I'm afraid I tend to agree with Intel on this lawsuit matter. AMD has long had fantastic engineers combined with the absolute worst marketing, business, and executive directors known to man. They've had no less than 3 amazing processor architectures and each time the non-engineering departments within screwed it all up. The production capabilities of AMD have always been an issue. Even if they did gain any significant amount of marketshare, they would never be able to handle it. This shortage has also contributed greatly to their already overly greedy accountants. They've had many opportunities to take the market by storm with low cost cheap athlon64s, but instead they decided to cling to releasing strictly the processors with the highest margins. They made a distinct choice there: marketshare + longterm growth vs short-term money, and they chose the money.

My current rig has an Athlon64, and my next one probably will too. But I just can't help but feel Intel is right in this instance: AMD's problems are and have always been their own. I only wish AMD's engineering team would move over to Intel so that their skills could be fully tapped instead of being bogged down by a horribly idiotic bureaucracy.

Does anyone have an idea how much power will the new Turion will save by using DDR2?

Quad core on server - Expensive, especially given the need for all 4 cores to work (given AMD's design), but maybe. But it's not a new thing - Intel tried adding cache :-D But this would have a very definite effect on performance though.

Tri-core: Never gonna happen on the desktop. Price is pretty much all anyone cares about, so unless AMD somehow slits it's profitable wrists, and loses all profits it's made recently in one quarter, not gonna happen.

Tweaks: How far can you tweak and goose a core before you can't do any more...? The performance benefits were pretty sporadic / not that great for Rev. E anyhow.

Shared L3 cache - Makes sense from a performance standpoint with dual cores and multithreaded code, but again, expensive. In order for the cache to have a large enough effect, it would have to be fairly massive too.

And DDR2 won't make much, if *any* difference, and could easily make performance worse.

Maybe a combination of these things will help though.
-Mole

I don't see any problem for AMD and I was kind of surprised to read how well the Turion does against the Dothan on a level playing field.( Dothan vs. Turion at "laptoplogic.com"). The shrink and new process technoloy should help the small lean company move well compared to the big old dinosaur.
Remember Intels new "built from the ground up processors of the futher will be bits and pieces of the ten year old p3 and failure p4. Their " New bus" will be made by putting two of their "ten year old buses" together.Their new chips will have more cache instead of new innovation. More of the same old crap, Oh yeah the new chips will fit in the same sockets but only on a "new mother board with dual buses". intel will use all of its marketing resources to sell this crap as they have always to Joe six-pack. Joe will need a new ViiV to replace all the dull machines sitting in his cellar and garage.

I'm surprised no one's mentioned "Two-rion" /kicks self

With the shrink to 65nm, I wouldn't be surprised to see amd selling something like Presler. wtf? you say? Well, Presler is two chips on a MCM package.

AMD could simply put two dual-core 65 nm A64s on one MCM and sell it as a quad-core CPU. It can even use chips with one faulty core and sell a 3-core CPU, alleviating one of the main drawbacks of dualcore chips - yields. Heck, it could even put two faulty chips on one module and sell it as a standard dual-core chip. The bastard may even OC better due to the non-functional silicon absorbing and dissipating heat for the working cores.

Both chips could be connected with a Hypertransport bus, ideally, there'd be a 4-channel memory bus so each chip gets 2 channels of memory. Kinda like 2-way Opteron systems (if I understand it correctly). But we'll have to see if M2 is ready for such an implementation.

As for PF2, I have no doubts the chip will perform really well. It's just that it's rather...boring from an architectural perspective. So far, anyway. When intel releases more details, I might perk up my ears a bit. But their current focus on marketing (V//V) does make me tune out, regardless of how talented their engineers might be. Ironically, due to their recent history of marketing over performance, it's going to be much harder to sell the idea of an intel solution to me, because I'll always be looking for the catch. Who says consumers don't catch on? Maybe intel should get David Manning to write them some good reviews :P

I don't know what to expect from AMD from 2006. However they are being quite ... secretive?

Of course 65nm is now operating at AMD, so processors using it will probably appear from early in 2006. That'll give AMD a die size advantage over their 90nm dies, or give them room to double the L2 cache sizes on their high-end processors. This will be a nice way to get some 'free' oomph in some of the SPEC benchmarks as well which they might need come Merom and server variants thereof.

Quad core? 90% chance of seeing prototype processors demonstrated during 2006, and then released in 2007 - the issue being the marketing number used - Opteron 280/285/290 will probably be released by 2007 for dual core, so where do AMD go beyond that? Opteron X4? At least they can use Athlon X4 in the high-end consumer area.

New core design? I don't think they'll release anything during 2006. However they might release information for a new core design they'll release in 2007 or 2008.

We know that there will be new sockets and DDR2 during 2006. Other aspects aren't known yet. FBDIMM in S1207? How many DDR2 channels in S1 and M2? HyperTransport 2.0? PCIe? Graphics (unlikely)?

Maybe AMD will get together with (a) chipset provider(s) and create an MCM with a Sempron and a Chipset on it. For example, a 256KB L2 Sempron with an SiS northbridge and southbridge on the same package. Note that this way you can remove all the major interconnect pins, get things closer together, and cool using a single heatsink. Maybe AMD will merely create a 'System On A Chip' socket specification, with pin-outs for DDR2, PCIe, Graphics, Audio, USB2, Firewire, Ethernet and so on. This would be a really neat idea in my opinion, but not for everybody. Great for non-upgradable media centres and the like however.

Maybe AMD will take a page from intel's book and crank up the clockspeed and FSB along with minor tweaks.

Looking at a lot of the things that Intel is doing to up the IPC of Fred (aprropriating Damage's name for Intel's Architecture Of The Future), such as a wider pipeline, combining multiple instructions, and the like, it would seem to my decidely non-computer-scientist mind that AMD could throw a lot of those tweaks at some revision of the A64. Or, they could take the approach they did towards the end of the Athlon XP era: slash prices (made bearable by smaller processes to manufacture the chips on) and reposition themselves as the value leader until R&D can cook up another Intel killer.

I hope amd doesnt come out with anything new for awhile atleast . I just bought a X2 4200+ and 7800 GTX . I dont really want to upgrade again but it seems like i always have to have the newest latest greatest fastest whatever it may be . I would be money ahead if they dont come out with any upgrades for awhile .

You think AMD might have a lot riding on its suit against Intel and is thus not really working on future technologies much?

Multi-core is still going to take a while before it is utilized well in the desktop space. Of course the quicker the market penetration the better. I think that AMD would have to do something more then just make a 3/4 core version of their processor available when Intel debuts their new cash cow.

My personal belief is that any changes AMD makes in the next year and a bit will be minor core revisions. Probably better branch preditiction and possible DDR2 support.

I think that Intel will take the performance lead for a bit and there is little AMD can do about it in the short term. However we can all hope that they have something cooking in R&D.

No matter if you like Intel or AMD, heated competition will only benefit the consumer more.

Editor: I believe that should read "cat bird's seat".

AMD is playing it's cards close to it's vest and that's okay for now. The Hammer core has been an unmitigated success so it's not surprising to see them milk that cash cow as long as they can. The big thing to come is PCI-E on the die. I'm not sure why they wouldn't be touting this. It will give AMD a tremendous I/O scalability advantage on the server side.

im first. Go amd