Intel (P4) Netburst is dead (it's _official_ now)

[halcyon]

From behardware (courtesy of golem.de originally):

In an interview with Golem.de, Patrick P. Gelsinger, Intel CEO, confirmed the abandonment of the NetBurst architecture. Future Merom, Conroe and Woodcrest processors, the second dual core generation for laptops, desktops and servers will all share a common new architecture.

The philosophy of this new architecture will be the same (but even more efficient) as the Pentium M: a relatively short pipeline which doesn’t reach very high frequencies but has a very high efficiency per MHz and per Watt. The Pentium 4 Netburst architecture long pipeline, which reaches high frequencies to the detriment of efficiency per cycles and which generates very power greedy CPUs should now be part of Intel past.

This puts to rest the remaining speculation as to whether the post-prescott CPUs based on Pentium M tech will use Netburst or not.

Also the goal seems to be high efficiency / Watt as well, which is good for us silent enthusiasts.

The problem is that we won't be seeing these till very late 2006 or early 2007, but at least they are coming.

Here's hoping that the other cpu speed demands will not have negated the wattage savings by then :)

[~El~Jefe~]

From behardware (courtesy of golem.de originally):

In an interview with Golem.de, Patrick P. Gelsinger, Intel CEO, confirmed the abandonment of the NetBurst architecture. Future Merom, Conroe and Woodcrest processors, the second dual core generation for laptops, desktops and servers will all share a common new architecture.

The philosophy of this new architecture will be the same (but even more efficient) as the Pentium M: a relatively short pipeline which doesn’t reach very high frequencies but has a very high efficiency per MHz and per Watt. The Pentium 4 Netburst architecture long pipeline, which reaches high frequencies to the detriment of efficiency per cycles and which generates very power greedy CPUs should now be part of Intel past.

This puts to rest the remaining speculation as to whether the post-prescott CPUs based on Pentium M tech will use Netburst or not.

Also the goal seems to be high efficiency / Watt as well, which is good for us silent enthusiasts.

The problem is that we won't be seeing these till very late 2006 or early 2007, but at least they are coming.

Here's hoping that the other cpu speed demands will not have negated the wattage savings by then

knowing intel, it wont give up any 'technology' but instead, created insanely hot chips that are truly useless in many ways. Whatever happened to the whole "relying on dothan" for future use?

These new dual cores are not riding that better highway. Id go for intel in a second if it made a more cold running dothan styled dual core setup.

[wsc]

Hell, I'll be satisfied if Dothan CPU/MB prices drop down 30% or so. That's all the processing power I need.

[Ralf Hutter]

Hell, I'll be satisfied if Dothan CPU/MB prices drop down 30% or so. That's all the processing power I need.

FWIW, a few months ago I remember seeing some Intel price move info that showed Dothan CPU prices are set to drop 28% sometime in late June/early July when Intel introduces some new CPUs.

[frostedflakes]

A step in the right direction for Intel IMO.

But I do not believe that Intel will completely abandon Netburst. Shorter pipelined CPUs like Dothan and Athlon K7 and K8 seem to be inherently worse than Netburst at certain tasks, encoding for example. I think Netburst could still flourish in a server environment, where heat, noise, and power consumption are less of a concern. But for a desktop, I would welcome a Dothan-esqe CPU with open arms.

[Green Shoes]

There's a new article over at THG (I know, I feel dirty) that explores the death of Netburst and the possibility of using Dothan in a desktop even more....and it certainly raised my eyebrows. Long but worth a good read. I question those benchmarking numbers, but if they're correct.......wow........

[Ralf Hutter]

There's a new article over at THG (I know, I feel dirty) that explores the death of Netburst and the possibility of using Dothan in a desktop even more....and it certainly raised my eyebrows. Long but worth a good read. I question those benchmarking numbers, but if they're correct.......wow........

Toms is about the biggest Intel shill on the internet. Combine that with their fairly poor methodology and note the results.

Note that in this particular "review", they're comparing an seriously OCed Dothan against a non-OCed AMD CPU. Their benches are also skewed towards ones that showcase Intel processors.

[Green Shoes]

Hmm...well, maybe I should trust my intuition a little more. I think they have a point, though, as trained monkeys could come up with a better architecture than NetBurst.

[ExpertNovice]

Toms is about the biggest Intel shill on the internet. Combine that with their fairly poor methodology and note the results.

Note that in this particular "review", they're comparing an seriously OCed Dothan against a non-OCed AMD CPU. Their benches are also skewed towards ones that showcase Intel processors.

I agree about comparing an overclocked chip with a non-overclocked is really assinine. However, even SPCR has touted the use of the Pentium-M for silent systems.

I'm happy to see that an SPCR reviewer is willing to slam results that confirms an SPCR philosophy. I assume that SPCR is now saying the Pentium-M should be shunned and AMD used instead?

[Ackelind]

In the review at tom's hardware, they praise the dothan chip for it's efficiency compared to AMD A64.

The only problem is that they are comparing it to the newcastle/clawhammer core.

Why the heck would they compare intel's newest flagship against a really old and sluggish AMD-core which has been outdated for quite some time now. I mean, they are two revisions back!

Sigh.

[Ralf Hutter]

In the review at tom's hardware, they praise the dothan chip for it's efficiency compared to AMD A64.

The only problem is that they are comparing it to the newcastle/clawhammer core.

Why the heck would they compare intel's newest flagship against a really old and sluggish AMD-core which has been outdated for quite some time now. I mean, they are two revisions back!

Sigh.

Already answered a few posts back:

Toms is about the biggest Intel shill on the internet. Combine that with their fairly poor methodology and note the results.

Do what I do, pay no attention to Tom's and you'll be the better for it.

[Shining Arcanine]

There's a new article over at THG (I know, I feel dirty) that explores the death of Netburst and the possibility of using Dothan in a desktop even more....and it certainly raised my eyebrows. Long but worth a good read. I question those benchmarking numbers, but if they're correct.......wow........

What impresses me is this:

http://www.tomshardware.com/cpu/2005052 ... y_cpu_load

The Pentium-M uses the same amount of power at different clock speeds. This means that it has no leakage problems. Of course its transistors will refuse to operate at insane clock speeds but the point is that they don't leak.

In the review at tom's hardware, they praise the dothan chip for it's efficiency compared to AMD A64.

The only problem is that they are comparing it to the newcastle/clawhammer core.

Why the heck would they compare intel's newest flagship against a really old and sluggish AMD-core which has been outdated for quite some time now. I mean, they are two revisions back!

Sigh.

Maybe they didn't have review samples. Anyway, the new core shouldn't make a huge difference in the benchmarks. It will still use more power than the Pentium-M and it will still be getting creamed by an overclocked Pentium-M (which doesn't use more power than the stock version ).

[darthan]

I seriously question the system power results that Tom's got. They are, in fact, quite impossible. There is a fixed amount of current that flows through a processor to perform an operation. This is a fixed power cost (unless the voltage is changed) to do work. For it to do more work per second it has to use more power (it's overcoming more entropy at the very least) and I'd hope it is doing more work per second if it is overclocked.

Also for all you people whining about wanting a Dothan dual core and not some other stuff, the Merom, Conroe, Woodcrest, Sossaman, and Yonah architectures are all improved (and multi-core) versions of Banias just like Dothan. Intel is switching entirely over to this architecture in the future. I wouldn't expect the desktop derivatives to be nearly as cool as the laptop chips because Intel will tweak them more for speed than for power efficiency but if they run twice as hot as Dothan the hottest chips will have a TDP of only 42W.

See Jon "Hannibal" Stokes' articles at Arstechnica here on the Pentium M or here for information on Intel's future cores ironically as they apply to Apple.