10.2GHz Intel "Nehalem" with a 1200MHz FSB by 2005

[ASIMOS]

There is no way to have in 2005 10,2Ghz x86 based proccesors.In order to feed the CPU with Data You need bandwidth, in order to have bandwidth you need memory speed.I don't want to be too technical since nobody like it from the responce i had in my previous to post (R350 & RV350 part 1 & part 2).Here is a table to clarify the possibilities Intel have from 2003 H2 0,9-micron to 2005 H2 0,65-micron:
YEAR MICRON BUS SPEED MEMORY MULTIPLIER MIN CPU SPEED MULTIPLIER MAX CPU SPEED
2003 H2 0,9 800 2X400DDRI 3,75 3GHZ 4,75 3,8GHZ
2004 H1 0,9 800 2X400DDRI 5 4GHZ 6 4,8GHZ
2004 H2 0,9 1066 2X533DDRII 3,75 4GHZ 4,75 5,06GHZ
2005 H1 0,65 1066 2X533DDRII 5 5,33GHZ 6 6,4GHZ
2005 H2 0,65 1600 2X800DDRII 3,75 6GHZ 4,75 7,6GHZ

So in no way you should expect anything higher than 7,6GHZ before 2006.Surely everybody knows that mass production DDRII modules is delayed until 2004 H2, so intel is forced to stay at 800Mhz bus for a year.Ofcource you should expect that prescott to move to 1600 bus speed a quarter before the next Intel chip make an appearance (2005 H1).

 

[DX2Player]

Well look at the RDRAM Roadmap by Samsung, if they can make 64bit RDRAM PC1200 by the end of the year thats 9.6GB/sec bandwith
Dont know if it will be out that fast but probably by begining of 2004
RDRAM Roadmap

 

[teutonicknight]

Originally posted by: BFG10K
10.2 GHz? Just what we need for the likes of UT2003. Bring it on, I can't wait!

Ha, have fun in that game! I'll be off playing UT2005 !

 

[Stunt]

lol you are a dumb@$$
bug me about a petty a instead of an e
well here is so more "usefull" info
its spelled USEFUL, so go screw yourself

 

[paralazarguer]

Stunt, you're exactly the type of person that I don't like to see anywhere. You start off by calling someone a dumba$$ for no reason and then when he puts out a well-reasoned response you attack his english skills.

huh, u're last post made no sense
no pass english?

When, in fact, you are three times worse. Just because I hate idiots like you, I point out that yours are not high-school level either and, like any true hypocrite, you can't take your own heat and turn on me.
You're not going anywhere in life like that kid.

 

[Stunt]

man shutup
you are bashing my misspelling of grammar
you brought that comment on yourself
"and please explain how because it has been "18 months" that "moore's law" is false oh great one. "

you try and make sense of that
and i like how you tried to quantify how much worse i am at spelling than he is
at least you can understand me
for example i wrote gammer and you knew it was grammar
wow! thats like "3 times" worse than the other guy

go back to your hole
stop twisting crap around...
i'll get further in life than you will
this is such a pointless arguement....
who the hell are you to criticize me?

 

[WyteWatt]

If it is true i love it but i am not holding my breath.

 

[CallTheFBI]

Uh, if they already know all this why don't they have the 10GHz CPUs now?

 

[Wingznut]

I don't mean to hop on the "bash on Stunt" bandwagon, but...

A 3rd year EE major with such a horrible attitude, poor grammar skills, and not knowing what Moore's Law is? Not to mention that it seems as though you hardly understand the evolution of CPU/FSB speeds...

I have my doubts.... I personally don't know anybody in this field with the above issues.

Sorry.

 

[Wingznut]

Originally posted by: CallTheFBI
Uh, if they already know all this why don't they have the 10GHz CPUs now?

If that's a very serious question, send me a msg (ICQ/MSN/AIM... whatever) and we'll hook up sometime for a conversation.

 

[Stunt]

ok
so im guessing that you think the fsb is good for a processor of that speed?
i was just comparing it to previous versions
i really dont have that bad of an attitude
heh
my grammar isnt that bad
i only spelled gammar wrong
i dont use formal writing skills on a message board but thats my casual style
im not about to criticize the writing skills of others on the board
your right about the history of the semiconductor industry
i havent learned that, im more into the technical side of things
besides
im not even heading into that field
im in power option so really, why would i care about that side of things?
this is just a side interest
for example you can be a mech eng taking aerospace and not know much about automotive...
especially trivial stuff
im sorry your company cant get the fsb up as high as it should be
maybe amd can

 

[MadRat]

Originally posted by: ASIMOS
There is no way to have in 2005 10,2Ghz x86 based proccesors.In order to feed the CPU with Data You need bandwidth, in order to have bandwidth you need memory speed.I don't want to be too technical since nobody like it from the responce i had in my previous to post (R350 & RV350 part 1 & part 2).Here is a table to clarify the possibilities Intel have from 2003 H2 0,9-micron to 2005 H2 0,65-micron:
YEAR MICRON BUS SPEED MEMORY MULTIPLIER MIN CPU SPEED MULTIPLIER MAX CPU SPEED
2003 H2 0,9 800 2X400DDRI 3,75 3GHZ 4,75 3,8GHZ
2004 H1 0,9 800 2X400DDRI 5 4GHZ 6 4,8GHZ
2004 H2 0,9 1066 2X533DDRII 3,75 4GHZ 4,75 5,06GHZ
2005 H1 0,65 1066 2X533DDRII 5 5,33GHZ 6 6,4GHZ
2005 H2 0,65 1600 2X800DDRII 3,75 6GHZ 4,75 7,6GHZ

So in no way you should expect anything higher than 7,6GHZ before 2006.Surely everybody knows that mass production DDRII modules is delayed until 2004 H2, so intel is forced to stay at 800Mhz bus for a year.Ofcource you should expect that prescott to move to 1600 bus speed a quarter before the next Intel chip make an appearance (2005 H1).

10GHz may be plausible, but in no way should you tie in the speed of the CPU to the memory technology supporting it. Current 533MHz fsb for the P4 is still limited by a true speed of only 133MHz. The 800MHz fsb will still presumably be limited to a true speed of 200MHz. The bandwidth of the memory doesn't necessarily have much to do with the latency of the memory, meaning the CPU-to-memory gap is inconsequential to this equation.

By the time they reach 65 nanometer technology in memory technology we should have quad-pumped memory interfaces running upwards of 500MHz or more.

 

[Stunt]

giving a 10.2ghz a 1.2ghz fsb
is like a p4 3.06 with a 366 fsb or 90mhz not quad pumped
or an athlon (barton) 2500+ with a 235 fsb or 118 not double pumped

sorry, but it just doesnt seem right
i expected more
oh well
im not going to change anything
intel can do what they like
just hope they dont take a huge hit in proformance
and they will loose a lot of oc'ers

assuming they stay with a quad pumped bus, they need a 300mhz bus
and a 34X multiplier
thats huge
for every 4mhz of fsb you get 34 mhz core
thats bad for oc yields
oh well

 

[Bluga]

Originally posted by: Wingznut
For those of you who don't see the need for faster processors, I ask you to have some vision...

Think about having perfect voice recognition... Perfect character and pattern recognition.
Think about having a fully rendered 3D environment to shop in. Places to visit in full 3d. How about a completely 3-dimensional operating system?

I'm not a creative person, and I will admit that I don't have a lot of vision. I am amazed at how so many of the people around me simply have no barriers... They don't constrain themselves with limits.

Sure, there's no use for 2000fps in Q3. But, think bigger!

Ya i'm sure Intel pentium can speed up internet, LOL.

 

[ASIMOS]

___________________________________________________________________________________
Originally posted by: MadRat
10GHz may be plausible, but in no way should you tie in the speed of the CPU to the memory technology supporting it. Current 533MHz fsb for the P4 is still limited by a true speed of only 133MHz. The 800MHz fsb will still presumably be limited to a true speed of 200MHz. The bandwidth of the memory doesn't necessarily have much to do with the latency of the memory, meaning the CPU-to-memory gap is inconsequential to this equation.

By the time they reach 65 nanometer technology in memory technology we should have quad-pumped memory interfaces running upwards of 500MHz or more.
___________________________________________________________________________________
CPU speed is tied with memory bandwidth and memory technology from the nature of computer technolohy itself.Yes everybody knows that Intel is using a quad pumped bus, but this doesn't imply 4x speed increase in proccesor speed since is extremely complicated design and everybody knows this if remember the 1GHZ PIII/133MHZ versus 1,3 PIV/400Mhz results when we compared these two differently architectured CPU.In fact the quad pumped Intel FSB is so complicated that I don't expect to launch in the three next years a more complicated aproach.there are 3 reasons that forced Intel to made this aproach:
1)SDRAM memory speed couldn't follow P4 needs for memory bandwidth
2)Intel Marketing department
3)Intel Marketing department
VIA has the rights for QDR & QDRII, but Intel is not going to invest in this technology (the reasons are obvious).
If you mean quad memory interface (like the dual ones Nforce2 or Granite bay) forget.If someone motherboard manufacturer would listen to you, you could be in danger.
Please believe me, the numbers I gave are correct.

 

[MadRat]

The memory could run at 100MHz and it will still outpace the hard drive speed.

My point is that the internal caching is more relevant than memory speed.

 

[Bovinicus]

I can't wait for AI to processed without killing framerates in RTS/Realistic FPS games. Maybe 10GHz will finally take care of that problem...

 

[MadRat]

Dual-processors would do it if they wrote the program for SMP.

 

[zsouthboy]

Originally posted by: tkdkid
Specs of the majority of machines available at Best Buy in 2005:

10.2 GHz Intel Processor
128MB PC133 SDRAM
20GB HD
Integrated graphics/sound
2 PCI slots
USB 2.0
still NO AGP SLOT
15 inch monitor with built in crappy speakers (or free speakers worth about $5)
free $40 printer included (yippee!)
MS "carpal tunnel" edition keyboard/mouse
Windows XP 2005 Home with tons of useless functions/services enabled by default that make the system just crawl.

Total price: $3000

and people will snap them up like hotcakes

SO FREAKING TRUE! LOL

 

[Sunner]

Originally posted by: MadRat
Dual-processors would do it if they wrote the program for SMP.

Not all programs are suitable for SMP systems.

Since games is an obvious target for high end CPU's, let's take Quake III as an example.
I remember reading John Carmack's .plan file about his effort to make the most out of SMP system, and in the end it seems like he came to the conclusion that in most cases, the performance gains wouldn't be too great.
And not to just on the "Carmack is God" bandwagon, but he is no doubt one of the most talented programmers out there in his field, and frankly, if he can't figure out how to do it, I don't think many, if any other programmers are going to.

 

[MadRat]

For every "legend" there is a reality. Carmack is one to rely on his judgement for gaming graphics programming. I wouldn't rely on his judgement for SMP programming as it is not relative to video. I highly doubt one person can be a jack of all trades in programming and be elite in everything at the same time. If that was so common then everything nowadays would be written in assembly. Plus those comments are pretty dated, with SMP in M/S products greatly improved over the NT 4.0 days.

 

[BoberFett]

Originally posted by: Sunner

Originally posted by: MadRat
Dual-processors would do it if they wrote the program for SMP.

Not all programs are suitable for SMP systems.

Since games is an obvious target for high end CPU's, let's take Quake III as an example.
I remember reading John Carmack's .plan file about his effort to make the most out of SMP system, and in the end it seems like he came to the conclusion that in most cases, the performance gains wouldn't be too great.
And not to just on the "Carmack is God" bandwagon, but he is no doubt one of the most talented programmers out there in his field, and frankly, if he can't figure out how to do it, I don't think many, if any other programmers are going to.

Carmack is a graphics programmer, not necessarily an AI programmer. The graphics do have to be done in some sort of order, there's not a whole lot that could be done that could speed it up beyond pure brute force Mhz. AI however could be split up. Perhaps a separate thread for every opponent in a game.

 

[rommelrommel]

Originally posted by: magomago
Dang....What will we be REALLY using it for? Games better not get realistic (unless its CS ) because I play games to get away from reality.

HRm, makes the idea of a ps3 seem kinda moot. Then again they did say they were going to try to make it a thousand more times powerful than the ps2

Yeah, and CS: Condition Zero should almost be out by then too

 

[ST4RCUTTER]

"The contents of the note appear to reveal future plans for future Intel desktop processors right up until 2005.
By then, according to the note, Intel will be able to deliver 10.20GHz desktop CPUs codenamed "Nehalem" and produced using 65 nanometer technology.

If Intel manages to migrate away from the 90 nanometer technology it will introduce towards the end of this year, by then the "Prescott" core will deliver at least 5.20GHz using the 800MHz system bus.

The immediate successor to Prescott after it tops out at 5.20GHz will be the "Tejas" core, also produced on a 90 nanometer process and delivering 5.60GHz using a 1066MHz system bus. That's slated to start appearing towards the end of 2004.

Tejas will increase in steady increments which appear to be 6GHz, 6.40GHz, 6.80GHz, 7.20GHz, 7.60GHz, 7GHz, 8.40GHz, 8.80GHz and topping out at 9.20GHz.

The first Nehalem is supposed to appear at 9.60GHz before Intel succeeds in its goal to produce a 10GHz+ chip, the Nehalem, and using a 1200MHz front side bus." --Vespasian



Ten gig ehh...that would be interesting. Although I'm less interested in how fast they can run the core clock compared to how they're going to pull it off without having the CPU's producing as much heat as a fuel rod undergoing fission. Leakage current at the 65nm node is said to be horrific. The idea of "sleeper transistors" doesn't seem too novel to me (when comparing it to clock gating) although it should be more effective. Anyone know if the EDA has even come up with the tools to implement this process??? Wingz? When are we going to see the first BBUL package? I think it's clear to most in the industry that process shrinks alone will never reach 10Ghz...physics just keeps getting in the way. New materials, new fabbing methods...so much more needs to be done. It'll be fun to watch them try though.

 

[krackato]

10.2ghz chip - 650 dollars
2 terabyte harddrive - 200 dollars
Geforce FX 4 - 399 dollars
Telling your son about YOUR first computer - priceless

Somethings, money can't buy. For everything else there's Mastercard.