IBM claims 22-nm SRAM success

[Idontcare]

IBM and its joint development partners Advanced Micro Devices Inc., Freescale, STMicroelectronics, Toshiba and the College of Nanoscale Science and Engineering, have claimed they have developed the first working SRAM cell implemented in a 22-nm manufacturing process. The cell was built at CNSE's 300-mm research facility in Albany, New York.
http://www.eetimes.com/news/se...HA?articleID=210101316

Good news for sure, although I won't be surprised if Intel shows off a 300mm wafer loaded with 22nm node sram at IDF here shortly.

 

[KingstonU]

Any news like this is exciting for competition sake. Correct me if I'm wrong, but AMD is scheduled to release 32nm and 22nm in 2010 and 2012 respectively. Compared to Intel with 2009 and 2011 respectively. So 1 year behind as always. I just worry that with AMD's woes of late that they will start getting even further behind than that. Does this sample mean AMD is still on track?

 

[Idontcare]

Well node's and timelines are a tricky business full of smoke and mirrors.

One company's node can have no competitive relevance whatsoever to another company's node. The concept of a node is a very very loose definition/label at best. Like GHz...means nothing without more information (such as IPC, among others).

So you have to be careful in getting what you hope for...would you prefer to see AMD release their 32nm node the same year as Intel only to find out the xtor's performance metrics are on par with Intel's 45nm node's xtors?

Or would you rather see AMD spend an extra year optimizing more aggressive xtor targets at 32nm and release a more competitive process technology?

For instance even with the near 1yr lag AMD's 45nm has on Intel, the performance metrics of AMD's xtors will no doubt still lag those of Intel's because Intel has HK/MG and AMD will not (at least initially, and possibly for the entire node life).

Where this all gets kinda skewed in a one-dimensional analysis is the lack of budget context. Intel's process development budget literally dwarfs AMD's development budget. The viewpoint should not be one of "I can't believe those slackers continue to lag a year behind Intel, node after node" but rather the viewpoint really ought to be "it is really impressive that with 1/4 the resources AMD somehow manages to stay within drafting distance of the sprinting pace that Intel's budget allows them to maintain".

And on the flip side, for Intel shareholders they have really got to wonder if R&D dollars are being put to work in as efficient a manner as AMD's management does. Tossing 400% more money into the R&D pit and only managing to maintain a 1-2yr lead over a company 1/10 your market cap and with red ink all over the place isn't exactly a testament to a well managed R&D policy, it just means you have more than enough cash to be liberally free with it and not worry too much about the accountability.

If Intel doesn't increase their lead over AMD by at least another year when the 22nm is released (say Intel releases in 2011 and AMD releases in 2013-2014) then I'd be a little irritated as a shareholder that Intel spent all that much more money than AMD and merely maintained their 1-2yr lead.

 

[piesquared]

Another article here:

http://www.tgdaily.com/content/view/38941/135/

This part is very interesting:

IBM said that it is on track with its 32 nm process and promises that it will use a ?leading 32 nm high-K metal gate technology that no other company or consortium can match.?

 

[KingstonU]

Yah I know what you mean about the node not being the whole story, like AMD 90nm versus Intel (first gen) 65nm. But node is a big advantage when you are trying to compete on performance, and also important is how much less a smaller node costs to manufacture and thus being able to make more profit or increase price/performance. It also increases yields doesn't it? (which means more chips per wafer pass quality checks thus less waste right?)

That's a good point about HK/MK, I forgot that AMD is not going to have this feature ready until 2nd gen 45nm or even 32nm, which also helps with performance and power consumption that Penryn shows over Conroe.

 

[KingstonU]

From the link above:

"IBM and AMD first at 22 nm, challenge Intel?s manufacturing lead:

IBM and its chip development partners made a stunning announcement today, apparently beating Intel in the successful production of the first functional 22 nm SRAM cell. 22 nm processors are still three years out in the future, but IBM?s news is a good sign that chip manufacturer will be able to easily scale to this new level by the end of 2011. It appears that, for the first time in several decades, Intel may have to put some extra time into its research and development efforts to make sure it can keep its manufacturing lead at 22 nm and beyond."


This is obviously optimistic and sounds like marketing hype but are they saying they can have it ready and available at Newegg for 2011? (note I said they are saying, not that I think it's likely). Or does that mean that it will be read for production by 2011?

 

[Idontcare]

Originally posted by: KingstonU
From the link above:

"IBM and AMD first at 22 nm, challenge Intel?s manufacturing lead:

IBM and its chip development partners made a stunning announcement today, apparently beating Intel in the successful production of the first functional 22 nm SRAM cell. 22 nm processors are still three years out in the future, but IBM?s news is a good sign that chip manufacturer will be able to easily scale to this new level by the end of 2011. It appears that, for the first time in several decades, Intel may have to put some extra time into its research and development efforts to make sure it can keep its manufacturing lead at 22 nm and beyond."


This is obviously optimistic and sounds like marketing hype but are they saying they can have it ready and available at Newegg for 2011? (note I said they are saying, not that I think it's likely). Or does that mean that it will be read for production by 2011?

Well sadly this is where you and I run into the limitations of ignorant journalists who like to spin and weave more into a press release than was actually there to begin with.

Here is IBM's press release, note there is zero mention of timelines or production years (such as 2011) or anything about dominating/threatening Intel's lead.

http://www-03.ibm.com/press/us...pressrelease/24942.wss

I should have forgone the EETimes link and just went straight to the source to begin with. EETimes has really gone downhill this past year.

 

[soccerballtux]

So you have to be careful in getting what you hope for...would you prefer to see AMD release their 32nm node the same year as Intel only to find out the xtor's performance metrics are on par with Intel's 45nm node's xtors?

Actually, the first, because then fab price is similar, and AMD can then cut prices more.

As it is now, they're fighting for a shred of similar performance, and since they're still on 65nm, price cuts hurt them more.

 

[Soulkeeper]

hopefully more actual results come from this news than strained silicon

 

[Idontcare]

Originally posted by: soccerballtux

So you have to be careful in getting what you hope for...would you prefer to see AMD release their 32nm node the same year as Intel only to find out the xtor's performance metrics are on par with Intel's 45nm node's xtors?

Actually, the first, because then fab price is similar, and AMD can then cut prices more.

As it is now, they're fighting for a shred of similar performance, and since they're still on 65nm, price cuts hurt them more.

I can't fully tell what you are trying to communicate with the first sentence of your post...are you speaking to the die-size reduction entitlement that would come from the simple scaling effects of a new node?

If so then I absolutely do agree, and that is where AMD needs 45nm (even without HK/MG) sooner rather than later.

Understand the gist of my posts are usually generic as to the IDM, but use specific IDM names (AMD, Intel) when trying to get the point across. In general most IDM's prefer to delay the release of a node if the trade-off in doing so is that it nets them a performance/cost competitive node in the end.

Although extreme cases will always invalidate such tradeoffs, being 1-2yrs behind your chief competitor is simply a fatal business strategy in any industry, semiconductor or other.

 

[bryanW1995]

Originally posted by: Idontcare
Well node's and timelines are a tricky business full of smoke and mirrors.

One company's node can have no competitive relevance whatsoever to another company's node. The concept of a node is a very very loose definition/label at best. Like GHz...means nothing without more information (such as IPC, among others).

So you have to be careful in getting what you hope for...would you prefer to see AMD release their 32nm node the same year as Intel only to find out the xtor's performance metrics are on par with Intel's 45nm node's xtors?

Or would you rather see AMD spend an extra year optimizing more aggressive xtor targets at 32nm and release a more competitive process technology?

For instance even with the near 1yr lag AMD's 45nm has on Intel, the performance metrics of AMD's xtors will no doubt still lag those of Intel's because Intel has HK/MG and AMD will not (at least initially, and possibly for the entire node life).

Where this all gets kinda skewed in a one-dimensional analysis is the lack of budget context. Intel's process development budget literally dwarfs AMD's development budget. The viewpoint should not be one of "I can't believe those slackers continue to lag a year behind Intel, node after node" but rather the viewpoint really ought to be "it is really impressive that with 1/4 the resources AMD somehow manages to stay within drafting distance of the sprinting pace that Intel's budget allows them to maintain".

And on the flip side, for Intel shareholders they have really got to wonder if R&D dollars are being put to work in as efficient a manner as AMD's management does. Tossing 400% more money into the R&D pit and only managing to maintain a 1-2yr lead over a company 1/10 your market cap and with red ink all over the place isn't exactly a testament to a well managed R&D policy, it just means you have more than enough cash to be liberally free with it and not worry too much about the accountability.

If Intel doesn't increase their lead over AMD by at least another year when the 22nm is released (say Intel releases in 2011 and AMD releases in 2013-2014) then I'd be a little irritated as a shareholder that Intel spent all that much more money than AMD and merely maintained their 1-2yr lead.

if it was just intel vs amd then this would be a truly herculean effort by amd. however, amd didn't even create this chip, ibm did. ibm is fighting intel and keeping them in check through amd (along with freescale, toshiba, etc). if amd falters, then via, nvidia, or the sultan of brunei will come up with a product that competes directly with intel, and ibm will support the new company too.

 

[Idontcare]

Originally posted by: bryanW1995
if it was just intel vs amd then this would be a truly herculean effort by amd. however, amd didn't even create this chip, ibm did. ibm is fighting intel and keeping them in check through amd (along with freescale, toshiba, etc). if amd falters, then via, nvidia, or the sultan of brunei will come up with a product that competes directly with intel, and ibm will support the new company too.

Confusing me there, what chip are you referring to?