The one-chip PC is a falicy... if the integration of all blocks produced higher performance, then you can charge more for that integration. The slower blocks (ATA, USB, PCI) don't perform significantly better when integrated, thus consumers don't value integration (board manufactures will, but only by a fixed amount).
If you integrated the gpu/ethernet/sdram/cpu, then all the high-speed blocks can talk with each other at die level speeds. It's clear as day that performance will significantly increase.
The southbridge IO (ATA USB PCI) requires many pins, and moving them to a single chip doesn't save enough money at board level to justify the added expense at die level. The power cost of running a bus to the southbridge is tiny compared to the total system budget.
It's quite clear that the AMD-ATI merger was to aquire the IP and expertise necessary to integrate gpu core(s) on the same die as cpu core(s). Nvidia does not have to actually market a design, but rather patent some key concepts, and this could provide revenue or protection.
I would very much doubt that they could compete with AMD and Intel who have already patented many x86 cpu concepts.
It's a shame that Intel has decided not to buy nvidia, and go it alone with it's own design staff.
I was under the impression that AMD's "native" quad-core design would be very hardware virtualization friendly. The fact that these chips have a very modular memory design and are not used in MCMs (multi-chip-modules), should allow for the simplest possible virtualization configuration.
I would really like to see desktop virtualization that allows fast access to my legacy windows apps (rather than my present approach of vncing to a windows box powered on all the time). The old windows box only costs electricity, but if any hardware on it fails, I really don't want to invest more money in a make-shift solution.
Smart consumers always stay away from a 1st generation product. I would gladly give-up some features for peace of mind. The more the cost of the equipment is, the more you want to stay with a proven vendor.
It's possible for any generation to be plagued with mistakes, so it's best not to be the early adopter.
80 cores on a chip and optical interconnects sounds great...but I wish they would talk more about the end application goals (i.e. a system that does 1080p ray-tracing and has 100% speech recognition). It's great that they are pushing the design limits as they are, but without clear vision of how the technology is to be used, it's likely that it will miss the mark.
I was hoping to hear about a single die with cpu/northbridge/southbridge/gpu all integrated (and for mobile use)... that would certainly turn the computer market on it's head. Nvidia knows this already, and has everything but the cpu integrated. Intel not buying Nvidia is the stupidest thing I have ever seen.
I would like to have fiber connections to each room in my house... that way I could have displays anywhere I liked (and keep the unsightly computer/video game/video equipment out of sight). I wouldn't worry about the form factor of my equipment, just the compactness of my displays and speakers (output devices).
Wireless just doesn't have the bandwidth for driving high-res displays.
VoIP and IM are interactive means of communication, where email is quite asynchronous. Of course you have to whitelist VoIP and IM, or else you could have to be online all the time.
I do the same thing. Hardware raid and software raid is just overkill. My wife's computer has a single 250GB drive, and my computer has a single 250GB drive... until we get close to filling them, I do auto-scheduled incremental and full backups to the other drive. I have a dvd burner that I will backup files to (mostly media files). Gigabit ethernet is very nice for full backups. We both use S3 suspend all day, so waking the other computer to do an incremental backup is simple (I setup both boxes for wake-on-magic-packet)... when finished the boxes return to S3 after 20 minutes of inactivity.
I really don't need high up-time or complete fault tolerance. Who does?
At least half off all power dissapated is by static leakage... even if you halt the clock, you only save half the power. That's why you have to remove the upper voltage rail to get all the savings. It's just like S3 suspend mode.
If single cores can have their clock halted and upper voltage rails removed... and still resume all previous threads, why can't both cores be suspended, and only one core resumed?
I am still waiting for Intel/AMD to announce a power saving feature of shutting down all but one of the multiple cores... if this is going to be a mobile chip, obviously you only want one core running while you are using battery power. It should be like that displacement-on-demand thing that newer cars and trucks have.
Medtronic's neurostimulator (for pain treatment) lasts at least 9 years. The Japanese government sponsored the development of a rechargable pacemaker, but found that IC and manufacturing advances made devices more than 5 years old obselete. Recently, advances in data storage and wireless distance/data transfer have given new features to mature treatments.
I want to see mixed VTs on a process. The "controller" core uses the high VTs, and additional cores use low-VTs that leak like a son-of-a-gun, but are held in S3 suspend and only used for peak processing. The key is to remove the power rails so that the low-VTs don't cost you when you aren't using them.
I beleive Cell uses this method, and throttles them with temperature to ensure that the cooling and power supply design is never exceeded with long heavy loads.
"That way you can carry half a dozen, dozen movies with you on trips, to your friends houses, wherever, and view them on TV."
So what happens when the movie is half over, and you need to get going? Do you leave the thing with your friends, or be a dick and take it with you?
If I ever bought one, I would just tell my buddies it's a old ipod. I don't want them to know that I keep a copy of "sisterhood of the traveling jeans" on it.
My knoppix CD and a knoppix-home-image (on the local HD) allows me to quickly image the drive over my home network. I do this whenever a large change has been made in windows. Before I go to bed I kick this off, and the script (on the home-image) shuts the computer down when it's done.
It requires more than one box... but it keeps me safe from any "soultion" that vendors offer (be they hardware or software vendors). Vendors will always have an angle or restrictions.
Knoppix with scripting IS the perfect and final imaging solution. Now making it easier is the remaining task.
It's a shame that the 970MP's two 1MB caches are not shared like the power4+'s cache is. A shared cache is great for single threaded performance and for sharing variables between threads (threads running on different cores).
Is shared cache a premium feature, maybe similar to power4+'s external L3 cache?
Cut-n-paste your sensitive logins and passwords one character at a time. You need to type-in the alphabet (upper and lowercase) and numbers into a different window. This is all the keylogger sees (that and cut-n-paste commands).
Hopefully nobody is looking at your screen remotely (and see the mouse movements)... anyone have a technique around that?
The data storage itself doesn't consume power, but the address_decode/read_mux is transistor based and comsumes static and dynamic power. The more storage you want to access, the larger the read_mux that is required to decode each unique address.
You are absolutely right that ROM storage does not consume any power... you just can't see the data without consuming power.
The threshold voltages for the process need to be low enough for the read_mux transistors to toggle quickly, but not too low that it leaks more power than you can spare.
It's nice to be the physically smallest rom, but what about power consumption (static and dynamic)? Also, what about access time (usually you have to balance power consumption and access time... you get one at the expense of the other).
Flash die can be die-stacked to increase the storage capacity... using die from the same wafer. If you want to die-stack rom die, all the die will be from different wafers (extra mask layers are required for each unique rom die). Flash wafers will be cheaper to produce because they will have a much greater volume.
I'm sorry but flash looks like a much more flexibile and cheaper solution (being relatively larger in size).
I am very impressed with my creative muvo tx mp3 player. It's a pen (flash) drive with an mp3 player built in. If you can mount a pen drive, you can load this player... (no usb cables needed)
I believe the ipod shuffle has a similar interface, but I like my LCD display, FM radio, voice and radio recording. Shuffle might require itunes software, where muvo doesn't require any software once mounted.
You are absolutely right... HTTPS is the extra security layer. Just all my other internet communications are subject to snooping to everyone and his dog.
Now I will be able to read all my neighbour's non-encrypted emails (and learn his slashdot username;-) )
The reader software becomes the equivalent of a restricted browser... no printing, no copy-n-paste, no save-as. It's been done before, but you can't blame them from trying to make some money of this scheme.
It's quicker to query a server for a key, than to re-download the content every time (they could cache the content, but it would have to be encrypted there).
Updates to the document would be nice... as long as they don't remove content.
amen. If there were atteched photos, they probably will not be saved in their native format. I had to laugh when I saw the dumbass slashdot moderator mod the parent up.
The same dumbass modded-up the spotlight comment too. Who wants 10,000 email pdfs to sort through? Or one pdf with 10,000 pages?
Mozilla maintains indexes of the sender and subject lines for lightning fast searches (and then you can go to body text if you can't find it with those fields). You don't even have to use mozilla to search the indexes, you can just grep them if you want.
Slashdot Mirror
The one-chip PC is a falicy... if the integration of all blocks produced higher performance, then you can charge more for that integration. The slower blocks (ATA, USB, PCI) don't perform significantly better when integrated, thus consumers don't value integration (board manufactures will, but only by a fixed amount).
If you integrated the gpu/ethernet/sdram/cpu, then all the high-speed blocks can talk with each other at die level speeds. It's clear as day that performance will significantly increase.
The southbridge IO (ATA USB PCI) requires many pins, and moving them to a single chip doesn't save enough money at board level to justify the added expense at die level. The power cost of running a bus to the southbridge is tiny compared to the total system budget.
It's quite clear that the AMD-ATI merger was to aquire the IP and expertise necessary to integrate gpu core(s) on the same die as cpu core(s). Nvidia does not have to actually market a design, but rather patent some key concepts, and this could provide revenue or protection.
I would very much doubt that they could compete with AMD and Intel who have already patented many x86 cpu concepts.
It's a shame that Intel has decided not to buy nvidia, and go it alone with it's own design staff.
I was under the impression that AMD's "native" quad-core design would be very hardware virtualization friendly. The fact that these chips have a very modular memory design and are not used in MCMs (multi-chip-modules), should allow for the simplest possible virtualization configuration.
I would really like to see desktop virtualization that allows fast access to my legacy windows apps (rather than my present approach of vncing to a windows box powered on all the time). The old windows box only costs electricity, but if any hardware on it fails, I really don't want to invest more money in a make-shift solution.
Smart consumers always stay away from a 1st generation product. I would gladly give-up some features for peace of mind. The more the cost of the equipment is, the more you want to stay with a proven vendor.
It's possible for any generation to be plagued with mistakes, so it's best not to be the early adopter.
80 cores on a chip and optical interconnects sounds great...but I wish they would talk more about the end application goals (i.e. a system that does 1080p ray-tracing and has 100% speech recognition). It's great that they are pushing the design limits as they are, but without clear vision of how the technology is to be used, it's likely that it will miss the mark.
I was hoping to hear about a single die with cpu/northbridge/southbridge/gpu all integrated (and for mobile use)... that would certainly turn the computer market on it's head. Nvidia knows this already, and has everything but the cpu integrated. Intel not buying Nvidia is the stupidest thing I have ever seen.
I would like to have fiber connections to each room in my house... that way I could have displays anywhere I liked (and keep the unsightly computer/video game/video equipment out of sight). I wouldn't worry about the form factor of my equipment, just the compactness of my displays and speakers (output devices).
Wireless just doesn't have the bandwidth for driving high-res displays.
VoIP and IM are interactive means of communication, where email is quite asynchronous. Of course you have to whitelist VoIP and IM, or else you could have to be online all the time.
I do the same thing. Hardware raid and software raid is just overkill. My wife's computer has a single 250GB drive, and my computer has a single 250GB drive... until we get close to filling them, I do auto-scheduled incremental and full backups to the other drive. I have a dvd burner that I will backup files to (mostly media files). Gigabit ethernet is very nice for full backups. We both use S3 suspend all day, so waking the other computer to do an incremental backup is simple (I setup both boxes for wake-on-magic-packet)... when finished the boxes return to S3 after 20 minutes of inactivity.
I really don't need high up-time or complete fault tolerance. Who does?
At least half off all power dissapated is by static leakage... even if you halt the clock, you only save half the power. That's why you have to remove the upper voltage rail to get all the savings. It's just like S3 suspend mode.
If single cores can have their clock halted and upper voltage rails removed... and still resume all previous threads, why can't both cores be suspended, and only one core resumed?
I should really take a patent out on this.
I am still waiting for Intel/AMD to announce a power saving feature of shutting down all but one of the multiple cores... if this is going to be a mobile chip, obviously you only want one core running while you are using battery power. It should be like that displacement-on-demand thing that newer cars and trucks have.
Medtronic's neurostimulator (for pain treatment) lasts at least 9 years. The Japanese government sponsored the development of a rechargable pacemaker, but found that IC and manufacturing advances made devices more than 5 years old obselete. Recently, advances in data storage and wireless distance/data transfer have given new features to mature treatments.
m l
http://www.medtronic.com/neuro/restore/noFlash.ht
I want to see mixed VTs on a process. The "controller" core uses the high VTs, and additional cores use low-VTs that leak like a son-of-a-gun, but are held in S3 suspend and only used for peak processing. The key is to remove the power rails so that the low-VTs don't cost you when you aren't using them.
I beleive Cell uses this method, and throttles them with temperature to ensure that the cooling and power supply design is never exceeded with long heavy loads.
I'm sure that BlueGene will keep it's customers happy, but the Japanese have similar if not more resources to equal anything out of IBM.
Unless IBM is using passive optical interconnects or molecular circuits... revolutionary is just not the right word.
It's just CMOS ANDs, ORs, NOTs and Flops. When it's something else, let us know.
"That way you can carry half a dozen, dozen movies with you on trips, to your friends houses, wherever, and view them on TV."
So what happens when the movie is half over, and you need to get going? Do you leave the thing with your friends, or be a dick and take it with you?
If I ever bought one, I would just tell my buddies it's a old ipod. I don't want them to know that I keep a copy of "sisterhood of the traveling jeans" on it.
My knoppix CD and a knoppix-home-image (on the local HD) allows me to quickly image the drive over my home network. I do this whenever a large change has been made in windows. Before I go to bed I kick this off, and the script (on the home-image) shuts the computer down when it's done.
It requires more than one box... but it keeps me safe from any "soultion" that vendors offer (be they hardware or software vendors). Vendors will always have an angle or restrictions.
Knoppix with scripting IS the perfect and final imaging solution. Now making it easier is the remaining task.
It's a shame that the 970MP's two 1MB caches are not shared like the power4+'s cache is. A shared cache is great for single threaded performance and for sharing variables between threads (threads running on different cores).
Is shared cache a premium feature, maybe similar to power4+'s external L3 cache?
Cut-n-paste your sensitive logins and passwords one character at a time. You need to type-in the alphabet (upper and lowercase) and numbers into a different window. This is all the keylogger sees (that and cut-n-paste commands).
Hopefully nobody is looking at your screen remotely (and see the mouse movements)... anyone have a technique around that?
"ROM is not classified as "static" or "dynamic.""
The data storage itself doesn't consume power, but the address_decode/read_mux is transistor based and comsumes static and dynamic power. The more storage you want to access, the larger the read_mux that is required to decode each unique address.
You are absolutely right that ROM storage does not consume any power... you just can't see the data without consuming power.
The threshold voltages for the process need to be low enough for the read_mux transistors to toggle quickly, but not too low that it leaks more power than you can spare.
It's nice to be the physically smallest rom, but what about power consumption (static and dynamic)? Also, what about access time (usually you have to balance power consumption and access time... you get one at the expense of the other).
Flash die can be die-stacked to increase the storage capacity... using die from the same wafer. If you want to die-stack rom die, all the die will be from different wafers (extra mask layers are required for each unique rom die). Flash wafers will be cheaper to produce because they will have a much greater volume.
I'm sorry but flash looks like a much more flexibile and cheaper solution (being relatively larger in size).
Intel is the best architecture for a single cpu with a butt-load of DIMM slots... all the memory controllers are in the non-integrated northbridge.
Good recommendation.
I am very impressed with my creative muvo tx mp3 player. It's a pen (flash) drive with an mp3 player built in. If you can mount a pen drive, you can load this player... (no usb cables needed)
I believe the ipod shuffle has a similar interface, but I like my LCD display, FM radio, voice and radio recording. Shuffle might require itunes software, where muvo doesn't require any software once mounted.
You are absolutely right... HTTPS is the extra security layer. Just all my other internet communications are subject to snooping to everyone and his dog.
;-) )
Now I will be able to read all my neighbour's non-encrypted emails (and learn his slashdot username
When the FBI can break wireless encryption in 3 minutes, I'll do my online banking over cable.
;-)
I like my location (trafic, shopping, schools), it would take more than $300/year to get me to move.
AT least my cable is going to get cheaper
The reader software becomes the equivalent of a restricted browser... no printing, no copy-n-paste, no save-as. It's been done before, but you can't blame them from trying to make some money of this scheme.
It's quicker to query a server for a key, than to re-download the content every time (they could cache the content, but it would have to be encrypted there).
Updates to the document would be nice... as long as they don't remove content.
amen. If there were atteched photos, they probably will not be saved in their native format. I had to laugh when I saw the dumbass slashdot moderator mod the parent up.
The same dumbass modded-up the spotlight comment too. Who wants 10,000 email pdfs to sort through? Or one pdf with 10,000 pages?
Mozilla maintains indexes of the sender and subject lines for lightning fast searches (and then you can go to body text if you can't find it with those fields). You don't even have to use mozilla to search the indexes, you can just grep them if you want.