Intel to build future on CPU from 1995?

[DemoCoder]

For more disconcerting to me is Intel's emphasis on performance/watt, this seems to be suggesting that they can no longer compete on pure speed and will focus all of their efforts to change the way consumers evaluate CPUs towards their new model, which is power efficiency. Fine for notebooks, but they seem to be pushing this even for desktops.

Does this mean future Intel chips will see diminishing returns on speed? As the numerator of the perf/watt fraction flattens out, the denominator can make big gains, and Intel will still claim great progress in their marketing, but I for one will weep.

 

[Saem]

Demo,

One way to reduce power consumption is to increase performance, this way one reduces the amount of time the chip spends in a high consumption state. This makes no difference in the absolute consumption metrics, but huge difference in, work load metrics. The latter is is part of Banias' approach.

Then there is also the fact that as they drop the wattage ceiling they can hack on more cores. With that and hope to God MS actually makes the situation better on the library and compiler front and we might not miss fast serial processors on the desktop.

Currently, I think the biggest thing lacking is not so much the software, but a nimble platform that can adapt well to the approaching massively parallel architectures, allowing the software to follow with greater ease.

 

[ANova]

Nope, this new arch is more than this, but I guess it'll be next year until we know more than your casual "4-issue wide end-to-end".

Oh yeah, it'll have alot of smaller improvements as well. Especially in the areas where the Dothan currently lacks.

Does this mean future Intel chips will see diminishing returns on speed? As the numerator of the perf/watt fraction flattens out, the denominator can make big gains, and Intel will still claim great progress in their marketing, but I for one will weep.

They will be faster and consume less power. win/win

 

[Khronus]

Anandtech has some info on a P-M w/ on package memory and voltage controller. No mention of a timeframe as I recall though.
http://www.anandtech.com/tradeshows/showdoc.aspx?i=2511

 

[pascal]

For more disconcerting to me is Intel's emphasis on performance/watt, this seems to be suggesting that they can no longer compete on pure speed and will focus all of their efforts to change the way consumers evaluate CPUs towards their new model, which is power efficiency. Fine for notebooks, but they seem to be pushing this even for desktops.

Does this mean future Intel chips will see diminishing returns on speed? As the numerator of the perf/watt fraction flattens out, the denominator can make big gains, and Intel will still claim great progress in their marketing, but I for one will weep.

Or this could mean market pressure. Not everybody is a gamer that will accept 5 times more heat for ~20% more linear performance.

Corporate likes cool chips. Mom and dad likes cools chips. Many people want those chips in the living room, cool and quiet. I love cool and quiet chips too

I miss my P3-S/Tusl2-c combo

Give me a dual-cpu, cool and quiet, some high-speed UMA, integrated GPU for vizualization oriented system, and a nice RTOS (concurrent threads) with a clean interface and I will be happy.

 

[SugarCoat]

This sucks. I think Guden is right that Intel is blowing marketing smoke into people faces again. They are trying to make people believe that this is the biggest announcement in 5 years but in reality it is nothing new. 30% faster than Netburst? AMD didn't even bothered to counter act to Intel. I am sure they are thinking that they can beat whatever Intel will bring by their dual core CPUs.

Dothan runs amazing and thats basically where conroe comes in. Conroe is a new core built from the ground up, but based on the more work, less MHz idea. People are speculating a 1333FSB but that may actually be reduced to 1066, shorter much more improved pipeline system with 14 stages (Dothan has 10, and a prescott of today has 31 for comparison), 65nm of course and all the added goodies of todays P4 plus extras. I dont think its smoke an mirrors because the Dothan line are amazing proven performers, and from the changes to the chips told, its going to be a very nice cpu. And if its not, on with AMD right? If it is, and especially if they undercut with prices, it maybe quite some trouble for AMD. Personally i want what ever is best.

Rough estimates from the changes in conroe compared to a P4 today, put it closer to a 34% improvment. Its really kinda cool, now just need some chips .

 

[Entropy]

For more disconcerting to me is Intel's emphasis on performance/watt, this seems to be suggesting that they can no longer compete on pure speed and will focus all of their efforts to change the way consumers evaluate CPUs towards their new model, which is power efficiency. Fine for notebooks, but they seem to be pushing this even for desktops.

Does this mean future Intel chips will see diminishing returns on speed? As the numerator of the perf/watt fraction flattens out, the denominator can make big gains, and Intel will still claim great progress in their marketing, but I for one will weep.

Er.. The flattening of performance increases already happened.
Your tears of lament are five years or so late.
Even while accepting huge power draw increases, doubling performance has taken Intel roughly three times as long as it used to. Now the CPU manufacturers are doing what they can to keep ASPs high to an audience that has been taught that speed is the figure of merit to look for. They are simultaneously trying to reeducate, and trying to get speed along alternative routes.

But the days of (single thread) performance doubling every 18 months are over. They won't be back, and programmers who factor in future performance increases along those lines in their projects had better wake up quickly. As in - a few years ago.

 

[DemoCoder]

I don't care about single thread performance, I care about total performance. On the high end, I am simply not interested in power draw. I'd put a 1KW PSU into my desktop if it guaranteed me I could buy 2x power.

But Intel's dual core is too little too late IMHO, and no one except maybe some startups and IBM, is trying to do anything really innovative anymore and break out of the Wintel paradigm that is shackling us.

 

[Entropy]

I don't care about single thread performance, I care about total performance. On the high end, I am simply not interested in power draw. I'd put a 1KW PSU into my desktop if it guaranteed me I could buy 2x power.

Well, here we disagree.
I feel that high power draw systems belong in server rooms, basically. I like my personal tools to be discreet and comfortable in all respects.

But Intel's dual core is too little too late IMHO, and no one except maybe some startups and IBM, is trying to do anything really innovative anymore and break out of the Wintel paradigm that is shackling us.

Here, however, I find myself in pretty much complete agreement.
The industry is stagnant, and unlikely to move much given established patterns and alliances, and with DRM issues an additional wet blanket to grass roots enthusiasm.

Bad times for enthusiasts, and unlikely to improve much. Time to cultivate other interests.