what this seems to point to, is AMD/ATI should have a graphics card that can handle Crysis at Very High Settings @ 1920x1200 .. sometime around mid -2009.

somehow I dont think Nvidia will take that long.

I was struck by the breadth of AMD at this point in time. What, barely a year ago, they were a chip company. Now it's chipsets and GPUs and integrated GPUs and so forth. If anything, the breadth of their presentation underscores the radical changes of the past year and certainly could account for some of their shortcomings. I suspect any business that changes so radically would stumble in the first year of so. It could work out, or it could a Daimler Chrysler. Time will tell I guess. But bottom line: AMD isn't just a small chip company anymore.

I'm not buying anything for the Shrike platform unless it has time-traveling capabilities.

You missed some big news. AMD was a bit sneaky about it of course:

#1. "Bulldozer", AMD's next-gen core, the Nehalem-competitor, no longer shows up in 2009.

Back in July, it was featured in a couple slides, and 2 parts: 'Falcon' (Bulldozer core + graphics, the first Fusion part), and 'Sandtiger' (Bulldozer core server part, up to 8 cores)

Now, Falcon is gone, replaced with 'Swift' (K10 core + graphics), and 'Sandtiger' is gone from 2009 roadmaps. Now there's just Montreal there.

So Bulldozer is delayed to at least 2010, or canceled.

Bulldozer was not mentioned anywhere in AMD's presentations yesterday. Not sure about 'Bobcat', another core from July.

#2 R700. Now showing up in Q2 09 at the earliest. WTH? That's a big slip, if true.

Basically, AMD's hope of challenging Intel and nVidia performance just went out the window for the foreseeable future.

Yeah those plans all call for a 45nm process cores by mid-2009, but they had a hard enough time just getting 65m cores inside of 2007–and even then they're a little buggy (TLB).

I got a laugh out of that AMD vs. nVidia graph projections on page two.

The "glass is half full" mentality also manifested in Meyer's assertion that AMD doesn't need the fastest chips around to succeed. "Many people feel AMD needs to have the best-performing CPU component . . . That perception is false," he bluntly stated.

He means from the sense of diversity in their product line up and the fact that not everyone is looking for top of the line? Simply looking for a product that can do what they want at a price they agree with? I'm sure many average consumers fit the description exactly.

The statement still hides some realities. The better performing parts are in fact dictating where their ASP can land. Competition forcing your product lines to be priced lower isn't a win, even if you can still balance the checkbook.

Edit: Some grammar and structure changes.

Of course I do hope that AMD can thrive and compete, although I still rue the day they bought ATI. However, the thing to remember about ATI is that they easily survived for many years going back to the 1980s by selling low-cost graphics solutions. There's nothing saying that AMD/ATI cannot survive by catering to the bulk of the PC market which buys low-end and mid-range parts. Only a small percentage of users actually buy the higher-end parts.

It may suck for enthusiasts not having competition at the top, but it may give a sigh of relief knowing that you can hang onto your hardware longer since there is less incentive to push the envelope on the part of hardware makers. As long as AMD can remain reasonably competitive on the mid-range, it will still move the hardware industry along, even if more slowly. I might not speak for everyone, but I for one don't like buying a new $300 to $400 video card and then only 12 months later seeing something new come out that is twice as fast at the same price point.

I think AMD needs to have a smaller, lower volume production line in Dresden (or with Chartered in Singapore) that is optimized for cranking out "the latest and greatest" desktop Phenom processors ... the ones in the "Leo/Leo Refresh Platform" for enthusiasts for 2008/2009.

I think it's important for AMD to have all the "good vibes" they can get from the gamers and PC enthusiasts. It's the gamers and enthusiasts who have a lot of influence on everyone else who don't have as much knowledge about PC hardware. Most people don't really want to put in the time to research what the latest "awesome PC gear of the year" is, so they usually just ask that guy at school or work is a gamer or a PC techie. They might even take the word of the guys working at Best Buy somewhat too. Other people just get on the internet and google up 5-10 reviews on PC hardware if they have a little knowledge of PCs. The people who write these reviews are definitely enthusiasts and usually bigtime gamers.

Because of these "influencers" (like the marketing people call them), AMD needs to try to match or surpass Intel's "latest and greatest" processors for the desktop as quickly as possible. They need to get their best benchmark performing processors out in the market quickly ... even if they are ridiculously expensive

Intel's QX9650s are selling for what $1,000+? Not many people are going to buy them, but at least people know that there are some good benchmark performing processors on the way that are about to come down in price in the next year or so.

AMD ought to get those 45nm, DDR3, AM3 "Deneb" and "Propus" processors out in the market as fast as possible! Try to surprise everyone and have them on the market in May if possible. Who cares if they are priced at $1,000 or $1,200!

As long as those expensive processors are getting good "buzz" from the enthusiasts in their town and in the reviews they read online, a lot of the people might make up their mind to buy a PC with that processor in it right there .... for 6 months later ... say Christmas 2008 or whatever. It's like in the old days of Chevy Camaros vs. Ford Mustangs. As long as your product has good "buzz" and people know that "new and improved" models are on the way that you can afford, they'll start saving up for it and planning to get it.

I believe you meant Leo will replace spider later NEXT year.

Errata:

"The Radeon HD 3870 and Radeon HD 3870 are certainly encouraging efforts, though."

In "Conclusion" - closing second paragraph.

Feel free to delete this comment when fixed.

Damn. They're cutting all their good ideas. Last year they had a better outline of what they were going to do. What happened to bulldozer (did I miss it)?

Well, they've got real nice lookin' graphics.

what? no mention of the R700 GPUs until 2009?

http://www.techreport.com/r.x/amd2007analystday/38.png

even the "enthusiasts" are expected to use the R6xx family all through 2008?

"Grose said AMD is aiming for higher margins [gonna be kinda hard to charge more for buggy chips], lower research and development costs [guess we won't be seeing any revolutionary new designs anytime soon... but potentially even more buggy chips], lower marketing, general, and administrative costs [you mean I'm going to see even fewer AMD ads than the zero I see now?], and a return to profitability in the third quarter." [You guys have been inhaling a little too much of that magic blue smoke, methinks.]

Things are looking pretty gloomy in Sunnyvale. I don't think we're going to see a turnaround in 2008. When Barcelona was released, the excuse was always a new stepping just around the corner, First it was BA, then B2, now we have to wait for B3. When is everyone going to realize this is just smoke-and-mirrors and a red herring hiding a much bigger problem. AMD's 65nm process is a bust.

The writing is on the wall, We know the K8 architecture can scale to 3Ghz and beyond since there are plenty of 90nm chips doing just that, but 12 months after the launch of Brisbane, there have yet to be any shipping bins to reach 3Ghz, let alone higher. Couple this with the fact that 2.6 Ghz Phenom 9900 samples are slated for a 140W TDP and it nearly seals the deal. Their process is running near a cliff and they have a nearly insurmountable thermal/power wall to get to 3 Ghz on K10.

The TLB erratum, while obviously a problem, becomes a very fortuitous mishap. From a PR perspective, they can blame the problem on a fixable design flaw as opposed to admitting a fundamental manufacturing/process problem. They now have the "luxury" of tweaking the process for a few more months in the hopes they can get decent binnings. They're also able to keep shipping K8's which are a proven product which are cheaper to produce (smaller dies) and targets a larger market (Quad's are certainly not mainstream).

If AMD manages to ship 3 Ghz Barcelona's by the end of Q2 '08 like they're guiding now, I'll eat crow, but I'm calling it right now. We're never going to see 3 Ghz Barcelona's on 65nm, atleast not until a Shanghai 45nm shrink.