When the 45nm process comes out the leakage problem will be completly fixed completely.
Yeah, and they'll get the Nobel Prize for that, since the power consumption due to leakage increases with the descrease in process size. In fact, it's getting so high in chips being currently designed that the static power consumption is becoming higher than the dynamic power consumption due to the signal switching.
But, Intel will fix it completely with their next process. It'll be easy.
Your answer, Windows Vista. Thanks to the hubris of Microsoft, Windows Vista will be ignored by gamers just as they ignored Windows2000 and shunned WindowsME. Doing stupid deliberate things like retarding the performance of OpenGL in Vista in favor of DirectX is enough to alienate the likes of id Software.
There's nothing deliberate about what they are doing. Please try to understand what's going on before coming up with generalizations like "Vista will be bad for Open GL so all games will go to Mac because all games are written by Id Software".
The Vista desktop uses Direct X to render the new desktop features, so it can't run Open GL natively at the same time... so, they provided a wrapper, which causes a performance hit. As far as the full-screen games are concerned, though, they don't use the desktop, hence the aeroglass support goes away and ATI or nVidia native OpenGL driver kicks in.
So, don't be sad. Your OpenGL games will run just as fine as they used to.
The OpenGL windowed applications might suffer performance degradation, but that's another story.
Watch out Speeders if the RF plates become a reality. Sensors along roads take your position, computers extrapolate speed and two days later you get your ticket in in the mail.
Well, they already do that with automated radar guns and red-light cameras... Having computers track you over a certain distance and calculate your speed seems like a more complicated way to do it.
And Big Brother Watching you? You wouldn't even need the software predictions mentioned a few weeks ago -- just follow the RF tag around town
An RFID tag is not a GPS receiver + data transmitter... It's just an ID tag. In order for the "big brother" to follow you around town, they'd have to install RFID readers on every corner of every town, and then network them all together to keep track of where you're going. It'll be quite a while before that's feasible.
The patent is not called 'dynamically generating a Web page in response to the request', as the poster would have you believe, but 'System and method for managing dynamic web page generation requests'. It describes a system in which a dispather takes the dynamic page requests and distributes them to the least loaded 'page servers' which actually do the data retrieving and page generation. It's a system for reducing the load on the main web server if it generally receives a lot of dynamic page requests.
It's not a particularly novel system, but if enforced, it will most certainly not 'shut down almost every dynamic site on the Internet'.
Yes, because the only possible uses a high-capacity optical disc could have on a game console would be for watching movies. You couldn't possibly use the much higher-capacity discs for games on this gaming system, right?
Is that going to make you not buy a gaming system? Poor game developers, they'll have to cram the game onto one DVD! Even though the game is good, just imagine how much better it could've been with all that extra space on the media!
Bah, look at the PC games with all the high-resolution graphics, still doing fine with the regular DVD media. If I see the games suffering dramatically from the lack of space, then I'll decide to sell my xbox and buy more ps3 games, but it's not going to stop me from buying xbox at launch if the games are good.
Right, and marketing research is always 100% accurate... Nobody has ever done marketing research that said a product would be a great success, only to have it become a terrible failure, right?
It's much harder to do research to figure out if a product will be a success than to ask people do you want HD-DVD on your xbox. If I may be so bold to say, I believe that research will be accurate enough.
And, as you believe, the HD-DVD is important for the extra space on the media, then that research is even easier to do -- call up all your game developers and ask them what they can do with it.
IF the 360 flops? Of course the 360 launch is going to flop. Let's not forget that Microsoft announced that 360s will ship with standard DVD drives and the HD-DVDs will be coming later (next year most likely).
I, for one, couldn't care less about the lack of HD-DVD on Xbox 360. I'm going to get an Xbox 360 to play the games, and when in some distant future HD-DVD movies start becoming widely available, I'll get the Xbox upgrade or buy an HD-DVD player.
Do you not think Microsoft did some market research to see how many people would hold out on getting their new gaming platform because of the lack of HD-DVD support? I'm pretty sure a good number of people didn't care, or you'd be seeing an HD-DVD drive in the Xbox.
Read Kim Stanley Robinson's Red/Green/Blue Mars trilogy to spark your imagination. At times its pretty slow and dry but it has a vision that is captivating. A must read for geeks, space enthusiasts and sci fi fans.
Or, if you'd like the same great vision without the endless pages describing Mars relief, try Greg Bear's Moving Mars, instead. It's short on poetic parts that are prominent in the Robinson's trilogy, but it's full of great ideas and characters.
There's a good book called "Guns, Germs, and Steel" that covers many topics on the rise of civilizations and proposes theories on why the world is the way it is now. There's an interesting section of animal domestication that explains why, even with the latest technologies, we simply cannot domesticate certain wild animals, like elephants.
The bottom line is that the closest we can get to "farming" of elephants is to have wild-life preserves where they can grow on their own and will not be killed by humans. With all that meat, muscle power, and ivory, the elephants have always been a perfect target for domestication -- but those attempts have never succeded.
It's like forcing a minority to take the stairs but letting the white guy use the elevator, even though both are perfectly capable of riding the elevator.
I'd say it's more like letting the condo tenants use the elevators, while the guests are restricted to using the stairs.
Re:SCAMS every where
on
Shopping Online
·
· Score: 4, Informative
Check out redflagdeals.com regularly for Canadian online (and offline) specials.
So, from one avid gamer to another, tell me what these complicated and involved games are?
First, to put things in perspective, I was comparing the complexity of the new games vs. the early games that the original poster was referring to -- the games that his 5 year old son enjoys. Compared to the original Zeldas, Marios, and going even further, Tetris, any of the modern games are a lot more complex.
While the basic idea for most of these new games might be rehashes, the implementation is quite complex. Simulation/stregy games like Civilization 3, SimCity 4, Football Manager 2005, and adventure games like Syberia II, are all very involved games, even though they had much simpler predecesors available some number of years ago. There's a whole new type of games in recent years in MMORPGs, and World of Warcraft is anything but a simple game. Splinter Cell requires you to master a wide variety of moves and tools in other to be succesful against other players, which is a big improvement over the original FPS games, and Chaos Theory now adds an element of teamwork to that. Rome: Total War puts in total control of giant armies, raising the complexity of RTS games to another level. Fight Night: Round 2 gives you almost perfect control of the boxer's complete body, comparing that to the simple button mashing of the original boxing games.
All of this is subjective, of course, and maybe you still find all these things overly simple, but, IMHO, most gamers will agree that the games listed above are quite a bit more complex that what was popular 10-15 years ago.
1. With the next generation of consoles becoming nothing more than computers,
The game console makers want you to feel like the next generation of consoles can do everything a computer can do. They've wanted you to do think that since PS2. Well, if you have taken the time to RTFA, you would've seen that they are, again, not able to do what the computers can do, and are just machines optimized for easy gaming.
PCs can still handle higher resolutions in generally higher FPS with adaptable display features to match your needs/hardware. They are just generally more complicated to use.
Tell me something: Why do games today *have* to be something I can't let my 5 year old son play? He still plays the old Nintendo
Get him a Gamecube. Nintendo still makes tons of games that are fun for young kids.
Maybe it's just me. Maybe I'm getting old.
Yeah, that's probably right. If you ask the teens these days, they'll tell you that there are many fun games out there that they enjoy very much. You probably don't have time to look for them or try them out, because there are so many more games available these days. A lot of these games are a lot more complex, which makes them more fun for the generation of teens that grew up with the latest technology, and doesn't find simple games as much fun as the previous generation did. It doesn't mean that they are not fun, it just means that you have to be more comfortable with the complexity of them, and you have to invest more time into playing them. The older you get, the less time you have in your life to devote to serious gaming... unfortunately.
Additionally, each prefetch would have come from the same IP as the actual hit, so they would not be counted separately.
Your other point is valid, but this one is not. The problem isn't that the prefetch and the hit are counter seperately as 2 hits, but that the prefetch without a hit is counter as a hit.
You might be misunderstanding what I'm trying to say, or I might've not been clear.
Sticking in the extra pipelines and still getting good yields is an achievement in process/production technology, and possibly layout if any improvements there resulted in a die shrinkage.
The design of each pipeline itself has not changed much from the previous generation, making the features/design of this card unimpressive compared to, say, switch they made for the previous generation of cards which introduced PS 3.0, higher clock speeds, and higher performance in a card of an equal number of pipelines.
So, I guess I was speaking about the design achievement of making this card as opposed to the benchmark scores when I said that it is hardly impressive.
Don't be silly. PS3 is coming out in a year, there's no way Sony will be buying a chip that has been out in mass market for a year for their new flagship entertainment product. On top of that, this GPU is the same as the previous one from nVidia, with extra pipelines -- it's hardly impressive!
IMO, the PS3 will be using the next-generation GPU that will most likely be available for PC at about the same time as the PS3.
ATI is already shipping DirectX 9 class IGPs for both AMD and Intel, and the reviews are looking pretty good: here's an ASUS board for Intel CPUs, and a review of the AMD chipset. If you're looking for very good IGP, these definitely look like a very good option!
Exactly. They pissed of an actual customer, who they had already spent hundreds of dollars in acquiring. He's gone away now. He's not coming back. All over a sloppy driver that shouldn't have been, and needn't have been (see your own argument) sloppy. Lost customers always hurt a company more than customers never obtained in the first place.
I agree that the drive shouldn't have been sloppy, but to say that it needn't have been is to neglect the complexity of a modern video driver. This piece of software has to facilitate maximum performance out of a piece of hardware on many different hardware configurations, while running in tandem with tens of thousands of games and applications, all without a single problem -- because even one crash will make that particular customer go up in arms and blame the driver.
It's not easy making an excellent video card driver, and just throwing more people at it is not going to do the job. I don't think ATI underestimated the importance of the driver and kept that team small -- they simply didn't have the right people or the right process to do it in an excellent way a few years ago.
Now, it's just emotional thinking to say "they had a crappy driver 3 years ago -- even though all the reviewers say that the new drivers are awesome, and that the cards are awesome, I'm still not going to buy them". Luckily, not many people think like that (witness the market share numbers now vs. 3 years ago), but those who do are more vocal about it, just because it is an emotional subject to them.
Another proof that most people don't think way -- look at cars... Hyundai used to make some pretty crappy cars, and now they are overtaking some well-known Japanese brands in market share, because they've figured out a way to produce quality products.
Do you think companies like ATI have the same people working on a card design with 512MB of ram, and on coding drivers?
These things work in parallel, they could be producing a new board design every day and still have the driver team working on the drivers for the same amount of time.
Besides that, ATI has put a lot of focus on their driver design over the last couple of years, and you'll find very little (outside of the ordinary) complains about their current state, apart from an occasional troll like the GP, who probably had a bad experience some number of years ago and has never bothered to reconsider his opinion.
So if it's just the design that's done here, it's as easy to outsource as software, right?
Not quite so easy. The design needs to be simulated, emulated, synthesized, and the chip needs to be properly tested and qualified. All of that requires expensive equipment and lots of workers skilled in many different types of EE work.
Some IP core design is being outsourced by smaller companies, but I've seen this work well only for fairly simple and small cores -- because of the reasons above.
Are you sure hardware is tougher to outsource? What percentage of hardware is made in USA?
The majority of high-tech PC-related design work, for example, is done in North America... Intel, AMD, ATI, nVidia, IBM, Altera, Xilinx, all do their design work here.
The manufacturing is done in Taiwan and China for the most part, but that's not what most EE graduates are aiming for anyway.
Yes, because your bills are sent to where you are with your laptop *at that moment* (highway, hotel, coffee shop), not the billing address you specified with you signed up for the service.
When you call 911 the "regular way", you don't get your local coffee shop's 911 service, you get the 911 for your municipality/region. That's most likely the same region as your billing address, unless you travel out of town a lot, as the parent said.
Put simply, dual core means that both CPUs are on the same piece of silicon. They can share a unified cache, access it faster, and resolve deadlocks...
Actually, the Intel dual-core CPUs are simply two core dies in the same package, not two cores on the same die. So, they do not share cache and resolve deadlocks in any faster way than two separate CPUs.
It's quite a bit different than a SoC design where you put a large number of components onto the same die. While SoC will suffer from yield rates because of a larger die, the dual-core strategy will not, because each die is still as small as the original single-cpu solution.
Yes, HT does connect the cpu to the memory controller, but the new AMD chips do not have a Northbridge, so you're off there.
As MagnusDredd pointed out, they indeed have a northbridge, but the memory controller is on the CPU. *But*, in a multi-CPU unit, HT will be used to make connection between the CPUs, and each CPU will have a memory controller. So, whenever a CPU has to access a memory location not controlled by its local MC, it will have to go over HT to other CPUs to retrieve memory. The other CPUs, in essence, act as a traditional northbridge with an MC for that particular transaction.
Slashdot Mirror
When the 45nm process comes out the leakage problem will be completly fixed completely.
Yeah, and they'll get the Nobel Prize for that, since the power consumption due to leakage increases with the descrease in process size. In fact, it's getting so high in chips being currently designed that the static power consumption is becoming higher than the dynamic power consumption due to the signal switching.
But, Intel will fix it completely with their next process. It'll be easy.
Your answer, Windows Vista. Thanks to the hubris of Microsoft, Windows Vista will be ignored by gamers just as they ignored Windows2000 and shunned WindowsME. Doing stupid deliberate things like retarding the performance of OpenGL in Vista in favor of DirectX is enough to alienate the likes of id Software.
There's nothing deliberate about what they are doing. Please try to understand what's going on before coming up with generalizations like "Vista will be bad for Open GL so all games will go to Mac because all games are written by Id Software".
The Vista desktop uses Direct X to render the new desktop features, so it can't run Open GL natively at the same time... so, they provided a wrapper, which causes a performance hit. As far as the full-screen games are concerned, though, they don't use the desktop, hence the aeroglass support goes away and ATI or nVidia native OpenGL driver kicks in.
So, don't be sad. Your OpenGL games will run just as fine as they used to.
The OpenGL windowed applications might suffer performance degradation, but that's another story.
Watch out Speeders if the RF plates become a reality. Sensors along roads take your position, computers extrapolate speed and two days later you get your ticket in in the mail.
Well, they already do that with automated radar guns and red-light cameras... Having computers track you over a certain distance and calculate your speed seems like a more complicated way to do it.
And Big Brother Watching you? You wouldn't even need the software predictions mentioned a few weeks ago -- just follow the RF tag around town
An RFID tag is not a GPS receiver + data transmitter... It's just an ID tag. In order for the "big brother" to follow you around town, they'd have to install RFID readers on every corner of every town, and then network them all together to keep track of where you're going. It'll be quite a while before that's feasible.
The patent is not called 'dynamically generating a Web page in response to the request', as the poster would have you believe, but 'System and method for managing dynamic web page generation requests'. It describes a system in which a dispather takes the dynamic page requests and distributes them to the least loaded 'page servers' which actually do the data retrieving and page generation. It's a system for reducing the load on the main web server if it generally receives a lot of dynamic page requests.
It's not a particularly novel system, but if enforced, it will most certainly not 'shut down almost every dynamic site on the Internet'.
Yes, because the only possible uses a high-capacity optical disc could have on a game console would be for watching movies. You couldn't possibly use the much higher-capacity discs for games on this gaming system, right?
Is that going to make you not buy a gaming system? Poor game developers, they'll have to cram the game onto one DVD! Even though the game is good, just imagine how much better it could've been with all that extra space on the media!
Bah, look at the PC games with all the high-resolution graphics, still doing fine with the regular DVD media. If I see the games suffering dramatically from the lack of space, then I'll decide to sell my xbox and buy more ps3 games, but it's not going to stop me from buying xbox at launch if the games are good.
Right, and marketing research is always 100% accurate... Nobody has ever done marketing research that said a product would be a great success, only to have it become a terrible failure, right?
It's much harder to do research to figure out if a product will be a success than to ask people do you want HD-DVD on your xbox. If I may be so bold to say, I believe that research will be accurate enough.
And, as you believe, the HD-DVD is important for the extra space on the media, then that research is even easier to do -- call up all your game developers and ask them what they can do with it.
IF the 360 flops? Of course the 360 launch is going to flop. Let's not forget that Microsoft announced that 360s will ship with standard DVD drives and the HD-DVDs will be coming later (next year most likely).
I, for one, couldn't care less about the lack of HD-DVD on Xbox 360. I'm going to get an Xbox 360 to play the games, and when in some distant future HD-DVD movies start becoming widely available, I'll get the Xbox upgrade or buy an HD-DVD player.
Do you not think Microsoft did some market research to see how many people would hold out on getting their new gaming platform because of the lack of HD-DVD support? I'm pretty sure a good number of people didn't care, or you'd be seeing an HD-DVD drive in the Xbox.
Guaranteed -- most consumers have no idea that AMD chips are faster.
Most consumers have never heard of AMD, in fact. Most consumers still think "CPU" is the computer case with "some stuff like memories" inside.
Read Kim Stanley Robinson's Red/Green/Blue Mars trilogy to spark your imagination. At times its pretty slow and dry but it has a vision that is captivating. A must read for geeks, space enthusiasts and sci fi fans.
Or, if you'd like the same great vision without the endless pages describing Mars relief, try Greg Bear's Moving Mars, instead. It's short on poetic parts that are prominent in the Robinson's trilogy, but it's full of great ideas and characters.
Those are just wild elephants that are tamed enough to be ridden. Nobody is growing domestic Indian elephants on a farm.
There's a good book called "Guns, Germs, and Steel" that covers many topics on the rise of civilizations and proposes theories on why the world is the way it is now. There's an interesting section of animal domestication that explains why, even with the latest technologies, we simply cannot domesticate certain wild animals, like elephants.
The bottom line is that the closest we can get to "farming" of elephants is to have wild-life preserves where they can grow on their own and will not be killed by humans. With all that meat, muscle power, and ivory, the elephants have always been a perfect target for domestication -- but those attempts have never succeded.
It's like forcing a minority to take the stairs but letting the white guy use the elevator, even though both are perfectly capable of riding the elevator.
I'd say it's more like letting the condo tenants use the elevators, while the guests are restricted to using the stairs.
Check out redflagdeals.com regularly for Canadian online (and offline) specials.
So, from one avid gamer to another, tell me what these complicated and involved games are?
First, to put things in perspective, I was comparing the complexity of the new games vs. the early games that the original poster was referring to -- the games that his 5 year old son enjoys. Compared to the original Zeldas, Marios, and going even further, Tetris, any of the modern games are a lot more complex.
While the basic idea for most of these new games might be rehashes, the implementation is quite complex. Simulation/stregy games like Civilization 3, SimCity 4, Football Manager 2005, and adventure games like Syberia II, are all very involved games, even though they had much simpler predecesors available some number of years ago. There's a whole new type of games in recent years in MMORPGs, and World of Warcraft is anything but a simple game. Splinter Cell requires you to master a wide variety of moves and tools in other to be succesful against other players, which is a big improvement over the original FPS games, and Chaos Theory now adds an element of teamwork to that. Rome: Total War puts in total control of giant armies, raising the complexity of RTS games to another level. Fight Night: Round 2 gives you almost perfect control of the boxer's complete body, comparing that to the simple button mashing of the original boxing games.
All of this is subjective, of course, and maybe you still find all these things overly simple, but, IMHO, most gamers will agree that the games listed above are quite a bit more complex that what was popular 10-15 years ago.
1. With the next generation of consoles becoming nothing more than computers,
The game console makers want you to feel like the next generation of consoles can do everything a computer can do. They've wanted you to do think that since PS2. Well, if you have taken the time to RTFA, you would've seen that they are, again, not able to do what the computers can do, and are just machines optimized for easy gaming.
PCs can still handle higher resolutions in generally higher FPS with adaptable display features to match your needs/hardware. They are just generally more complicated to use.
Tell me something: Why do games today *have* to be something I can't let my 5 year old son play? He still plays the old Nintendo
Get him a Gamecube. Nintendo still makes tons of games that are fun for young kids.
Maybe it's just me. Maybe I'm getting old.
Yeah, that's probably right. If you ask the teens these days, they'll tell you that there are many fun games out there that they enjoy very much. You probably don't have time to look for them or try them out, because there are so many more games available these days. A lot of these games are a lot more complex, which makes them more fun for the generation of teens that grew up with the latest technology, and doesn't find simple games as much fun as the previous generation did. It doesn't mean that they are not fun, it just means that you have to be more comfortable with the complexity of them, and you have to invest more time into playing them. The older you get, the less time you have in your life to devote to serious gaming... unfortunately.
Additionally, each prefetch would have come from the same IP as the actual hit, so they would not be counted separately.
Your other point is valid, but this one is not. The problem isn't that the prefetch and the hit are counter seperately as 2 hits, but that the prefetch without a hit is counter as a hit.
You might be misunderstanding what I'm trying to say, or I might've not been clear.
Sticking in the extra pipelines and still getting good yields is an achievement in process/production technology, and possibly layout if any improvements there resulted in a die shrinkage.
The design of each pipeline itself has not changed much from the previous generation, making the features/design of this card unimpressive compared to, say, switch they made for the previous generation of cards which introduced PS 3.0, higher clock speeds, and higher performance in a card of an equal number of pipelines.
So, I guess I was speaking about the design achievement of making this card as opposed to the benchmark scores when I said that it is hardly impressive.
No doubt that this is the heart of the PS3.
Don't be silly. PS3 is coming out in a year, there's no way Sony will be buying a chip that has been out in mass market for a year for their new flagship entertainment product. On top of that, this GPU is the same as the previous one from nVidia, with extra pipelines -- it's hardly impressive!
IMO, the PS3 will be using the next-generation GPU that will most likely be available for PC at about the same time as the PS3.
ATI is already shipping DirectX 9 class IGPs for both AMD and Intel, and the reviews are looking pretty good: here's an ASUS board for Intel CPUs, and a review of the AMD chipset. If you're looking for very good IGP, these definitely look like a very good option!
Exactly. They pissed of an actual customer, who they had already spent hundreds of dollars in acquiring. He's gone away now. He's not coming back. All over a sloppy driver that shouldn't have been, and needn't have been (see your own argument) sloppy. Lost customers always hurt a company more than customers never obtained in the first place.
I agree that the drive shouldn't have been sloppy, but to say that it needn't have been is to neglect the complexity of a modern video driver. This piece of software has to facilitate maximum performance out of a piece of hardware on many different hardware configurations, while running in tandem with tens of thousands of games and applications, all without a single problem -- because even one crash will make that particular customer go up in arms and blame the driver.
It's not easy making an excellent video card driver, and just throwing more people at it is not going to do the job. I don't think ATI underestimated the importance of the driver and kept that team small -- they simply didn't have the right people or the right process to do it in an excellent way a few years ago.
Now, it's just emotional thinking to say "they had a crappy driver 3 years ago -- even though all the reviewers say that the new drivers are awesome, and that the cards are awesome, I'm still not going to buy them". Luckily, not many people think like that (witness the market share numbers now vs. 3 years ago), but those who do are more vocal about it, just because it is an emotional subject to them.
Another proof that most people don't think way -- look at cars... Hyundai used to make some pretty crappy cars, and now they are overtaking some well-known Japanese brands in market share, because they've figured out a way to produce quality products.
Do you think companies like ATI have the same people working on a card design with 512MB of ram, and on coding drivers?
These things work in parallel, they could be producing a new board design every day and still have the driver team working on the drivers for the same amount of time.
Besides that, ATI has put a lot of focus on their driver design over the last couple of years, and you'll find very little (outside of the ordinary) complains about their current state, apart from an occasional troll like the GP, who probably had a bad experience some number of years ago and has never bothered to reconsider his opinion.
So if it's just the design that's done here, it's as easy to outsource as software, right?
Not quite so easy. The design needs to be simulated, emulated, synthesized, and the chip needs to be properly tested and qualified. All of that requires expensive equipment and lots of workers skilled in many different types of EE work.
Some IP core design is being outsourced by smaller companies, but I've seen this work well only for fairly simple and small cores -- because of the reasons above.
Are you sure hardware is tougher to outsource? What percentage of hardware is made in USA?
The majority of high-tech PC-related design work, for example, is done in North America... Intel, AMD, ATI, nVidia, IBM, Altera, Xilinx, all do their design work here.
The manufacturing is done in Taiwan and China for the most part, but that's not what most EE graduates are aiming for anyway.
Yes, because your bills are sent to where you are with your laptop *at that moment* (highway, hotel, coffee shop), not the billing address you specified with you signed up for the service.
When you call 911 the "regular way", you don't get your local coffee shop's 911 service, you get the 911 for your municipality/region. That's most likely the same region as your billing address, unless you travel out of town a lot, as the parent said.
Put simply, dual core means that both CPUs are on the same piece of silicon. They can share a unified cache, access it faster, and resolve deadlocks...
Actually, the Intel dual-core CPUs are simply two core dies in the same package, not two cores on the same die. So, they do not share cache and resolve deadlocks in any faster way than two separate CPUs.
It's quite a bit different than a SoC design where you put a large number of components onto the same die. While SoC will suffer from yield rates because of a larger die, the dual-core strategy will not, because each die is still as small as the original single-cpu solution.
Yes, HT does connect the cpu to the memory controller, but the new AMD chips do not have a Northbridge, so you're off there.
As MagnusDredd pointed out, they indeed have a northbridge, but the memory controller is on the CPU. *But*, in a multi-CPU unit, HT will be used to make connection between the CPUs, and each CPU will have a memory controller. So, whenever a CPU has to access a memory location not controlled by its local MC, it will have to go over HT to other CPUs to retrieve memory. The other CPUs, in essence, act as a traditional northbridge with an MC for that particular transaction.