AMD showcases 45 nm silicon - Page 2

Sorry for the typos, English isn't my primary language. I usually use a spell checker but that time I was unfortunate enough to type on an awful keyboard with no spell checker around.

If they are carrying around a wafer with “fully functional” 45nm chips on it instead of packaging and testing them, things are worse off with AMD management than I thought.

Awesome! Is it a wafer full of (what would become) QX6800s?

Nope. I think they were a batch of chipsets. Pretty small-ish die. I do have a 6-inch wafer, too.

Intel did so over 1 year ago with WORKING silicon

I can say with confidence that after reading this tortured thread that Intel's marketing budget will make a significant dent in the coming quarterly profits.

Awesome! Is it a wafer full of (what would become) QX6800s?

Nope. I think they were a batch of chipsets. Pretty small-ish die. I do have a 6-inch wafer, too.

I've done a google search on Typhoon and chipset, but not come up with anything. I'm pretty sure these chips are test chips, produced for process characterization, not an actual product.

This is a good question .... perhaps they cannot get passed the QA for the Opty brand on thier 65 quite yet.

You do you job well JJ, infuse doubt regarding AMD whenever u have the thance, never do that with Intel, always defend Intel when the facts are not correct, you try to do this for AMD but aways with a twist and a "but" somewere, miss key points regarding AMD and aways shift things in Intel's favor no matter what. You have an ridiculouse high post count per day with the overwhelming majotity of posts defending Intel or explaining why Intel is beter. You start postig at first working hours in the morning and sometimes finish posting when working hours are over.

You are just one more example of how low Intel can get.

P.S. Tell me you're not a Intel drone! Please!
Has it occurred to you that Jack is just looking at Intel and AMD and likes what he sees from Intel and is unimpressed by what he sees from AMD? Intel clearly has a superior product, and is concentrating on the kinds of processing that are becoming more important right now. Their long term plans are well thought out, and consist of technologies they have already demonstrated real product capability with, notably MCMs, low power and energy efficient graphics processing. AMD is struggling with poor products and markets openly everything but their next generation of processor. Their plans aren't grounded in their current products and they aren't competitive in the fastest growing segment, low power CPUs.

It is pretty darn easy to get excited about what Intel is doing, particularly if you understand semiconductors. They've knocked my socks off with their execution and strategy. I'm pretty sure Jack has been similarly affected. Obviously, such a situation will make one inclined to give Intel the benefit of the doubt because they are performing so well. However, the bias exists for good reason. It isn't that Jack necessarily wants Intel to do well because he likes the blue man group or thinks AMD sucks. It is that he thinks they are the better performer right now. He isn't a fan that likes the local team because they're the local team. He obviously esteems Intel's recent performance and likes their plans for the future. It is much better than the people who think AMD can do no wrong. They are clearly AMD fans, not technology fans.

Bingo. I have been a die-hard AMD fan for a while, and I really want AMD to start kicking arse again, but facts are facts. Right now Intel is doing good stuff, which nobody can deny. How it came about was strange (the whole Pentium M VS netburst team) and AMD didn't have the resources to split into 2 CPU manufacturing teams, so instead of going down the radical redesign road they are trying to do what they can to stay competitive and make slower tweaks to their design.

Anyway, it really does freak me out to see x2 6000s going for under $300, and I will probably buy AMD just to support the underdog (because you really just HAVE to feel bad for them right now). I feel bad for the people working at AMD and Intel right now because the competition is just insane. But none of this stuff changes the facts, and people always distort reality because of their emotions. Argh.

I doubt AMD can release 45nm CPU's within the next 2-3 months

So basically they're damned if they do and damned if they don't?

I will change my mind on AMD as soon as they show something that tells me they're competitive. I'm not a fan of any particular company. Intel won me over with the Pentium M and is showing that they are moving in the right direction with low power and systems in a package. AMD is not showing signs of performing, so I'm not going to be impressed when they show a test chip and people react to it like its a product.

The thing that really gets me, is that they are going to miraculously speed up 45nm, even though it is appearing as though they are still a year or more behind Intel. Since I know exactly what goes into technology development, I'm not buying that they're going to speed up without making the investment to do so. They aren't investing in manpower and testchips at a higher level, so they won't get there any faster than before.

It is possible that AMD has something going on that will have meaningful impact on the market, but until I see a product that indicates that they're moving in the right direction, I will scoff at stupid stuff like this.

We wont see benches until cards hit reviewers. Why would anyone suspect theyd see benches at a convention? (in reply to bench seekers)

They adopted 45nm faster as they will likely make 65nm a very shortlived program. And you will likely see 32nm adopted early as well to drive the cost of complexity down in layering. What smaller nodes means to AMD is highly significant. And the sooner they get smaller ,the better. That and fusion relys heavilly on faster node transitions, I wouldnt doubt it if they are working at a 32nm sample. Id say expect a dram test wafer from them or IBM as soon as physically possible.

Hopefully before Intel ,but there is more than enough reason to doubt it.

Companies don't just move on to the next technology node because they want to. They have to make investments and do the work. AMD will not move to 45nm just because 65nm sucks for them. They have to do something to get there, and it doesn't look like they are doing what is necessary to gain on Intel's 12 month lead in 65nm.

Well, for one thing, 45nm will offfer better power consumption for them, and higher yeilds, so I don't see why they wouldn't push it out to keep up with intel

In today's market it is a poor product that survives by being aggressively priced at the low end and through clocking to the point of extreme heat and power consumption at full load. It was once a good product. That day is gone.

K8 isn't the whole problem at AMD. Their strategy for the future is predicated upon a shared system bus with individual components in separate sockets on a board. This approach is known to have bandwidth issues. It is most certainly not going to perform as well as a system in a package, where Intel has been advocating and demonstrating competence. Heat may become an issue for Intel, but it appears as though heat is becoming less of an enemy as power efficient techniques are improving very fast.

We will see if AMD is shoring up their processing weakness when Barcelona has been benchmarked and we find out when it is available and how much its going to cost. Then we will see if they have plans to become MCM capable.

In fact, the analysts only can predict in their field, not the technology side. As we are in a CPU forum, we know what "would happen next".

I will not be surprised if AMD will raise the CapEx again in the near future. And this is the way to go for AMD as the cost of CPU, GPU and chipsets should be kept low.

What I'm wondering is; is it possible for amd to completely skip one portion like 45nm or 32nm, and go straight to the next process, such as 32nm or 22nm?

The transition to immersion lithography is a big change. I am wondering where AMD is testing with 45nm. It will be a great news to me if Fab 30 is now ramping down.

The die size difference (118mm^2 vs 219mm^2 for Brisbane / Windsor) makes AMD to push 65nm. And without 65nm, Barcelona will be a dream only (near 600mm^2 at 90nm).

Skipping Rev. G for Opteron is not unpredictable as AMD is going to introduce quad-core Opteron. And from news I have read, AMD does not seem to introduce Next. Gen. Arch. based dual-core Opterons. That makes 65nm extremely important for Opteron series.

Because they still can enjoy the increasing market share from the hot furnaces (Preschott) before the Core Arch. debut.

Not really, amd just didn't want to have to do any more work and if I recall correctly, they even claimed "90nm has plenty of life left in it", boy are they regretting that now

If they would have converted to 65nm there just to stay competitive, they actually would have gained more profits because people would pay more for a better ocing core (the original x2 cores didn't oc much at all) that ran extremely cool, and they wouldn't have ran into the whole 90nm cpus cost too much to manufactur during a price war

Just like robsx2 and bm

Some people just never seen to learn though, and for our and their sake, I hope they get their act together and hire better management

Whatever happens, Let us all bow our heads and pray!
Oh most powerful and merciful God.
Let not one of the two producers of our beloved computer processor
go into the darkness.
Least we have to pay a price ten thousand fold for the product produced
by the remaining one.
And let us be ever wary that only through compitition can we afford
the much needed "Mana" of life for our systems of choice.
And please embiggenate the world wide introweb so that new pron sites
can find their way to us.

AMEN*

See, this is why you and I should take over AMD

And here's two great emoticons for the next time you want to display what goes on in your amd meetings
I may not know the most about gfx and cpu hardware, but I definitely know what will save amd's ass