Intel finaly goes for a big change

[SpiralDragon]

http://www.theinquirer.net/?article=15749

http://www.theinquirer.net/?article=15760

http://www.theinquirer.net/?article=15768

glad i waited.... ...

[TransformX]

Eventually this means that most of the next lines of Intel CPUs are coming from Haifa-Israel.

[SpiralDragon]

lol... thats true... .... you've got some realy smart people there...

[az]

Well that certainly makes them better

But I guess it's a good thing, if true - Prescott was really only the last step in the wrong direction.

AZ

[SpiralDragon]

hopfuly...

[TransformX]

Apparantly those nut-cases from Haifa do have some bright ideas. I just hope they're not too bright so we still have a competition between Intel and AMD to keep prices reasonable.

[GT98]

So if they are going the Penitum-M chip...aren't they in accualitly going back to the good ole P3 design, abit with one 1MB of cache on it?

I have a Penitum-M 1.6Ghz CPU in my laptop and its great considering its a laptop. I was able to play Jedi Academy on it last fall

[dZeus]

not p3, PPro you mean

P6 is a real nice architecture.. let's see what Intel can do with it

[K6-III]

Good direction to see Intel heading...

[bsdgeek]

I'd think that there's quite a bit of clockspeed headroom for Pentium-M when it doesn't have to worry about working in a low-power environment. I don't know what it would be, but if it's there, P-M would be immediately adaptable for desktop use. Very cool.

But, what does this announcement do to the future of BTX? Is it still coming?

[rylan]

Well Intel essentially admitted to finally fu*king up here, as they've been going for higher clock speed at the expense of operations per clock and added heat. Thats what they get for creating Marketecture where 'higher clock is better'. I think they realized it when AMD gave them a big bitch-slap with their dual core designs and much more efficient processors in terms of ops per clock, and therefore less heat problems.

[rugger]

rylan,

That is maybe a little harsh.

I am sure intel had NO idea 5 years ago when it designed the netburst architecture that it would be facing the problems it does today.

Five years ago, when the Netburst was on the drawing board, we had already reached the practical limits of IPC for the X86 architecture. Intel decided to head in another direction for improving overall performace rather then attempt for better IPC. Intel also decided to focus heavily on multimedia performance, where being able to move and process large volumes of data was more important then just low volume processing.

It is hard to deny that the P4 chip was going reasonably well until the move to 90mn fabs and Prescott rendered it dead in the water.

Finally, nobody has pushed CPU design over the last few years nearly as hard as intel, and they are paying the price for learning these limitations first.

Hopefully, Intel can well intergrate the lessons learned from the P4 into the P-M, giving us a fast, cool and multimedia focused processor.

[rylan]

Yeah it is kinda harsh, but for a huge company like Intel that invest billions of dollars into microprocessor R&D, they should've known that the track they were on would be unsustainable. With the P4 they shifted tracks towards a higher latency, higher clock rate design that is inherantly inefficient. Incidently, I would argue that the P4 was going reasonably well... Intel was hitting thermal design limits and power usage issues since they pushed the core past 2.4GHz. If you remember, they had problems with their own thermal limiting diode kicking in too early and slowing down the chip because their reccomended cooling wasn't sufficient at first.
Yes there are some good points in both AMD and Intel chip archetecture, but I have to say that AMD has done a lot more lately for the industry with their 64bit dual core design. I just think Intel got cought with their pants down again (as they did with the initial Athlon release) so have to spend more money to tweak things and catch up.

[TnT]

As long as a whole bunch more people in Oregon don't get laid off I'll agree to whatever change they make.

[Marshmallowman]

Intel and most si fabricators have been aware for a long time about aproaching frequency/density limts for Si, hence all the new things that are being tried.

So they go out and design a CPU that relies on high frequencies for performance...hello!?

I think a cynical reliance on marketing and FUD allowed them to make such a stupid choice. Luckily there is enough people with who don't rely on reality to make purchaseing decisions to keep intel the CPU business.