AMD Bulldozer pushed out to 2011

[Shintai]

I was talking about specific Itanium design,not vliw architectures in general...

Itanium is the dominant VLIW design. Itanium keeps its revenue growth. Hardly sinking...Sparc however might be sinking soon.

x86/x64 is simply too limited for the long term. We can barely go beyond 3 issue wide. We basicly need things like SMT not to have the execution units to sit idle.

[xVeinx]

The basic point is that we won't be seeing much from AMD for the next 2 years. A die shrink with associated tweaks could increase performance more, but the roadmap is pretty quiet, and I imagine that AMD will continue in that until Bulldozer. There is only so much "left" in an architecture in terms of optimization, so for AMD to focus on a new architecture makes sense. AMD needs a new architecture and a new paradigm.

[gOJDO]

What if in 2011 AMD postpones Bulldozer to 2013+?

[v_rr]

What if in 2011 AMD postpones Bulldozer to 2013+?

What if we keep the spam at minimum

[Sparky]

How did itanium get dragged into this?

Screw itanium, has nothing to do with the subject

And from the looks of things, nothing has gotten "pushed out" just some are jumping on this "news" for some weird reason

[Chad Boga]

How did itanium get dragged into this?

A clueless attempt to slag off Intel.

And from the looks of things, nothing has gotten "pushed out" just some are jumping on this "news" for some weird reason

It is no big deal to have raised this as a news item and I would say that the "jumping" was done by the usual suspects who demand nothing less than completely gushing, never critical assessments/comments about AMD.

[Solus Corvus]

It is no big deal to have raised this as a news item and I would say that the "jumping" was done by the usual suspects who demand nothing less than completely gushing, never critical assessments/comments about AMD.

It's a little less one sided then that. It's no mistake that bad AMD "news" was posted by terrace or that Shintai/Informal are tied for the highest number of posts in this thread. Both sides of the AMD/Intel spectrum have made themselves look pretty bad with this thread.

And seriously, it's not news. Anyone who was paying attention knew this back in november.

[FlanK3r]

Yep. First AMD 32nm CPU. Deneb is in for about 2½ years more in the frontline fighting Yorkfield, Nehalem, Westmere and Sandy Bridge.

Bulldozer is out about 3-6months before the first 22nm Intel CPU (Ivy Bridge).

Do u know all about Deneb chip and potencial Denebs? And x6 based on Istanbul, hm? Just waiting man and u will see...

Orochi come in 4Q 2010 or 1Q 2011 to shop
Intel tick tock?Hm?....

Westmere 32nm 4Q 2009
SandyBridge 32nm same quartal as AMD Orochi
Ivy Bridge not at 2Q 2011

[Extelleron]

Do u know all about Deneb chip and potencial Denebs? And x6 based on Istanbul, hm? Just waiting man and u will see...

Orochi come in 4Q 2010 or 1Q 2011 to shop
Intel tick tock?Hm?....

Westmere 32nm 4Q 2009
SandyBridge 32nm same quartal as AMD Orochi
Ivy Bridge not at 2Q 2011

Orochi will apparently sample in late 2010, actual product launch cannot be expected until mid-2011. And that is AMD's current projection....given their track record it is not impossible that Bulldozer will not appear at retail until late 2011. Bulldozer will compete with Sandy Bridge for a little while but it will compete with 22nm Ivy Bridge for most of its lifespan, assuming a 22nm shrink from AMD sometime in late 2012.

AMD has nothing faster than Deneb planned for the desktop market, until Bulldozer (approx 2.5 years). A six-core Istanbul part may appear on the desktop but I doubt it will ever be a mainstream part. Anyway it will do nothing once Intel has Gulftown out, much less Sandy Bridge.

AMD's product portfolio for the next 2 years is going to be the low end. I would expect that Propus will be the focus of much of AMD's lineup with 1-2 Deneb SKUs at the top end ($150-200), with Regor and Propus-based tri-cores forming the <$100 market. Deneb's die size is too large for it to really be worth it on the desktop IMO at least, considering the kind of ASP AMD is going to be getting by 2010-2011. It is going to be a rough two years, that's for sure.

[v_rr]

Some people here need a time machine. Live the present not the future

[terrace215]

At least we have hardware.info making fools out of themselves today....

[largon]

Or, perhaps it is Xbitlabs who is making a fool of themselves.
No, really, these sites are only reporting what people are saying. Some (sites and people) are just more informed than others...
And considering the timetable of these statements and positions of the persons giving them, one would think the CEO (Meyer) has a better understanding what the company will be doing and when, compared to a random PR person.

But, hey, let's just all keep on bashing... It has worked this far.

[vietthanhpro]

What if in 2011 AMD postpones Bulldozer to 2013+?

I think !
K6+K7+K10--> K11 Bulldozer.

[saaya]

itanium is the future... in some distant parallel universe maybe... come on, i know that some people at intel still think itanium will take over one day, but thats ridiculous...
they are pulling graphics back ont cpus by building an x86-64 based gpu emulator that will then later be merged with the cpu to then run on itanium? that doesnt make sense..

build a x86-64 cpu to emulate a gpu
build a itanium cpu to emulate a x86-64 cpu?

thats just silly... when is still supposed to happen anyways? 2025?
intel thinkis it has itanium as a backdoor option to take over the market, and it was a great plan back then but it just never worked, and it will never work for the same reasons... companies are run to make money, anybody at intel trying to steer them towards blowing even more billions to shift away from x86-64 to itanium with is completely unecessary would lose support so quickly hed be replaced within months.

[savantu]

I think !
K6+K7+K10--> K11 Bulldozer.

K10=K5 x 2

[Sparky]

K10=K5 x 2

yeah, sure

[Rammsteiner]

Some people here need a time machine. Live the present not the future

Skew their brains with a 2yr delay

K10=K5 x 2

, stupid post.

Im quite curious for Bulldozer, but well, anything could happen which might delay or advance it for a quarter. Im looking forward to it but comparing it to a possible release from its competitor, come on We cant even predict on a solid base what's going to happen in 6 months, let alone end '10/ start '11,

[gosh]

Maybe they have lowered the priority for bulldozer because AMD has seen that they can improve Phenom much more. I think this is a good thing because it will mean that the AM2 socket will last longer. Also the need for bulldozer performance isn't needed for most PC users. Spending a lot of money that isn't going to sell that well, how smart is that? Better to wait as long as you can and build bulldozer as with the best possible design.

I have been using Phenom II 940 in vmware. Phenom II is a huge improvement in speed. It is almost as you don't notice that you work in a virtual machine compared to work on the main machine. Maybe this speed improvement is because RVI.

[Chad Boga]

Maybe they have lowered the priority for bulldozer because AMD has seen that they can improve Phenom much more.

It is more likely that they blundered the development process and had to either start again or do a substantial redesign, necessitating the mooted timeline.

[terrace215]

It is more likely that they blundered the development process and had to either start again or do a substantial redesign, necessitating the mooted timeline.

Yep.

Part of that might have been related to the performance numbers of EP and EX Nehalem. Those being so strong, design #1 might have been a non-starter performance-wise, especially projecting out another couple years.

[demonkevy666]

I don't care, really.

1 take more time to design new core. it can still use L3 cache and connect them like a K10. refinement in latency's in other parts too.
2 make sure it works right, not too hot, not power sucking, no tlb errors.
3 do not get pushed into a rushed product again, work with there schedules and not others schedules.

[terrace215]

one would think the CEO (Meyer) has a better understanding what the company will be doing and when, compared to a random PR person.

Clearly you have little experience with the AMD executive habit of making positive-spin statements that get misinterpreted (by design), but then, when the misinterpretation gets too broadly published, AMD execs get a little nervous about the whole lying-to-investors thing, and send out a minion to clarify.

Happened all the time on their investor CCs. They deliberately confuse "profits!" with "operating profit", or try to confuse "starting production ramp" with "product shipments" or "availability".

And here we had Dirk spinning the 32nm schedule in the usual AMD way, but when several folks ran with it and talked about AMD 32nm *products* in H2 10 (not what Dirk said, just what he wanted you to think), a week later, the PR spokesman is dispatched to say, "uh, no, we mean, um, products in 2011, thanks!"

Why AMD execs can never help themselves from overpromising/underdelivering is beyond me. They might try it the other way around for a change, though I guess that is easier to do when you're ahead.

[gosh]

It is more likely that they blundered the development process and had to either start again or do a substantial redesign, necessitating the mooted timeline.

Moving good solutions from bulldozer design to Phenom and extendending lifespan seems more logical. Don’t think that AMD has the economic margins to "play" with new technologies. They need to be in control.

[Chad Boga]

Moving good solutions from bulldozer design to Phenom and extendending lifespan seems more logical.

Logical? More like highly improbable.

When has such a thing happened previously?

Don’t think that AMD has the economic margins to "play" with new technologies. They need to be in control.

Recovering from a misstep isn't "playing".

[gosh]

Logical? More like highly improbable.
When has such a thing happened previously?

If I am right that is what AMD has always done. They make small enhancements all the time (think it is called new "revisions" that get new tweaks). Maybe we soon will se six core Phenoms that fits in the AM2 socket.

Recovering from a misstep isn't "playing".

"misstep" was your conclusion. You have no clue what or why they do things. There are a lot of things that decides how a company will advance.