30 stage Prescorr pipeline?!

hahaha intel and there extra long pipes its not the size of someones pipe ist the smooth foward thrust of performance in their pipe.

AMD 4 LIFE
 
That rumor was started by Inq, I believe, about a month ago. There is a thread discussing it. I would easially say it is BS, and probably stems from the different ways you can count the P4 pipeline and Inq's desire for hits.

The Prescott rumors from about a year ago, which pegged 1meg L2, PNI, and improved HT, have said Prescott has about 2 extra stages.
 
So many negative naysayers out there really seem to dread any boost Intel would get from a successful Prescott design. Wouldn't adding in an extra extended instruction set (PNI) be good reason for an extra pair of stages? Plus Prescott had a non-FPU routed integer design, which is supposed to help integer performance scale better. Add in the 90nm construction on 300mm wafers and its bound to drive costs of the processors to lows not found in previous flagship releases.
 
"It will have a larger pipeline," the representative said. "The larger the pipeline, you can do less work per clock and speed up the processor."
Click to expand...

Atleast Intel is admitting that they get less work done per Hz. :p

I think the Prescott will do just fine from everything I have heard. I guess we will just have to wait a few more days till the NDAs are lifted.
 
I wouldnt call 30 stages completely crazy. This is the next-gen ground-up design, it should be nice. While the current northwoods will make it look like a lost cause, it will ramp in time...far farther than northwood could dream.

Now then, the timeframe of this ramping depends solely on the prowess of intels process engineers...
 
Adding stages is a good thing wrt the P4 - it enables higher frequencies which the P4 lives by. An extended pipeline will only be problematic in the beginning/transitional stage of prescotts life, when its around the 2.8 to 3.4 ghz frequency.
 
Originally posted by Surly
Adding stages is a good thing wrt the P4 - it enables higher frequencies which the P4 lives by. An extended pipeline will only be problematic in the beginning/transitional stage of prescotts life, when its around the 2.8 to 3.4 ghz frequency.
Click to expand...

Not really, because this time around Intel is beefing it up more in other important areas. Overall the prescott should be faster than the northwood clock for clock.
 
Originally posted by Big Worm
Not really, because this time around Intel is beefing it up more in other important areas. Overall the prescott should be faster than the northwood clock for clock.
Click to expand...


Of course this remains to be seen whether the increased cache and 'enhanced branch predictors' are enough to maintain or increase performance on a clock to clock comparison with northwood.
 
Originally posted by Surly
Of course this remains to be seen whether the increased cache and 'enhanced branch predictors' are enough to maintain or increase performance on a clock to clock comparison with northwood.
Click to expand...

There are some credible sources floating around, The expected prescott performance isn't as secret as the holy grail or anything.
 
FPU garnered through the SIMD routes will be probably be Prescott's strongest suit, with integer and raw FPU being weaker yet. Just seems like that is the trend of Intel when they move to longer pipelines.
 
i think ars hit it on the mark. this is what i thought was planned with hyperthreading anyway. besides an established multithreaded code path for multicore cpu's.
 
Some people are expecting a drop almost 20% to prescott from northwood, assuming the clock speeds are equal. I HOPE the longer pipelines don't lead to that large of a degradation :rolleyes:
 
Fortunatly it has around about 1~5% total IPC increase.

resulttable.jpg


The sandra scores though I think are bogus. When haven't they been anyway? I have not taken sandra scores seriously for quite along time.

Check this link for some better sandra scores.
http://www.ocheaven.com/article/0310/readgoodarticle.asp?id=32
 
According to JCViggen the Prescott has 32 total stages in the pipeline, up from 28 in Northwood. This is the first time I've seen anyone lay a specific number on Prescott.

Someone else goes on to define multiple ways to measure the pipeline length, too, one using ALU and the other FPU. What the bickering about Prescott and 30 stages be much ado about nothing. Quote: "And the band played on..."
 
makes sense to increase the pipeline stages since its the way to get higher frequencies
 
doesnt matter how you look at it, higher frequencies means better performance
 
Originally posted by dderidex
Overclockers.com has an article on this, as well.

Kinda more information - sort of.
Click to expand...

I think this comment sums it up nicely:

I really don't think I'm wrong.
Click to expand...

Well I think this guy sure is basing alot of claims about the future of prescott without any evidence as of yet. He didn't even take into consideration any of the improvements of the prescott, only harping on the increased pipeline and thats IT. The prescott isn't simply a 3.4 Ghz northwood with an increased pipeline:rolleyes:

I would take that article with a grain of salt, even if he did have any links or sources in the first place.
 
Originally posted by 3dfx
doesnt matter how you look at it, higher frequencies means better performance
Click to expand...

Riiiiight....

That's why the Celeron 2.4 pwns all over the Athlon 64 '3000+' (a mere 2.0 ghz chip), right?
 
Originally posted by dderidex
Riiiiight....

That's why the Celeron 2.4 pwns all over the Athlon 64 '3000+' (a mere 2.0 ghz chip), right?
Click to expand...

Ouch, Pwned! :eek:

Personally I still say this will be a wait and see thing, just be patient.
 
Originally posted by CentronMe
Ouch, Pwned! :eek:

Personally I still say this will be a wait and see thing, just be patient.
Click to expand...


yeah, good point, one thing we do know is that Prescott will be an overclocking king!
 
Originally posted by Big Worm
Wow, I havent been to that site in years, I think I first registered around 98/99. Is it still fairly popular?
Click to expand...

Yeah if you want to find a pretty good plethora of older members that is a good place to go. Not alot of that young flamer crap and we have only one cpu forum for everyone. Intel and AMD discussion is not seperate.
 
Originally posted by dderidex
Riiiiight....

That's why the Celeron 2.4 pwns all over the Athlon 64 '3000+' (a mere 2.0 ghz chip), right?
Click to expand...

in terms of DVD encoding, and MP3 encoding, and when o/c'd yes, yes it does.
 
Originally posted by rayman2k2
in terms of DVD encoding, and MP3 encoding, and when o/c'd yes, yes it does.
Click to expand...

I use to use my Duallie AMD rig 2 1700+ XP's @ 2300mhz each. for encoding my DVD's "backups":D

My Shuttle box was faster with w 2.2 celeron @ 3.3ghz. I just upgraded it to a 2.8 celeron @ 3.6ghz.


But my AMD rigs F@H way better than my P4 and celery rigs.



Simple build what you need for each purpose. What do you think credit cards are for anyway!!!
 
the dual AMD vs the celeron
eheh yea needed more clarification
celeron wins encoding
 
You must log in or register to reply here.
Back
Top