AMD showcases 45 nm silicon

[eregular]

Monterey (CA) – At an event held in Monterey today, AMD showed a wafer with “fully functional” 45 nm silicon for the first time.

According to chief technology officer Phil Hester, the 300 mm wafer code-named “Typhoon”, shown in public for the first time today, holds 45 nm dies combining SRAM and logic. The executive said that AMD’s 45 nm process is on track and recent notes from Intel that AMD is facing yield issues “are rubbish” and “wishful thinking on their side”.

The 300 mm wafers with 45 nm structures are expected to ramp in production at the end of 2007. First 45 nm processors are expected to be commercially available by mid of 2007.
More details tome come.
CLICK ME!!

45nm amd x2's with L3 cache and maybe clocking around 3ghz...*drool* oh yah and of course it'll be 65-89watts.

 

[r0ck]

http://www.intel.com/pressroom/archive/releases/20060125comp.htm

Intel demoed something similar over 15 months ago.

 

[fidgewinkle]

I'm pretty sure you meant that the first commercially available processors will appear in mid 2008, not 2007. I'm pretty sceptical, as they are showing us a freaking test vehicle. You know a company is desperate when they're showing test vehicles to the public. This is making me wonder if AMD is afraid that Barcelona will be a flop and they will be struggling to get through the year. If you're an AMD fan, this is not cause for celebration. It is cause for worry.

 

[Jakc]

http://www.intel.com/pressroom/archive/releases/20060125comp.htm

Intel demoed something similar over 15 months ago.

Didn't AMD compete just fine so far, using a higher nm than intel?
This would mean that AMD is slowly catching up...
Or would that be too optimistic?

 

[xpresso]

http://www.intel.com/pressroom/archive/releases/20060125comp.htm

Intel demoed something similar over 15 months ago.

So this officially confirms that AMD is about 15 months behind Intel on 45 nm. My estimate has been a range 12 to 18 months

 

[xpresso]

My guess they're rushing 45 nm due to Barcelona not good enough to compete with Intel's 45 nm offerings.

 

[ajfink]

Isn't that a cute lil' thing, then?

 

[Iam9376]

http://www.intel.com/pressroom/archive/releases/20060125comp.htm

Intel demoed something similar over 15 months ago.

Didn't AMD compete just fine so far, using a higher nm than intel?
This would mean that AMD is slowly catching up...
Or would that be too optimistic?

Yes..when their microarchitecture was better implemented. We can't say for certain about their newest architecture, but one is inclined to believe it doesn't compete*

* pure speculation based on lack of data.

 

[zornundo]

My guess they're rushing 45 nm due to Barcelona not good enough to compete with Intel's 45 nm offerings.

I disagree, the reason why they would rush it out is to catch up for better yeilds

better yields = higher gross margin %

looking at their Q1 2007 results, AMD needs all the gross margin help it can get!

 

[Iam9376]

My guess they're rushing 45 nm due to Barcelona not good enough to compete with Intel's 45 nm offerings.

I disagree, the reason why they would rush it out is to catch up for better yeilds

I think it's more marketing than anything. Intel is ready to ship 45 nm chips, AMD is desperate. A headline like this looks better than "AMD almost ready to produce 45nm". Just a thought.

 

[Slobogob]

The 300 mm wafers with 45 nm structures are expected to ramp in production at the end of 2007. First 45 nm processors are expected to be commercially available by mid of 2007.

Given the AMD typical delays a realistic market entry would be around Q1 2009.

 

[ajfink]

http://www.intel.com/pressroom/archive/releases/20060125comp.htm

Intel demoed something similar over 15 months ago.

Didn't AMD compete just fine so far, using a higher nm than intel?
This would mean that AMD is slowly catching up...
Or would that be too optimistic?

Not really --- costs wise, Intel has enjoyed the cost advantage significantly, and we see this in the bottom line results of the price war.

Also, you are purporting the same flawed argument seen by most AMD fanatical base -- equating process with performance of an architecture. Performance is based on a combination of both architectural computational efficiency (IPC) as well as clock speed (design + process). Put a crappy architecture (Netburst) on a superior process (Intel's 65 nm) and you will still be strattled with poor IPC.

AMD's current 90 nm is no match for Intel's current 65 nm.... as is their current 65 nm no match for Intel's current 65 nm....

On a node for node basis, Intel's process technology has always exceeded AMD's technology on raw process performance fundamentals... i.e. Vt, Idsat, Idlin, etc.

Jack

Arguably not a match. They do enjoy similarly low power usage, it doesn't clock as high.

But yeah, Intel's process steps are better.

 

[NMDante]

Umm....

If this was a wafer with 45nm "fully functional" die, why didn't they show the actual die in a final package?

I've seen plenty of wafers, and believe me, I can take a picture of a wafer I have in my house, and claim that it's a fully functional whatever.

I guess I just don't see the excitement over showing a wafer with die over an actual product running. Sure, it's 45nm, but again, I can show a wafer and claim the same thing.

 

[BaronMatrix]

I'm pretty sure you meant that the first commercially available processors will appear in mid 2008, not 2007. I'm pretty sceptical, as they are showing us a freaking test vehicle. You know a company is desperate when they're showing test vehicles to the public. This is making me wonder if AMD is afraid that Barcelona will be a flop and they will be struggling to get through the year. If you're an AMD fan, this is not cause for celebration. It is cause for worry.

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

 

[djgandy]

First 45 nm processors are expected to be commercially available by mid of 2007.

So they will be released before Barcelona?
haha what a huge fail.

EDIT....

Sounds like another annoucement to stop their stock dropping.

Also check the precise wording. I hate to be pedantic but..

a wafer with “fully functional” 45 nm silicon for the first time.

A wafer? Surely you would show a chip. Maybe Jack can clarify, but showing off a wafer doesn't mean anything is functional?

the 300 mm wafer code-named “Typhoon”, shown in public for the first time today, holds 45 nm dies combining SRAM and logic.

So once again they showed a wafer. Was there any proof the components worked?

The 300 mm wafers with 45 nm structures are expected to ramp in production at the end of 2007. First 45 nm processors are expected to be commercially available by mid of 2007.

So they are going into production at the end of the year but available before they go into production.

This is either a huge error on AMD's part or a huge error on toms part. I'm not sure what I which I would prefer it to be.

 

[Jakc]

Uhm.
Almost all the athlon x2 processors currently out there are 90 nm, whereas C2D's are 65 nm.

Only the brisbane is 65 nm, but that one does not even run at the top speeds.
And saying that current AMD 90 nm is no match for current intel 65 nm is an overstatement. AMD has been keeping up with intel fairly well, except in the highest end.

What this means? This means that AMD has been keeping up fairly well, using an older architecture(athlon 64 vs C2D), but also older technology(90 nm vs 65 nm) than what intel is currently using.

AMD is not doing bad at all from what I can see.
Of course the C2D are awesome architecture, and intel does a great job pushing technology to the next stage, but it does seem to me the gap between the two companies is getting less and less all the time.

Of course quite a lot will depend on barcelona.

 

[djgandy]

First 45 nm processors are expected to be commercially available by mid of 2007.

So they will be released before Barcelona?
haha what a huge fail.

Sounds like another annoucement to stop their stock dropping.
Now, that's a misprint, he meant 2008

Okay Makes more sence now. AMD has been so quiet I was literally waiting for them to surprise us with something huge like this. Maybe not

 

[Jakc]

False, amd also has a single core 65nm line, and don't forget that the amount of 90nm cpus to 65nm cpus isn't as much as it used to be, at least before the 5400-6000 anyways

From what I can see, most is 90 nm at this point.
http://en.wikipedia.org/wiki/Athlon_64_X2

 

[ajfink]

I'm pretty sure AMD is shipping as many / more 65nm processors now than they are 90nm. Also, 65nm mobile processors are reaching the channel as well.

 

[ajfink]

I know, the 65nm turon is supposed to debut any day now, though they seemed to have taken their sweet time with it

It took them a little while to get the process good enough for mobile processors, I assume.

 

[djgandy]

False, amd also has a single core 65nm line, and don't forget that the amount of 90nm cpus to 65nm cpus isn't as much as it used to be, at least before the 5400-6000 anyways

From what I can see, most is 90 nm at this point.
http://en.wikipedia.org/wiki/Athlon_64_X2

well wikipedia isnt the best source for latest tech news. It's a good resource for a lot of things but usually lags behind and even lacks in the specifics that are going on.

 

[djgandy]

I know, the 65nm turon is supposed to debut any day now, though they seemed to have taken their sweet time with it

It took them a little while to get the process good enough for mobile processors, I assume.

Another good logical point. Points like these get us called Intel fanboi's but when you think it through, if Intel are releasing solid tech on the mobile platform (core/penryn), "the sky is the limit" when it comes to desktop.

It reinforces the december 65nm paper launch. Not many processors appeared. The yields were not good enough. Still no 65nm mobile parts.....What do we conclude...? I see the chain as desktop->server->mobile. Mobile being the best quality parts considering the price they are sold and the low voltage they operate at.

I write this from a turion laptop which is slower then a 2800 barton

 

[r0ck]

Where are the 65nm Opterons?

 

[xpresso]

My guess they're rushing 45 nm due to Barcelona not good enough to compete with Intel's 45 nm offerings.

I disagree, the reason why they would rush it out is to catch up for better yeilds

Nah, not so fast, you can't have yields yet. That 45 nm process isn't really working yet until we see some published benchmarks... Thinking about it now, this is just a wafer and if I had to guess, this is a pure publicity stunt :?

The real significance will be when the dust settles post Barcelona lunch to see how it performs in comparison to C2Q & Penryn. AMD does not have 45 nm toolset, so most likely this is a test chip ran at East Fishkill (IBM). Hector needs all the publicity he can get to slow down the loss of market share. If AMD loses another 3% and more, they will be in real trouble. Once we see barcelona benchmarks, we can tell why this shiny 45 nm wafer was revealed today

 

[Mex]

Where are the 65nm Opterons?

This is a good question .... perhaps they cannot get passed the QA for the Opty brand on thier 65 quite yet.
To build on your answer...AMD is in a rather delicate situation. Opteron is AMD's most important product and logically the one they should take the least risk on in order to avoid jeopardizing sales. The last thing they need to do is gamble their flagship processor on a (potentially) unproven and low-yielding process. After all, nVidia and ATi do the exact same thing with their high-end GPUs.