Detailing is good, we need that since leaving people in the dark is not good for business.

However, a bit more followthrough is required. 64bit has taken too long to get to the desktop. Nevermind this "when there is an OS" BS, they didn't wait for support for SSE. Intel releases abilities and then brings the software to support it. So the current idiocy over releasing 64bit is the fact that either they suck at it, or something functionally is wrong, or both. Its a smokescreen and a failed promise, imo, and one that has turned away a ton of customers (enthusiasts at minimum).

Let's get a plan, a map and then a nice upgrade path and a bit more fire under their pants to get things TO THE MARKET. That is unless they'd prefer to become a commodity chip company and let everyone else lead the industry.

Intel is screwing around and the customers are paying the price. Enough of them wake up to that reality and you can wave bye bye to even a huge company. It takes 1000 voices to get a smidgen of a good reputation and only a single voice to throw a lot of reputation in the trash.

Intel has a LOT of proving to do, and it has to be in more than just speeches and marketing fliers...

I think these will be the biggest crap if will go desktop, since only some server and workstation applications benefits from multi thread.

Now for server and workstation market this will be a huge leap, since motherboard designs will be simpler and cheaper.
I think there might be a 50% price cut down over current dual systems, unless dual core processors cost the same of two processors, but even so, the motherboard will go down.
2->1 | 4->2 | 8->4

Like some one already said here, this is the way of making the issues they are having in design and manufacturing microprocessors and pass them over to software.

Lets see what they bring to the table in IDF (Sept).

Anemone, you don't know shit if it was from your arse, and held under your nose.

Intel is delibrately screwing around with EMT64/AMD64 as long as possible.

Itanium architecture is what they're really aiming for. You don't spend billions on R&D, to develop a 64bit architecture to throw it away for some competitor's implementation. That's just shooting yourself in the foot. Why do you think WinXP for AMD64/EMT64 is delayed and will be sold in OEM form? (ie: you have to buy it with a piece of hardware).

AMD64 is nothing to them, and if they can delibrately delay AMD64 and fuck it up, it paves the way for Itanium and variants to be in all markets. (server/workstation/desktop).

The overall process means that AMD64 is getting dragged into the gutter. This is what Intel wants. EMT64 is not advertised heavily. The feature is there because MS isn't gonna create another 64bit variant of Windows alongside AMD64, the don't have the manpower for it. Intel is essentially put into a position where they either add EMT64 or loose big OEM customers like Dell.

So they slap it in to satisfy OEM customers.

Intel has proven in the past that their "cheats" do work. (HT and SSE2)

Current example, Itanium II's IA-32 Execution Layer is an example of improving Itanium II's 32bit performance. Its far better than Itanium I in that, it now performs clock for clock equal to a Xeon. Previously 32bit was done with on die hardware...It was shit...Then they designed a software layer that runs on IA64...It whooops Itanium I's ass.

All Intel needs to do is scale to high speeds and provide more optimisations to their "Execultion Layer". This will not need software developer's to optimise 32bit apps for the Itanium II.

I won't be surprised if EMT64 is very even to AMD64 in performance...And not costing a cent more. (AMD's overcharging is gonna bite them in the ass when Intel is offering EMT64 versions of their CPUs at no extra cost. See how Intel is trying to say: "Look, we're gonna give you 64bit for free, while AMD is chargin a premium for it.)

The next fight is Itanium III vs K9, both true 64bit platforms. None of this half-breed 32bit/64bit "transition to 64bit" CPU bullshit.

I'm not a fan of either company. Intel and AMD can treat their customers like shit, because its business. Being a fan of a company is just plain stupid, your just asking for a serious ass fuck from either company. Just consider their products as tools for your uses and nothing else.

Hey "I"...Have even used an SMP setup before? Because it sounds like you have no clue either. Another no clue looser that hangs out at Xbitlabs.

SMP (and dual-core) allows you multi-task far heavier than some piece of shit P4 with HT, where no apps are multi-threaded. You can play Doom III while encoding a video. (Using processor affinity, you can assign each CPU for each tasks).

Its not that hard for software developers to re-organise their coding and compile it to be multi-threaded. Heck, Intel is already providing documents on how to do this. Its freely available on the web.

The only disadvantage of dual-core is that it won't be scaled to similar speeds as single-core CPUs. The heat problem is easily solved by using the Pentium-M architecture but with dual-channel memory and faster bus speeds.

Ok, I've read and I'm learning, though you don't seem to disagree that Intel is screwing around. Itanium on the desktop? You're talking about 2010 like it was here now, imo. What makes you think we should even be talking like it was a performance comparison, when it won't scale down to that for at least 2-3 generations, or in other words likely till production lines are running at 45nm.

Granted you are quite correct that the second generation Itanium products execute much better 32 bit code.

SMP is a classic industry case of when you can't go higher serially, you branch out and go parallel to see if you can get enough speed parallel to exceed in performance. I have no opinion on whether it will bring more benefits than issues yet, there are rocks in both sides of the river.

I'm not entirel clueless :) but I'll take your 'education' with the merit its worth. Interesting read :)

And on the current generation abilities on the desktop today or near today, it would certainly help Intel if they were to release 925XE in Sept. It's probably not possible, but it would help.