Will Next Gen Intel CPU regain the performance crwon from AMD.?

X

Xcom_Cheetah

Limp Gawd
Joined
Apr 12, 2002
Messages
197
Reading all the coverage of the IDF i m still not sure if next generation Desktop Core of Intel will be able to win back the performance crown... Although it doesn;t matter much for intel baseline but for the prestige of having the fastest processor it surely is pretty important...
But going through the coverage i didn;t read anywhere of Intel such claim... All they are talking about performance/watt... which is good but will it be good enough to take back the crown... i think i have some doubts..
What you think about this matter.? will Intel be introducing any such processor like extreme edition just to take back the fastest processor crown.? or intel will like to keep ahead in performance/watt. and not talking about actual performance number.?
 
no one really knows, but intel dominates the mobile market where as amd with desktop performance/price ratio.
 
You are all right but i was just hoping that after going through the information so far known about the next generation Intel architecture you will be able to make some informed guess about it. I mean let see, the biggest improvement for me is
1. the more decoders. should give some minimum 20+% performance.
2. Shorter pipeline some 7-8% improvement here.
3. then micro and macro code fusion 5% improvement.
4. then you have more 4 pipelines ... some 10+ % improvement here. i mean for additional ALU units.
5. then these minor improvements like better lower latency, high bandwidth, low cycles for some specific instructions.. etc.. they should be around 5+%.

i think these all improvements should make a pentium M some 50% faster then its current performance. I think if it matches these performance estimate then a 2.4 Ghz will be pretty good compared to 2.8 Ghz X2... and i think it might be a just a tad faster then the AMD's offering at that time.
 
Xcom_Cheetah said:
1. the more decoders. should give some minimum 20+% performance.
2. Shorter pipeline some 7-8% improvement here.
3. then micro and macro code fusion 5% improvement.
4. then you have more 4 pipelines ... some 10+ % improvement here. i mean for additional ALU units.
5. then these minor improvements like better lower latency, high bandwidth, low cycles for some specific instructions.. etc.. they should be around 5+%.
Click to expand...

Well, that's why I say it's all speculation. We could all pull numbers out of our a$$. Who knows, perhaps you're right.

The other thing one needs to remeber is that this is a totally new architechture (depending on who you ask) for Intel. It will take Intel a year or so to tweak the hell out of it like they have with Netburst. If I were you, I would expect initial performance to be lack-luster at best and slowly ramp up over time. Who knows, by the end of the arch's lifetime, we may see it pass AMD.
 
No one knows, We can only speculate. I say it "should" be, if Intel has any sence at all. But no one knows for sure.

The only thing that is for sure is, It will be hella faster than Intels Previous Generation Netburst Processors.... and thank god for that.
 
Intel already has the crown (what u talking about regaining)? ;)

Intel had, has and always will have the crown...
AMD can either go bye bye, or go help build CPU's for Apples...

Opps... Intel beat AMD to the punch! ;)
 
The P4 line lost intel the "Gaming" crown, however they do have a little CPU the P-M which is a great little CPU for everything including gaming. However I just wonder if they will create a new desktop chip that doesnt run so hot that you could heat up your room in teh winter off its output :)
 
I think Intel will very likely regain the crown next year. I dont think AMD has anything planned for next year thats going to increase performance much from where their at right now and if Intel will just stay away from long pipeline style CPU's like the Northwood and Prescott cores and move to Pentium M style processors with Yonah and Conroe then the future looks very bright for Intel on the desktop. The Pentium M's are already as fast or faster then most high-end Athlon 64's in the majority of games and they dont even have an on-die memory controller. The Pentium M 770 can stay with an FX-55 in gaming usually. They need more memory bandwidth performance and need to go dual core in order to compete in the application spectrum though. The Pentium M's drag up the rear in multithreaded applications.
 
Saeid said:
Intel already has the crown (what u talking about regaining)? ;)

Intel had, has and always will have the crown...
Click to expand...

we were not talking about your opinion here which this obviously is
 
O-P-M said:
Doesn't AMD have the m2's coming out?
Click to expand...

Oh, you mean, the Athlon64 with a DDR2 memory controller and Pacifica virtualization, plus a higher TDP? Yes, they do, sadly.

Saeid said:
Intel already has the crown (what u talking about regaining)? ;)

Intel had, has and always will have the crown...
AMD can either go bye bye, or go help build CPU's for Apples...

Opps... Intel beat AMD to the punch! ;)
Click to expand...

What crown is this?
 
Does Intel plan on putting the memory controller directly onto the processor like AMD has?

It's too bad that memory performance can't keep up with the processor's. Hopefully with the present speed wall, memory can finally catch up with the processor's. A processor with a 1x multiplier? :D

Also does Intel plan on getting back on track with their system chipsets? The last real decent ones were the i865/i875 series.

It's interesting to be seeing Intel going back to, essentially the P3 core, which clock for clock, was always faster than the P4.
 
Sure it was faster clock for clock than the P4, but the P3 was limited to 1.4ghz (1.7ghz overclocks were possible, but they can't compare to a 3.8GHz P4.)

And what's wrong with the 945 / 955 chipsets?
 
No. It won't. I'd explain it, but .....(insert over inflated technical response of your choice here)....
 
robberbaron said:
Oh, you mean, the Athlon64 with a DDR2 memory controller and Pacifica virtualization, plus a higher TDP? Yes, they do, sadly.
Click to expand...
Socket M2 will also add support for quad-core at sometime during it's lifetime. Also, AMD has socket F for servers comming next year. There are about 267 pins unaccounted for on this new socket, so AMDs server CPUs have some, as of yet undisclosed, surprises comming.
 
Eickst said:
No. It won't. I'd explain it, but .....(insert over inflated technical response of your choice here)....
Click to expand...

...Come on... we're waiting!
 
I think it will be really impressice on Intel's part if they manage to beat AMD's processors in gaming without adding an on-die memory controller. and looking at the next generation Intel's processor it does look like its going to atleast equal AMD's best offering at that time. Which i think will put a lot of pressure on AMD for their next generation processor. because Intel is going to get an on-chip memory controller processor some time in the future which will straightaway give them 8-10% performance boost.

So definitely Intel is learning fast..
 
visaris said:
Socket M2 will also add support for quad-core at sometime during it's lifetime. Also, AMD has socket F for servers comming next year. There are about 267 pins unaccounted for on this new socket, so AMDs server CPUs have some, as of yet undisclosed, surprises comming.
Click to expand...

Probably for integrated PCI-E or more data pins for the quad core solution.
 
I hope not, because we need intel's stock to go down that way they'd have to sell their processors at a lower price even still. Market balancing = good for me. If their stuff is awesome, people would again be less likely to buy AMD.
 
GoHack said:
It's interesting to be seeing Intel going back to, essentially the P3 core, which clock for clock, was always faster than the P4.
Click to expand...

The P3 core had a pipeline of 10 stages long, and the new architecture will have 14 stages.
So there will be at least a 40% difference between the P3 pipelines and the new architecture, based on the number of stages alone.
And then there may be a lot of technologies that they might lift from P4, like double-pumped ALUs, trace cache or HT. Ofcourse it must also support EM64T.
And they will probably also introduce some new technologies, as they have previously done with every new generation.
So really, I think the end-result will have very little in common with the P3 core, other than the fact that it has a higher IPC than the P4 aswell.
But I suppose this whole "P3-core" thing sounds nice to people from the 'other side', it has a nice negative ring to it. To me it sounds like "Look, they're back to where they were in 1996". Trust me, they're in 2005, like the rest of us.
 
robberbaron said:
Probably for integrated PCI-E or more data pins for the quad core solution.
Click to expand...

Sorry to go off-topic, but on-die PCI-e might be a pretty wicked feature... now the important thing is that it should support SLI and crossfire. ;)
 
Josh_B said:
Sorry to go off-topic, but on-die PCI-e might be a pretty wicked feature... now the important thing is that it should support SLI and crossfire. ;)
Click to expand...

Good luck with the licensing on that!


Personally, I wouldn't be suprised if AMD is already planning (even though Intel just announced it at IDF) on die voltage regualtion. Just because they have a lot of pins unaccounted for doesn't mean that they are all data pins ;)


On board PCI-E would be nice, but what if they are working on "Quad Channel" DDR? It's pretty much anyone's guess, unless someone knows some people who know some people that robbed some people at AMD.
 
On board PCI-E would be nice, but what if they are working on "Quad Channel" DDR? It's pretty much anyone's guess, unless someone knows some people who know some people that robbed some people at AMD.
Click to expand...
doubt it. the fact taht you would need 4 DDR DIMMs and twice hte number of traces on the board for memory makes that very unfeasable, just like going from 256bit to 512bit memory interfaces on graphics cards.
 
Josh_B said:
Sorry to go off-topic, but on-die PCI-e might be a pretty wicked feature... now the important thing is that it should support SLI and crossfire. ;)
Click to expand...

Transmeta chips have on-die AGP, heh. I think 4x...
 
You must log in or register to reply here.
Back
Top