Next Intel Socket???

[mrmez]

Just looking @ this pic a few days ago...

And i cant help thinking the next Intel CPU socket (8xx? 9xx?) MUST be on its way.
Possibly in the next 6 months rumours of a new socket will appear, maybe with a new NForce killer mobo?

Anyway... just thinking out loud (so to speak)
Feel free to whore away :twisted:

 

[Groveling_Wyrm]

I would guestimate another 6 months, unless they want to get it out before the christmas rush, based off of previous practices....

 

[Action_Man]

I cant see a socket change for quite some time. New chipsets is another thing.

 

[mesarectifier]

Well Kentsfield (the next 'big thing') is LGA775, but I sort of see your point - they've used 775 for P4, PD, Celeron and now Core 2.

Only time will tell, but I don't think they'll jeapordise their newly obtained performance lead over AMD by changing sockets already.

 

[Julian33]

Is there really any need for a socket change? AMD seems to have to change more frequently because of the IMC, but even when DDR3 comes along, the fact that the memory controller is off chip means that there's no reason why LGA775 won't still be up to the task.

The only reason I could see Intel changing socket is if they end up reaching the limits of the FSB and end up having to introduce dual busses on the desktop line of chips

 

[slicessoul]

i think it would be better if it's not a socket anymore. As to mention that not a few people has its processor's pins are bent. I can say Socket + Pin = Fragile.

Slot could be a good idea, just like the old timers....

Or maybe it will be better if they make a board only for processor and the mainboard only for put the VGA cards, soundcards, and other peripherals. Then between processor board to mainboard are connected with fiber optics...or i don't know...just an idea.

 

[kmjohnso]

I hope they don't go back to that... Those where such a pain to cool. I don't think intel/amd makes decsions on the poor experience of a few unwieldly enthusaists.

 

[ranger90]

I did read.. I think it was here on Tom's actually... that 1033 is getting near the max of the FSB.

Plus, with the Memory controller being on-chip, that Quad-core was going to have memory access issues...

AMD seems to have the advantage when going to multiple core because the memory controller is off chip.

Makes me think Intel is going to have to rethink their design philosophy again. They already did when they stepped back from their "Make the processor faster is the best way for more performance." Now they've put in Hyperthreading, dual core, and more "pre-fetching" etc, and slowed the chips back down again. More efficiency. Me thinks they my change their memory controller to be off-chip.

Also.. If they're planning on dropping IDE disk controllers, they might change chipsets radically at the same time.

P.S. Why is there still a Parellel Port on Mobos???? Like.. does anybody still make a parellel controlled.... anything?? I say, drop the Parellel and Serial ports, drop IDE. Might as well..

 

[Sonic_Reducer]

Also.. If they're planning on dropping IDE disk controllers, they might change chipsets radically at the same time.

P.S. Why is there still a Parellel Port on Mobos???? Like.. does anybody still make a parellel controlled.... anything?? I say, drop the Parellel and Serial ports, drop IDE. Might as well..

Parellel ports and serial ports make mobos backwards compatibel with old big printers some companies have, i do agree with losing them for home market, but for corporate it's sometimes useful to have them , on the ide side it's time for change, but the companies that produce the optical drives have to go alog by putting the sata interface on the drives, there are a few but little solutons, besides that you have to fill up the board with sata connectors because imagine that 1 for dvd-rom or dvd-rw other one for the HD dvd-reader and recorder and another few for hard drives , at least 4 it's needed and may be little for enthusiats with raids and alot of hard drives, and how about the 3 1/2 connector ?? it remaisn the same , maube it would need some changes to, despite being almos dead some times it's useful. on the socket thing i find this new socket from intel quite nice it's easy to assemble but if you do bend a pin on the board it's hard to repair , it was easier to repair a pin on skt 478 times, but its evolution baby...

just my 4 cents

 

[slicessoul]

Well if you see, what is the fastest processor's speed today ? 3.2 gig, 3.4 gig ? and what's the speed of new processors ? 1.8 gig, 2.9 Gig but they put more core. I just see that we are going back.
Why they don't make a more faster processor and with more cores ?

It's just look like : A Sports car who can run 350 km/h with 2 people inside and a Bus who can run 120 km/h with 35 people inside. Why don't we make a bus who can run 350 km/h with 35 people inside ? so what do you think ? lets make a train...he..he..he.. :lol: i'm out...

i wish i own intel, i'll give you guys a free processor each time Moore's Law is realized.

 

[kmjohnso]

I have a train at work... 95-node cluster. It's hard to drive, won't go off the tracks, hard to stop, and when it crashs a thousand children are killed.

 

[slicessoul]

Well at least when you make a fast train, you don't forget to make the brakes and the monitoring program... :lol: formula one is a fast car right ? 350 to 60 km/h in 4 seconds.

CPu clocks may be dropping while cores increase but performance is makeing big increeses. They can make a doul core CPU that clocks at 3ghz but the heat and power cosumption would be outragious. The whole CPU race isn't about clock speed any more. As AMD shows with there lower clock speed CPU;s out performing Intels equivalent CPUs in the single core arena.

This is correct. So, what about the Moore's Law then ? or maybe they will start to reducing the process production ? lets say from 0.65nm to 0.11nm or maybe lesser and lesser or they will find another material which is will be more higher speed can be achieved and less power consumption ? that would be a good idea too.

 

[306maxi]

AMD seems to have the advantage when going to multiple core because the memory controller is off chip.

You might like to check that.

 

[exit2dos]

Intel has stated that it wants to release a new platform every two years, so I doubt they'll change the socket until the next platform upgrade.

But, as Action Man said, chipsets are another thing.

 

[NMDante]

Intel has stated that it wants to release a new platform every two years, so I doubt they'll change the socket until the next platform upgrade.

But, as Action Man said, chipsets are another thing.

It would be cool if a chipset could just be popped off and replaced, but alot of chipset changes also include voltage regulator changes and other things, like number of capacitors, resistors, or even a few in-line diodes.

But man, I would love to just take the northbridge cooling device out and pop a new chipset in, and still use the same board. That would be cool.

 

[captaincharisma]

Well if you see, what is the fastest processor's speed today ? 3.2 gig, 3.4 gig ? and what's the speed of new processors ? 1.8 gig, 2.9 Gig but they put more core. I just see that we are going back.
Why they don't make a more faster processor and with more cores ?

It's just look like : A Sports car who can run 350 km/h with 2 people inside and a Bus who can run 120 km/h with 35 people inside. Why don't we make a bus who can run 350 km/h with 35 people inside ? so what do you think ? lets make a train...he..he..he.. :lol: i'm out...

i wish i own intel, i'll give you guys a free processor each time Moore's Law is realized.

yea im olf fashoined so i haven't broken away from that logic yet. its hard for me to get a new CPU these days because I have a P42.4 GHZ CPU and almost all retail CPUs now no matter what there called are still at 2.4 GHZ now. now I wish there was a way to see how much faster the newer CPU's will be at 2.4 compaired to my current one.

 

[MatTheMurdera]

I did read.. I think it was here on Tom's actually... that 1033 is getting near the max of the FSB.

Plus, with the Memory controller being on-chip, that Quad-core was going to have memory access issues...

AMD seems to have the advantage when going to multiple core because the memory controller is off chip.

Makes me think Intel is going to have to rethink their design philosophy again. They already did when they stepped back from their "Make the processor faster is the best way for more performance." Now they've put in Hyperthreading, dual core, and more "pre-fetching" etc, and slowed the chips back down again. More efficiency. Me thinks they my change their memory controller to be off-chip.

Also.. If they're planning on dropping IDE disk controllers, they might change chipsets radically at the same time.

P.S. Why is there still a Parellel Port on Mobos???? Like.. does anybody still make a parellel controlled.... anything?? I say, drop the Parellel and Serial ports, drop IDE. Might as well..

I can tell you right now, FSB of 1066 is NOT near the max. In fact Ive gotten 1500 FSB on my first OC and I havnt even tried going above that!

 

[exit2dos]

Intel has stated that it wants to release a new platform every two years, so I doubt they'll change the socket until the next platform upgrade.

But, as Action Man said, chipsets are another thing.

It would be cool if a chipset could just be popped off and replaced, but alot of chipset changes also include voltage regulator changes and other things, like number of capacitors, resistors, or even a few in-line diodes.

But man, I would love to just take the northbridge cooling device out and pop a new chipset in, and still use the same board. That would be cool.

Would be a good idea. Unfortunately (or maybe fortunately) I look for Intel to use an IMC next time out.

I wonder, when they shrink to 32nm, if they could put all the north and southbridge logic on the CPU. This way the motherboard could just be a blank PCB with card and memory slots. You could upgrade your whole system with just the CPU.

 

[mesarectifier]

I wonder, when they shrink to 32nm, if they could put all the north and southbridge logic on the CPU. This way the motherboard could just be a blank PCB with card and memory slots. You could upgrade your whole system with just the CPU.

Great idea. Would kill the motherboard industry overnight, but great idea nonetheless.

 

[ervinshiznit]

Plus, with the Memory controller being on-chip, that Quad-core was going to have memory access issues...

AMD seems to have the advantage when going to multiple core because the memory controller is off chip.

u got it flipped, amd is memory controller on chip and intel is off chip

 

[exit2dos]

I wonder, when they shrink to 32nm, if they could put all the north and southbridge logic on the CPU. This way the motherboard could just be a blank PCB with card and memory slots. You could upgrade your whole system with just the CPU.

Great idea. Would kill the motherboard industry overnight, but great idea nonetheless.

With AMD now making their own chipsets, and Intel has always done so - I don't think either would care if the smaller mobo makers were forced out.

 

[weilin]

ya know, i would rather have a socket change whenever there is a substantial upgrade. Take core 2 duo for instance, if they were to have made it a socket 774 cpu we could eliminte 99% of the will this work threads. Think about it, its not like the chipsets before will support Core 2 so a socket change wouldn't affect people trying to upgrade at all. they need to get a new board for a newer chipset after all.

Just my 2 cents.

 

[exit2dos]

ya know, i would rather have a socket change whenever there is a substantial upgrade. Take core 2 duo for instance, if they were to have made it a socket 774 cpu we could eliminte 99% of the will this work threads. Think about it, its not like the chipsets before will support Core 2 so a socket change wouldn't affect people trying to upgrade at all. they need to get a new board for a newer chipset after all.

Just my 2 cents.

Good point. I also wish they would change the chip name when they change the chipset requirements or sockets, or at least give a designation letter.
"This motherboard will take Pentium 4Da through 4Dc", would be a lot less confusing than for the consumer to try to figure out if he has a Northwood, Prescott, Wiliamette, etc.

 

[weilin]

would also help in the upgrading process. I love AMD's design for 754 and 939. Pratically all 754 chips fit in any 754 board and any 939 chips work in any 939 board. It makes IT's job so much easier upgrading stuff. (socket A did have some issues though.. kinda like 775... so confusing esp. if there's no manual for the old board.)

 

[sepuko]

CPu clocks may be dropping while cores increase but performance is makeing big increeses. They can make a doul core CPU that clocks at 3ghz but the heat and power cosumption would be outragious. The whole CPU race isn't about clock speed any more. As AMD shows with there lower clock speed CPU;s out performing Intels equivalent CPUs in the single core arena.

You shure need to improve your spelling.
There might be a new socket for the mobile CPUs. The new 800MHz FSB of the mobile Core 2 Duo might require one. Looking at the 771 pin socket for the server CPUs that have 1333 MHz FSB i don't think Intel is going to introduce any new socket soon for the mass comsumer. I would expect the appearance of DDR3 compatible motherboards(and DDR3 itself in volume) before any new socket for intel CPUs. The same thing for AMD i guess. Only speculations...