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

[bsr]

why is clock speed even relevent ?? Isnt amd's new cpu going to have lower clock speed and still beat P4 3ghz ??

 

[dullard]

Originally posted by: nowayout99
Companies have been touting the voice recognition horn and all its glory since pre-1GHz. In fact I remember an interview with an Intel rep where he tried to sell voice recognition hard as a new grand possibility on their new $1000 1GHz CPU. At 3GHz, voice rec. still hasn't made much of a dent.

You missed the fine details of them touting the voice recognition horn back then. Here is what they said (paraphrased of course):
"Our new 1GHz CPU will let you begin to attempt voice recognition. Of course we realize for it to work well you need a CPU the speed of a 3.5 GHz PIII minimum, but don't let that stop you since you can still attempt it now and hope it barely works enough for you."
Notice that 3.5 GHz PIII part? We still don't have processors that speed. Thus we still don't have good voice recognition. In my opinion it will realistically take a 4.0 GHz P4 minimum though. And those are minimum numbers to get a pretty accurate voice recognition. For perfect voice recognition you need a 5 GHz or faster P4.

 

[Confused]

Originally posted by: dullard

Originally posted by: nowayout99
Companies have been touting the voice recognition horn and all its glory since pre-1GHz. In fact I remember an interview with an Intel rep where he tried to sell voice recognition hard as a new grand possibility on their new $1000 1GHz CPU. At 3GHz, voice rec. still hasn't made much of a dent.

You missed the fine details of them touting the voice recognition horn back then. Here is what they said (paraphrased of course):
"Our new 1GHz CPU will let you begin to attempt voice recognition. Of course we realize for it to work well you need a CPU the speed of a 3.5 GHz PIII minimum, but don't let that stop you since you can still attempt it now and hope it barely works enough for you."
Notice that 3.5 GHz PIII part? We still don't have processors that speed. Thus we still don't have good voice recognition. In my opinion it will realistically take a 4.0 GHz P4 minimum though. And those are minimum numbers to get a pretty accurate voice recognition. For perfect voice recognition you need a 5 GHz or faster P4.

Uhm, I'd take a 3.5GHz P3 over a 5GHz P4 any day


And for all those who can't think of what to do with 10GHz+, think compression (movies, files, music), image processing, distributed computing will be done even quicker than it can now......the list goes on



Confused

 

[DX2Player]

So slowly my but! Sit back for a second and compare, oh I dont know, the last 100 years to say all time before that. Than explain how slow we are going.

 

[Ruckas]

I love how AMD chips are clocked lower than the intel chips. But get the same benchmarks give or take. I could go read about how this is accomplished. But I'll just asume that AMD trys to make chips cheap, and perform well. While Intel uses the latest technology ALL the time. Without fine tuning the chips they already have. That's my opinion.

Now obviously Intel will win in a race to achieve 5ghz. However, I have 50 bucks cash says that the intel 5ghz cpu will cost 3-400 more dollars than the 3-4ghz chips AMD releases around the same time. Both will have the same benchmarks, but it always comes down to cost. How much is this crap worth?

Guess I have this super talent of stating the obvious. But I'd like some feedback. Maybe someone could explain how the AMD chips are able to get the same benchmarks at lower clock speeds with the same size wafer. In lamenst terms

Ruckas-

 

[dullard]

Originally posted by: Ruckas
I love how AMD chips are clocked lower than the intel chips. But get the same benchmarks give or take. I could go read about how this is accomplished. But I'll just asume that AMD trys to make chips cheap, and perform well. While Intel uses the latest technology ALL the time. Without fine tuning the chips they already have. That's my opinion.

Guess I have this super talent of stating the obvious. But I'd like some feedback. Maybe someone could explain how the AMD chips are able to get the same benchmarks at lower clock speeds with the same size wafer. In lamenst terms

The performance of a CPU can be simplified down to this equation: performance = f*I. In that equation let 'f' be the frequency of the CPU and let 'I' be the average number of calculations the CPU can perform per clock cycle (sometimes people call this IPC). How can a company increase the performance? Well there are 3 ways:
1) Increase f.
2) Increase I.
3) Increase f and I. Unfortunately this is quite difficult.

Intel has taken option #1 on its P4 design. With a given I, as f increases, the performance will increase. AMD's Athlon chips were originally deisigned for a lower f and a higher I. It doesn't matter if you take 2*3 or 3*2 they both equal 6. Thus no chip is inherently better or worse. They just had different design goals. Note: Intel also has chips that were designed the other way. The Intel Itanium line has a low f, but a very high I and thus outperforms both the P4 and the Athlon in many benchmarks. So it isn't an AMD vs Intel thing, since Intel has tried both routes.

That is about as non-technical as I can get.

Now obviously Intel will win in a race to achieve 5ghz. However, I have 50 bucks cash says that the intel 5ghz cpu will cost 3-400 more dollars than the 3-4ghz chips AMD releases around the same time. Both will have the same benchmarks, but it always comes down to cost. How much is this crap worth?

As for future pricing we don't really know for sure what will happen. But look at the current official prices of similar performing chips:
Links are here for AMD prices and here for Intel prices.
2400+ XP: $193, 2.4 GHz P4: $193
2600+ XP: $297, 2.66 GHz P4: $305
2800+ XP: $397, 2.8 GHz P4: $401
3000+ Barton: $634, 3.06 GHz P4: $637.
Looks like the official 1000 lot prices are almost identical for similar performing top-end chips. Note: street prices vary. AMD is in the midst of their 1st quarter price cut. Intel's 1st quarter price cut will shave $50 off every 2.5 GHz and higher chips in 3 weeks. For that reason street prices will fluctuate wildly in the next month with AMD usually a few dollars in the lead. For the vast majority of AMD's life it only undercut Intel by a small amount for similar performing chips. There was a period from March of 2001 to October of 2002 where AMD slashed prices to gain marketshare. But unless that returns I see no reason why AMD will be significantly cheaper than Intel when we reach 5 GHz.

 

[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

 

[HokieESM]

Originally posted by: Ruckas
I love how AMD chips are clocked lower than the intel chips. But get the same benchmarks give or take. I could go read about how this is accomplished. But I'll just asume that AMD trys to make chips cheap, and perform well. While Intel uses the latest technology ALL the time. Without fine tuning the chips they already have. That's my opinion.

Now obviously Intel will win in a race to achieve 5ghz. However, I have 50 bucks cash says that the intel 5ghz cpu will cost 3-400 more dollars than the 3-4ghz chips AMD releases around the same time. Both will have the same benchmarks, but it always comes down to cost. How much is this crap worth?

Guess I have this super talent of stating the obvious. But I'd like some feedback. Maybe someone could explain how the AMD chips are able to get the same benchmarks at lower clock speeds with the same size wafer. In lamenst terms

Ruckas-

Well, Ruckas, you have a good point about the LOWER clock speeds. But AMD really isn't cheaper than Intel at the top of the scale. If you look around, you'll see that a retail 2400+ is about the same as a retail P4 2.4B. And at the top end, its the same. Remember when AMD won the race to 1GHz? Guess what? They had the most expensive chip out there.

Now AMD DOES destroy Intel at the lower end--a 1800+ makes a 1.8A look ridiculous (if you take overclocking out of the mix.... and MOST consumers do).

Too many people try to say "AMD is the cheap solution and Intel is too expensive" or "I want a stable machine so I'll go Intel" or something really biased and stupid like that. In all realism, this world is competition. AMD and Intel took different roads to chip design. Hence, AMD is much more "efficient" MHz for MHz, so you get the whole 'power rating' bit. And its a VERY good number, to be honest (comparing the XP to the Intel--PR to MHz--you get a "win some, lose some" in benchmarks). But the Athlon XP isn't capable of the raw frequency that the P4 is.... so it was a compromise. Intel chose a different route. Note that I'm not being biased here... I actually don't care. I'm going to do the "math" on each computer I build and figure out which is better for my personal usage.

In any event.... as far as the progression in clock speed/CPU speed.... I think we're really going to need a new input/output device for the common person to really take advantage of the speed. Not to mention a "cleaner" operating system.

But I think what we'll REALLY see is the integration of computers into home entertainment more completely..... I built my parents a HTPC and wrote a script for them to "tape" a show and it encodes it into DivX. This is the ONLY way that you EVER see my parents tax even a XP1800+ (and they're the average college educated, upper middle class, 50 year olds).

 

[HokieESM]

very good points dullard. you just type faster than me.

 

[Stunt]

first off id like to say that moores law is dead if this is true
10.2ghz sounds sweet
the fsb sounds really bad though
comparison of fsb to core clock in some other processors

p4 3066 with 533 = 17.4% of core clock
p4 2400 with 400 = 16.7% of core clock
p3 1000 with 133 = 13.3% of core clock
p3 800 with 100 = 12.5% of core clock
athlon 800 with 200 = 25% of core clock
athlon 1800+(1533) with 266 = 17.4% of core clock

finally
10200 with 1200 = 11.8% of core clock

sounds crappy to me
but core clock is impressive

 

[dowxp]

Originally posted by: Stunt
first off id like to say that moores law is dead if this is true
10.2ghz sounds sweet
the fsb sounds really bad though
comparison of fsb to core clock in some other processors

p4 3066 with 533 = 17.4% of core clock
p4 2400 with 400 = 16.7% of core clock
p3 1000 with 133 = 13.3% of core clock
p3 800 with 100 = 12.5% of core clock
athlon 800 with 200 = 25% of core clock
athlon 1800+(1533) with 266 = 17.4% of core clock

finally
10200 with 1200 = 11.8% of core clock

sounds crappy to me
but core clock is impressive

uhm, what does moore's law have to do with clock/fsb ratio? furthermore, why does the ratio even really matter.

 

[Stunt]

it doesnt, they were two distinct points

nm i looked it up and moores law refers to # of transistors doubling every year
but moores law is already at 18months, so its not true anyways...

also i think the fsb should be much higher...
or are we back in the 1000 p3 with 100fsb days...
to be quite honest, the 1200 fsb is not impressive at all looking at past ratios
hell prescott will have an 800 fsb
making it over 20% of core clock
i like to see the fsb progressing with the core clock
oh and buddy if fsb doesnt really matter for you thats just great because there are many ppl that do
like the ppl with 1833mhz xp's with 400fsb
it makes a difference
btw what if it was a 100ghz cpu with 10mhz fsb, whould you be impressed?
im sure you would, look at that clockspeed!
"why does the ratio even matter, he-yuk!"
good point dumb@$$, lol

 

[dowxp]

Originally posted by: Stunt
it doesnt, they were two distinct points

nm i looked it up and moores law refers to # of transistors doubling every year
but moores law is already at 18months, so its not true anyways...

also i think the fsb should be much higher...
or are we back in the 1000 p3 with 100fsb days...
to be quite honest, the 1200 fsb is not impressive at all looking at past ratios
hell prescott will have an 800 fsb
making it over 20% of core clock
i like to see the fsb progressing with the core clock
oh and buddy if fsb doesnt really matter for you thats just great because there are many ppl that do
like the ppl with 1833mhz xp's with 400fsb
it makes a difference
btw what if it was a 100ghz cpu with 10mhz fsb, whould you be impressed?
im sure you would, look at that clockspeed!
"why does the ratio even matter, he-yuk!"
good point dumb@$$, lol

lol what an idiot. why are you posting from high school? theres no point to be arguing with you!

 

[Stunt]

actually posting from my house, and your right i am at school
university
also, highschool gets out at 3 or something does it not?
lol
please tell me your brilliant opinion on why fsb doesnt matter
trust me i think i'll understand
not trying to argue
just poking some fun at you horrible reasoning
of course you could not answer since you seem just too damn smart for the rest of us idiots...

 

[dowxp]

Originally posted by: Stunt
actually posting from my house, and your right i am at school university also, highschool gets out at 3 or something does it not? lol please tell me your brilliant opinion on why fsb doesnt matter trust me i think i'll understand not trying to argue just poking some fun at you horrible reasoning of course you could not answer since you seem just too damn smart for the rest of us idiots...

lets see. you are AT your school university right now. but you are posting from your HOUSE. nice.
Anyways, fsb does matter, but thats wasnt my point. you listed % ratios of different cpu's and architectures and finally concluded because the ratio of of 10.2ghz and 1.2ghzfsb is the lowest of them all, you call it crappy. i dont even think thats a legit way to compare performance of cpu's.

 

[dowxp]

and please explain how because it has been "18 months" that "moore's law" is false oh great one.

 

[Stunt]

huh, u're last post made no sense
no pass english?
anyways, yes i live in my house, at school, with 6 friends
thanks for pointing that out, because i said i lived with my family right?
oh wait, you are assuming sh!t...
i looked back at all the previous architectures to prove that the fsb seems quite low
but ya you are right they could have some special technology up their @$$
i'd still like to see it higher for other reasns though...
like oc'ing
lower multiplier is much better since it is locked.
just an observation...i like how you take criticism towards intel personal.
i think the cpu will be bottlenecked by such a low fsb, like the p3 did.
sure it will be fast, its 10ghz, but its low percentage wise for this day in age.

 

[dowxp]

Originally posted by: Stunt
no pass english?

I cant say the same for you. My point however is that there are too many factors such as mem arch, speed [ddrII], latency, to even guess that 10.2 ghz would be bottlenecked. why? because its all speculation and assumption. My view is that, not everything is raw speed. Other things can influence performance from SDR -> RDR -> DDR, or Hyperthreading (who would have thought 2 threaded cpu?) ; Moreover, on the AMD side, Hypertransport which virtual doubles the memory speed or something like that AFAIK, and about hammer well, no one knows.

 

[Ruckas]

Thanks alot dulard, and hokie. That explaination with the F*L was really informative. When I bought my p4, 1.5ghz was the hottest thing out. I remember looking at the prices of the AMD chips, then looking at the benchmarks. At the time I had wads of cash. And the benchmarks said the p4 (socket 428) was the best. So I got it. Man what a mistake that was. If I would have waited just 1 month, the northwood would have come out.. Now I'm just stuck with this 'ok' computer. I'll get around to making another one soon.

Anyway! My point. Because I got stung so bad, I've had this dwelling memory stuck in my head that...... Amd is just better. Because it's cheap, and if stuff breaks it's so much easier to fix. Apparently that isn't the case anymore. Intel actually brought thier prices down. That; or they were forced to do so because AMD was taking all thier bussiness. I guess I've been out of the loop for awhile heh...

Well, I'm all caught up, thanks almost entirely to you guys..

Ruckas-

 

[Sohcan]

Originally posted by: Stunt

i think the cpu will be bottlenecked by such a low fsb, like the p3 did.
sure it will be fast, its 10ghz, but its low percentage wise for this day in age.

Given that Nehalem is at least two years away and nearly zero details are publically known about it, it's still far too early to know what the FSB will be clocked at and if the CPU will be bottlenecked by its bandwidth.

* Not speaking for Intel Corp. *

 

[Stunt]

ya for sure lots of things determine speed of a system
like memory etc
but newsflash! intel isnt in charge of making memory, if it was we'd be using RD right now
the memory will be no different
intel/amd etc
intel is into cpus, and thats what we are looking at now
you can say chipsets etc as well but via/sis/nvidia/ati/ali can do that just as well
so fsb/cache/coreclock/extentions is what intel is looking after
cache is expensive, hinders yields, and scalability.
if they want to make a fast proc in terms of speed (rather than clock), they are going to want to up that fsb a lot.
i personally think it is bogus
i expect more out of intel than that...

just think, 2 years ago:
p3 1000 with 133 fsb
today:
p4 3060 with 533 fsb

so within the last 2 years, they uped core 3times, and fsb 4times
thats impressive

now look at 2 years into the future:
10200 with 1200

that would be uping the clock by 3.3times and the fsb 2times
i hate to see all those mhz go to waste

sounds kinda lacking from a fsb standpoint
duno bout you but i think 10.2ghz with 2.5ghz fsb sounds much more logical esp with 800 coming by fall.
just me, lets hear what other ppl think bout the fsb...

does it make sense?

 

[paralazarguer]

I'm not going to get involved in the actual topic of this conversation. I just wanted to drop a note to let stunt know that he's being an idiot and that he shouldn't pass judgement of other people's gramatical abilities when he writes at a sixth grade level. I also find it hard to believe that he's in University because they teach open-mindedness and the ability to express thoughts clearly, not to mention the english language.
Maybe he just started this January.

 

[Stunt]

oh my grammer...
for typing on a message board!
hey...you are right...
i forgot to f'n proof-read
or do a final draft
well sh!t
yes i am at university 3rd year elec eng
so yes im not the best at grammer (i am an engineer, lol)
but my point can still be seen
i didnt make fun of his grammer, i just chose not to answer a question that i couldnt even understand.
im normally not a pain in the @$$, sorry bad day.
anyways if you think they teach you english just cuz you go to university...
someone forgot to tel me and my 8 classes a term and my 40hours a week...
anyways pointless arguement, i was analysing the "Nehalem" processor.
sorry if im too offensive for your liking
post something useful next time

 

[Stunt]

btw this forum seems pretty active
does anyone know what xbit posted about the 9900pro?
ive been trying to get it
but cant
maybe a linky too
thatd be great
thx

 

[paralazarguer]

oh my grammer...
for typing on a message board!
hey...you are right...
i forgot to f'n proof-read
or do a final draft
well sh!t
yes i am at university 3rd year elec eng
so yes im not the best at grammer (i am an engineer, lol)
but my point can still be seen
i didnt make fun of his grammer, i just chose not to answer a question that i couldnt even understand.
im normally not a pain in the @$$, sorry bad day.
anyways if you think they teach you english just cuz you go to university...
someone forgot to tel me and my 8 classes a term and my 40hours a week...
anyways pointless arguement, i was analysing the "Nehalem" processor.
sorry if im too offensive for your liking
post something useful next time

Sure, here's something useful:
It's "grammar" not "grammer."