Hitachi also claimed was that Rambus should have disclosed their existing patents when they joined the group. This would have given them forewarning to demand free use or perhaps to steer around the technology. Any way you look at it this was a desperate ploy by Hitachi and had a snowballs chance in hell of working in court. You could claim they dropped the suit to help sell the business but the truth is that Hitachi were crapping themselves at the prospect of losing their shirt on this. The memory business is booming right now (they can't make enough of the stuff and prices are rocketing) and they can't have been in any hurry to sell, that's just more smoke.
So which of us actually knows what they're talking about? I think I have my facts straight.
This is one ambiguous Rambus response to a question. I read the quote I don't need to go back, for someone so cynical about Rambus you're awful quick to take YOUR interpretation of one ambiguous statement as gospel truth. The objective is clearly to maximize the hype. Unlike you I've also read other Rambus pronouncements over the last few days. The simple fact is that Rambus have already stated in other communications that they will charge HIGHER royalties for DDR SDRAM than RDRAM. So you explain to me why they will now want RDRAM to win. Another fact you are ignoring is that if RDRAM wins in terms of Intel mobo share then Rambus has to hand over a huge chunk of their company to Intel, so factor that into your reasoning and again explain your conclusion. Don't jump on the rabid monopoly theory based on one remark, think for a second and tell me why on Earth they'd still want RDRAM to win.
As I've already stated in other posts, if all they can do is influence Hitachi and Toshiba, that isn't enough to affect pricing or volumes significantly. Just look at the price of SDRAM at source recently. It's lept by a huge ammount, it may hit 100% price increase that hasn't translated to anything near the kind of a price hike predicted at the hands of Rambus royalties. The simple fact is this whole theory is baseless. Even if Rambus charged huge royalties it still wouldn't make SDRAM significantly less competitive against RDRAM. It isn't driven by the royalties they charge, it's driven by the volumes it's made in.
That must be why the Times incompetently attempted to cover the names, and why even Jim Risen, the great sleuth who couldn't sleuth out how to write pdf files properly now thinks lives are being endangered. It isn't just about the spies, it's their extended families. In anycase you don't know who is left alive.
Maybe you can testify at their trial, they have religious courts you know. No need for judges when you have a cleric to run the proceeding you see. All the CIA et.al. needed to do was NOT release the names. There was a zero activity required to keep those few lives safe.
There's nothing Hollywood about throwing Jim Risen to the DOJ in an attempt to find his source. No cover needs to be blown. It would be worth testing to see if freedom of the press extends to leaking secret CIA documents and endangering lives in an attempt to promote your web site / newspaper.
As for getting his disk, I doubt getting a job as a janitor in his office building represents a serious risk, then again his system is probably on the network and that wouldn't exactly represent a risk of exposure if hacked from Tehran.
Now if I can think of it in the time it takes to write a post how long do you think it would take counter intelligence to think of it? Ofcourse those Iranians live in caves and don't have computers, silly me.
Your trouble is you are too busy being cynical about CIA doctrine to expect anything useful to be done about this. So who's given up on the USA, you or me?
Subvert technical excellence? Get a clue. They clearly invented some of the SDRAM & DDR technology before anyone else, otherwise Toshiba and Hitachi wouldn't have agreed to hand over many millions in royalties.
Rambus are investing a fortune in R&D. They want their technology to be adopted. They now have a foot in BOTH camps so actually care LESS now about which technology wins. RDRAM actually might have some merit with enough channels and in terms of simpler layout despite all the FUD. The price depends entirely on how many manufacturers start pumping it out.
If Intel are able to ship enough chipsets supporting RDRAM then Rambus must hand over 20% of their company. Since they will get more cash from their SDRAM & DDR patents then I expect even Rambus wants RDRAM to fade away.
Aside from anything else, it's not the license fees which drive up the price, it's the volumes of the product shipped. There's not much Rambus can do on the royalties front to change that.
Even if the doomsday scenario were plausable, if Rambus manage to stitch up all the manufacturers with license agreements then they win with SDRAM & DDR, if they don't then they can't force the RDRAM issue because some manufacturers would be royalty free.
Damn it the more I read the worse this gets. Jim Risen is the guy who's caused this problem now he's accusing the people who exposed his treachery of endangering the very lives his incompetence has thrown to the wolves.
I say arrest him and show him the sharp end of a 10 year stretch unless he comes up with the name of his CIA mole. Even if he hadn't screwed up like this, he'd still have advertised to all & sundry that he had the original file somewhere on his systems. How difficult would it be for a half competent agent to wander in and take his whole file system?
Well the bible may say don't hold the children responsible for the sins of the fathers, but that doesn't really cut it in Iran.
Ignoring the fact that you seem to think it's OK to abandon former allies once they hit 70 years old, what about their families? There is also a more significant contemporary credibility problem here. The CIA can't afford to look this foolish to people they need work with TODAY.
This is a pretty significant lapse, then again it's the CIA, you gotta ask yourself if it's entirely accidental too. Maybe this is a CIA plot to deliberately discredit the people named in the document. What better way to 'accidentally' leak a few names.
Slashdot Mirror
No, this is different, here you have a dominant graphics card company threatening to cut off the air supply of a web site because they advertise or even mention a competing product.
If Coke had threatened your station to stop advertising Pepsi or to not run a movie because it had Pepsi product placement, Coke may have have had a law suit on their hands. This kind of crap is anticompetitive.
This is not the same thing.
There are plenty of links on this, not just the Hard OCP article. There are at least five sites making the same kinds of claims, some of them have evidence and threaten to post even more. Also I don't see any denials from nVidia, it appears that they promised a policy statement on this kind of abuse and didn't deliver in the hopes that the story would grow cold. The apology Kyle linked to was not for making invalid claims but for getting folks at nVidia into trouble, it never said anything about those claims being false and infact Kyle is still obviously displeased with nVidia's subsequent actions. If you actually read the original contents of that link it's pretty damning.
There's just too much evidence here indicating a pattern of abuse. I like nVidia's products but I want to see them win on merit, not through uncompetitive practices, there's already been too much of this in the computer industry. If they were more dominant this kind of thing would be illegal, infact it may be anyway. They are clearly asking that some sites remove advertising collateral (logos & reviews) from a competitor (3Dfx) in exchange for free review hardware. Is this legal?
If you read some of the ham fisted crap that the nVidia PR group has pulled (in the more detailed online accounts) you end up questioning their sanity never mind their tactics. It's a litany of how to piss independent web masters off. Those are the very people who are going to be reviewing the product, that's not smart.
Now it's really hitting the fan, and rightly so. PR are paid to prevent this kind of thing not CAUSE it. What a bunch of maroons.
This has nothing to do with the game client.
It's in the graphics driver.
There's a lot of evidence to suggest that when open source clients get out there the hacking runs out of control, beyond a few variable tweaks it's hard to modify a client without the code but with the code you can make wholesale changes and this happened with quake.
It takes much more work to make an open source client difficult to hack, and bot's written into these clients make them deadly, that is almost entirely to do with them being open source.
Technical solutions to this are tricky. It's below the level of even the game client in this case. With this cheat you could make it less useful by the game client doing extra work and visibility culling player characters. For hacked clients not even transmitting player positions when very non visible is problematic because in addition to the extra work you kind of need a server to arbitrate. The need to fix this undermines some of the advantage of having a fast hardware zbuffer.
It's interesting to note that this doesn't even require a special driver. A technically savvy developer could implement a graphics library wrapper which intercepted graphics calls to implement the equivalent functionality of these driver modifications. This kind of utility is already available and when done correctly only has a small performance impact on lot's of modern hardware.
The transparency doesn't need to be depth buffered at all.
Just enable blending with say 33% alpha and turn off the zbuffer and you're in business. You don't depth sort. You have your choice of blends. It won't really matter too much what order the tris draw in and will make NO difference for some blend functions.
Apple is going to use them. ATI let the cat out of the bag. Apple hasn't mentioned this yet but it looks like this is going in the high end G4.
Oh jeeze...
NO, OpenML is a different thing. It's for streaming video and such, it does not replace OpenGL.
NO, it looks like the XBox is independent enough to escape the ravages of conflicting corporate interests which would drag it down.
It will support OpenGL, that has already been announced. Why the heck are you commenting on this when you don't even no the most elementary and well publicised facts about the XBox? I know folks here have various axes to grind over the API issues but get the FACTS right. XBox will support OpenGL period. How many times does it have to be said before it sinks in?
Why the heck is a post with blatant factual inaccuracies moderated up?
Oh dear,
.plan he said that the pain and time taken for D3D to gain parity with OpenGL wouldn't be worth it. So, even if he were to start using D3D (and he isn't AFAIK) he still wouldn't be selling out, he owes no allegiance to an API, API's are tools not religious icons, the only reason he disliked D3D was it's technical flaws and obfuscation. The cross platform non proprietary nature of OpenGL is a more recently recognized bonus for the games industry (but OpenGL's primary objective initially) and one I expect everyone except Microsoft loves.
you are misguided.
nVidia and Microsoft have stated that nVidia will support OpenGL on the XBox, INCLUDING support for all the advanced graphics features. DirectX is NOT the equivalent of OpenGL. D3D is the 3D component that competes with OpenGL. You can write code which calls DirectX AND OpenGL and Carmack has, he DOESN'T use the D3D part of DirectX.
DirectX will NEVER exist on Linux, Microsoft's whole strategy is to own the API's and deploy them exclusively on Win32 to lock the applications to that platform.
Carmack is NOT "in team with M$", at most he's advising them. He actually advised them on D3D a long time ago, OK, the advice was "throw it away", but he had lengthy meetings with M$ at the height of the D3D vs OpenGL wars years ago. Guess what his advice on XBox will be today? I'd guess he'll be advising them to support all the great features using OpenGL, including advice on OpenGL extensions. That isn't selling out, it's consistent with what his position has been for years. Infact Carmack has said from the outset that if D3D were better he'd use it but way back in his famous
This is not the death of OpenGL. In fact with D3D copying so much of OpenGL it would almost be a semantic argument if D3D weren't such a darned mess. Even with your nightmare scenario (which isn't going to transpire) porting D3D games to OpenGL should actually be pretty easy and Loki have already demonstrated this.
The dire warnings seem overstated considering what is already accepted practice. They just pull the suspects emails in question prior to searching. Omnivore sounds like it was open to abuse and if that was deployed it should never have been, it's like wire tapping a small town to get evidence on one individual. Carnivore sounds like a right minded attempt to restrict scanning to the suspects account.
So what's new?
They still need a court order and they could always tap the suspects phone any time as things stand. This just let's them tap an account than might be moving on a dial in from different locations. The whole system has always been build on trust and controlled by the fact that any abuse of the system won't pass muster as evidence in court anyway.
So, if a Judge let them deploy Omnivore it sounds like there's a need for some legislation to prevent this sort of dragnet approach in future but the Carnivore system is exactly the kind of thing I'd expect the FBI to be getting up to, why is everyone so surprised? The intention of developing Carnivore as a discriminating filter seems to be a move in the right direction IF it only traps and searches the email of the suspect, and that's the whole point of the newer system.
Move along folks, there's nothing to see here.
Yes NASA is aimless.
The whole problem is that they don't see the true intrinsic value of manned flight. Instead of lofty goals we get expensive boondoggles in the shuttle which would be better spent on unmanned missions or a manned mission with a worthwhile goal. Shouldn't the Sojourner landing have told NASA something? Right now we're building the most expensive tin can in history to orbit the Earth (IF the Russians don't blow up a critical part on the pad). It's a joke to suggest that this has proportionate scientific value.
When China start manned missions and India put a man on the moon it will really show the world what incredible under accomplishment NASA has been guilty of these last few decades. Hopefully it will prompt some action.
The Challenger disaster and the accompanying Disneyland philosophy that no risk is acceptable for manned flight is equally foolish. Tourists die climbing Everest each year, thousands die in auto accidents, hundreds die in plane crashes fulfilling the most mundane objectives. Historically huge risks have been undertaken by celebrated pioneers of new frontiers. Unfortunately that isn't acceptable for NASA management, they can easily find the few brave souls to shoulder the burden, but instead they are mired in ineffective safety procedures because they have forgotten what it is like to excel. NASA has given up on the dream. Here's a plan to cut costs AND save lives; half the budget for manned space, cut back on the excessive safety and associated bureaucracy and spend the savings on road safety awareness in high schools or on medical equipment at a few hundred hospitals. I only make this point to illustrate how completely imbalanced the pioneering spirit has become at NASA, and it isn't for the want of brave astronauts, it's done for the political careers of their masters. The upshot is that the very people who would volunteer for worthy manned missions get to orbit the Earth in (safe) tin cans instead of doing some real exploration. I think I know what choice they would make given the option.
Well like you say you'd be bus limited. You still get way more bandwidth over AGP4x with fast writes. You couldn't compete with hardware (on the card) t&l. I expect some stuff would be faster, you could reject back faces and cull to frustum per primitive so if you had a lame app with display lists you might win, but you'd have to store the display list geometry on the PCI card, you would just get hosed trying to send it to the cards over PCI then get it back transformed.
You wouldn't need a driver per processor but all told it's only uninteresting for the stated problem because of the bandwidth requirement. This is exactly the kind of problem this card is there to solve provided you have a greater compute to i/o bandwidth ratio.
This is cool, for the right class of problem. I doubt you'd see this in a mainstream server, remember this will be running specialized cross compiled code which just reads & writes to system memory. There would be no real system level interface except back on the native x86 processor once it got the results. Anything running on it would be specially ported although it looks like it can support interface cards of it's own accord. It sounds like the ultimate seti@home processing system once someone ports the client :-). One thing that isn't clear, it runs ELF binaries but I assume you'd have to upload the executable to it's memory first it wouldn't see the native file system, is this correct? Would it be able to read & write to a memory mapped file? One other issue, memory endianess would be reversed with this processor vs x86 right? So your memory communications would have to twiddle all the bytes based on data type of you wanted to do anything with the results on an x86 or other linux native application right?
Does anyone here have any experience using this system?
Rambus have filed hundreds of patents last year. If you look at the rate they are filing (it's accelerating) and given that they make NO product themselves they are in a very strong position.
They will never have to pay anyone royalties because they have no need to license others ideas, just invent incrementally on existing ideas patented or not. Anyone hoping to challenge on a few patents is sure to run foul of their portfolio in some other area. This is a really powerful play by Rambus. I don't think people on slashdot really understand the magnitude of what's going on here or the strength of the Rambus position. Rambus is likely to win big time, they have a slew of patents in the works which nobody even knows about yet and a strategy to license them aggressively. I doubt they really give much on a damn about which memory technology wins. Patents are almost always written to be general enough to cover many implementations, hence the SDRAM claims, and Rambus has raced ahead of everyone in terms of focusing on one area of herdware design and filing as many ideas as they can think of.
A patent dispute can start with one claim and then snowball afterwards as the stakes increase and more ideas are disputed, the normal defense is to counter sue and then horse trade IP. Unfortunately for the rest of the world, Rambus use no IP from anyone else, they are just an ideas company. So Hitachi puts up this lame defense that Rambus should be forced to hand over their IP. The next Rambus move was probably to threaten them with breaching 50 other patents they happen to have at which point Hitachi caved. This last part is just idle speculation but probably close to the mark. Even if they won on the SDRAM issue they couldn't escape all the patents filed by Rambus.
Your prediction is self defeating. If they don't have the influence over the SDRAM vendors then they don't have the influence to force the RDRAM issue. The absolute best they could manage would be to try and pressure Hitachi & Toshiba to agree to higher volume RDRAM shipments, and that will only serve to reduce the price of RDRAM, not exactly a nightmare scenario. If they do have the influence with other vendors then they don't need to wield it because DDR will be more lucrative for the anyway.
Even if we use your figure of 10% to Intel (I thought it was 20%) that's still HUGE, at current market cap 10% that's just shy of $1.1B (that's B as in billion). You have to remember that in terms of the stock actually controlled by Rambus and their overall assets this is an even more significant number. In addition it would look pretty devastating on their books when they actually have to write it down. When they originally cut the deal they obviously thought they had nothing to lose, they'd have given their right stone for a majority market share but as they near the end of the Faustian pact things will be looking very different to them.
Latencies are only larger for memory on the same channel, and the simpler layouts are not unimportant if you want to try and suport additional channels. The 2 sticks are a non issue, the sticks would obviously be smaller. You're also talking about DDR memory which is faster than RDRAM, it's the comparrisons with 133 SDRAM which have been unfair. You'll have a point when DDR mobos are on the shelves but then Rambus might be pleased about that.
On the issue of Rambus' memory IP, if the advantages were unimportant, the memory manufacturers would eschew them and escape the royalties. Your rhetoric on the standards committee is lifted straight from the Hitachi countersuit against Rambus which has now been dropped like a stone. In fact you didn't get it quite right (maybe you got Dr Tom's contorted version of the facts), because one of the things that Hitachi also claimed was that Rambus should have disclosed their existing patents when they joined the group. This would have given them forewarning to demand free use or perhaps to steer around the technology. Any way you look at it this was a desperate ploy by Hitachi and had a snowballs chance in hell of working in court. You could claim they dropped the suit to help sell the business but the truth is that Hitachi were crapping themselves at the prospect of losing their shirt on this. The memory business is booming right now (they can't make enough of the stuff and prices are rocketing) and they can't have been in any hurry to sell, that's just more smoke.
So which of us actually knows what they're talking about? I think I have my facts straight.
This is one ambiguous Rambus response to a question. I read the quote I don't need to go back, for someone so cynical about Rambus you're awful quick to take YOUR interpretation of one ambiguous statement as gospel truth. The objective is clearly to maximize the hype. Unlike you I've also read other Rambus pronouncements over the last few days. The simple fact is that Rambus have already stated in other communications that they will charge HIGHER royalties for DDR SDRAM than RDRAM. So you explain to me why they will now want RDRAM to win. Another fact you are ignoring is that if RDRAM wins in terms of Intel mobo share then Rambus has to hand over a huge chunk of their company to Intel, so factor that into your reasoning and again explain your conclusion. Don't jump on the rabid monopoly theory based on one remark, think for a second and tell me why on Earth they'd still want RDRAM to win.
As I've already stated in other posts, if all they can do is influence Hitachi and Toshiba, that isn't enough to affect pricing or volumes significantly. Just look at the price of SDRAM at source recently. It's lept by a huge ammount, it may hit 100% price increase that hasn't translated to anything near the kind of a price hike predicted at the hands of Rambus royalties. The simple fact is this whole theory is baseless. Even if Rambus charged huge royalties it still wouldn't make SDRAM significantly less competitive against RDRAM. It isn't driven by the royalties they charge, it's driven by the volumes it's made in.
This whole story is a lot of nonsense.
That must be why the Times incompetently attempted to cover the names, and why even Jim Risen, the great sleuth who couldn't sleuth out how to write pdf files properly now thinks lives are being endangered. It isn't just about the spies, it's their extended families. In anycase you don't know who is left alive.
Maybe you can testify at their trial, they have religious courts you know. No need for judges when you have a cleric to run the proceeding you see. All the CIA et.al. needed to do was NOT release the names. There was a zero activity required to keep those few lives safe.
There's nothing Hollywood about throwing Jim Risen to the DOJ in an attempt to find his source. No cover needs to be blown. It would be worth testing to see if freedom of the press extends to leaking secret CIA documents and endangering lives in an attempt to promote your web site / newspaper.
As for getting his disk, I doubt getting a job as a janitor in his office building represents a serious risk, then again his system is probably on the network and that wouldn't exactly represent a risk of exposure if hacked from Tehran.
Now if I can think of it in the time it takes to write a post how long do you think it would take counter intelligence to think of it? Ofcourse those Iranians live in caves and don't have computers, silly me.
Your trouble is you are too busy being cynical about CIA doctrine to expect anything useful to be done about this. So who's given up on the USA, you or me?
Subvert technical excellence? Get a clue. They clearly invented some of the SDRAM & DDR technology before anyone else, otherwise Toshiba and Hitachi wouldn't have agreed to hand over many millions in royalties.
Rambus are investing a fortune in R&D. They want their technology to be adopted. They now have a foot in BOTH camps so actually care LESS now about which technology wins. RDRAM actually might have some merit with enough channels and in terms of simpler layout despite all the FUD. The price depends entirely on how many manufacturers start pumping it out.
If Intel are able to ship enough chipsets supporting RDRAM then Rambus must hand over 20% of their company. Since they will get more cash from their SDRAM & DDR patents then I expect even Rambus wants RDRAM to fade away.
Aside from anything else, it's not the license fees which drive up the price, it's the volumes of the product shipped. There's not much Rambus can do on the royalties front to change that.
Even if the doomsday scenario were plausable, if Rambus manage to stitch up all the manufacturers with license agreements then they win with SDRAM & DDR, if they don't then they can't force the RDRAM issue because some manufacturers would be royalty free.
This whole idea is just some scaremongering.
See Cliffs thread on Linux Development.
Damn it the more I read the worse this gets.
Jim Risen is the guy who's caused this problem now he's accusing the people who exposed his treachery of endangering the very lives his incompetence has thrown to the wolves.
I say arrest him and show him the sharp end of a 10 year stretch unless he comes up with the name of his CIA mole. Even if he hadn't screwed up like this, he'd still have advertised to all & sundry that he had the original file somewhere on his systems. How difficult would it be for a half competent agent to wander in and take his whole file system?
Well the bible may say don't hold the children responsible for the sins of the fathers, but that doesn't really cut it in Iran.
Ignoring the fact that you seem to think it's OK to abandon former allies once they hit 70 years old, what about their families? There is also a more significant contemporary credibility problem here. The CIA can't afford to look this foolish to people they need work with TODAY.
This is a pretty significant lapse, then again it's the CIA, you gotta ask yourself if it's entirely accidental too. Maybe this is a CIA plot to deliberately discredit the people named in the document. What better way to 'accidentally' leak a few names.
Wheels within wheels my friend.