Yes! Intel and AMD's Cross-License Agreement is very broad. Also IBM and Intel have Process deals that go even deeper. So AMD has other agreements by Proxy LOL! They (AMD) don't get blown away but Intel will be faster=P C2D is faster right now and IMHO, AMD isn't blown away. Slower, yes!
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Intels Penryn @ 45nm
- Thread starter FrgMstr
- Start date
Yes! Intel and AMD's Cross-License Agreement is very broad. Also IBM and Intel have Process deals that go even deeper. So AMD has other agreements by Proxy LOL! They (AMD) don't get blown away but Intel will be faster=P C2D is faster right now and IMHO, AMD isn't blown away. Slower, yes!
All the major companies (Intel, AMD, IBM, Samsung etc) are working on high-k gate oxides, I'm sure IBM has it already, maybe AMD as well. It's just a matter of when it goes into production, and that's where Intel is king, getting technology into production.
These announcements are mostly marketing anyway - a way to show their customers how high tech they are. Since they didn't announce the k value, or what metal they're using for the gate, you have to take their performance claims on faith.
Not really, some showed working DC and QC parts , others presented research papers...I wouldn't put an = between them.
Jimmy the Greek
Gawd
- Joined
- Aug 20, 2002
- Messages
- 657
Are we talking about a tangible difference from C2D, in the neighborhood of 3 GHz for the mid-level processors? Or is it not worth waiting for if I plan to build a new system?
Not really, some showed working DC and QC parts , others presented research papers...I wouldn't put an = between them.
Research papers on high-k gate insulators & metal gate oxides have been out for a few years now.
Since we don't know how IBM's high-k technology compares to Intel's, Intel's being 1st to production may not be that impressive. There's a huge heap of marketing:
Intel: "Biggest breakthrough since the 60s"
IBM: "Our high-k is going to make Intel's look like breakfast cereal"
(i'm paraphrasing but it was something silly like that).
All the sites have reviews and a lot of the info is there. I like the way this one was laid. There guys are a very long way from being Intel Lapdogs BTW.
http://pcper.com/article.php?aid=353
I read somewhere that A0 stepping of Core 2 Duo was only 1.6GHz. It seems that last time Intel tried making changes to something that already worked they failed. That was NorthwoodC to Prescott.
It seems this time it worked. IMHO, nothing else but that, that puts Penny at 3.3 at first and maybe up to 3.73GHz Tops. But with a free 20% boost, hey!I agree with them about that last Demo or the so called V8. So much for folks complaining about FSB. Dual Independent Buses and two separate banks of RAM, no chokage there
I am wondering what metals they are using for the process. The insulator is probably a stack of hafnium oxide and zirconium oxide. It's shown to have a ton of promise. Since I'm not a process engineer at Intel, I've gotta wonder from the outside
Though a quick patent search reveals much.What I find interesting is that they've patented a process that includes 7 or 8 different metals, but obviously they're not all being used. I wonder if it's too broad.
I am wondering what metals they are using for the process. The insulator is probably a stack of hafnium oxide and zirconium oxide. It's shown to have a ton of promise. Since I'm not a process engineer at Intel, I've gotta wonder from the outside
What I find interesting is that they've patented a process that includes 7 or 8 different metals, but obviously they're not all being used. I wonder if it's too broad.
one of the articles said the gate oxide layer was Halfnium oxide. They will keep the gate metal material secret probably.
An ALD process for the HfOx would probably include all the deposition gasses, that could be what you saw?
Wolfdale.
The article mentioned a hafnium based layer, but didn't say specifically oxide.
Nah it was a ton of metals that were mentioned in the patent, not gasses.
An ALD for HfOx would probably wouldn't contain W, Zr, Al, etc.
Here's a link to Intel's patent: http://patft.uspto.gov/netacgi/nph-...50&s1=7148548.PN.&OS=PN/7148548&RS=PN/7148548
Yes, metals must be for the gate material, to replace poly silicon. The gate material is going to be secret, so maybe the list of metals is just to obscure what they're actually using.
The gate insulator has to be an oxide, or it won't be an insulator. Hafnium oxide silicate, is what the industry is planning to use - so I'm sure that's what it is.
And I'm sure IBM, AMAT, TEL, AMD - all have patents on high k gate oxides.
I agree on the obfuscation.
I looked in the database and there are no AMD/IBM patents with the word hafnium in them. Does the patent agreement with Intel run that deep?
Companies file patents like crazy, they don't mean that much. I've done one myself. One thing's for sure - Intel didn't invent HfO2 gate oxide.
Patents in the US are first to invent, not first to file - so AMD or IBM could have a patent for HfO2 gate oxide they haven't even filed yet. It's who invents it (and documents it) first, not who files first. It might even be the ALD manufacturer who has the patent (AMAT, TEL etc).
Patents don't mean as much as you might think in IC manufacturing - everyone ends up using patented technologies, and you're lucky if you can get royalties from them. Even if Intel had a patent, IBM could change the doping in the HfO2 film or something and claim it's a different method.
wouldnt it be in powers of 2... so itd be 2 cores 4 cores 8 cores then 16 or something like that...
only makes sense easier numbers if you trying to split stuff instead of 33.3333% going to each core... lol u can have 25% go to each of the 4 cores and so on...
i dont know much but thats waht i figure...
the aliens that originally gave intel its 8086 came back and were saddened that they actually took that up to the prescott cores until changing things around.
hence the outrageous performance from C2D...
aliens....
DUH!!
mlambert890
Limp Gawd
- Joined
- Jan 8, 2007
- Messages
- 175
Just want to pull this out to clear up this misinformation. As indicated in a post a few above this, Intel/AMDs cross-license deal is BY NO MEANS "broad" and certainly does not cover process. Deals like this do not work through proxy! Hugely complex patent settlements are not transitive. I dont think Intel has any significant deals with IBM at all. They are fierce competitors. IBMs 45nm process is a joint effort with AMD, Sony and Toshiba. Intel worked on its own. It hardly matters anyway. IBM doesnt compete directly with Intel in its strongest market (esp now with the loss of Apple to Intel). IBM basically makes procs for consoles, UNIX boxes, set top boxes and other vertical industry applications. Intel competes with them in those spaces, but has totally lost the console battle and really relies on desktop/server/notebook sales for their revenue.
Im not going to pretend to know how the 45nm parts will shake out, but I think that right now, AMD *is* blown away. Intel has significantly faster parts that are cheaper and use less power. Not good. We'll see what AMDs 45nm parts bring to the table, but even if AMD/IBM/Sony/Toshiba have created a brilliant process, it wont necessarily translate to significant real world gains. The architecture is extremely important, obviously, as is the ability to execute on manufacturing. Intel is tough to beat when they are geared up. AMD struck when they had made a very bad architecture decision with the P4. They have corrected that now.
Personally, I see AMD going a totally different route. Their ATI acquisition, and now some announcements they've made, hint at what I had suspected. They are looking to lock up the potentially lucrative (in the future) media device market. They've made some announcements about working on a configurable at time of manufacture part that would be X number of CPU cores and Y number of GPU cores. So an OEM could potentially order a 2/2 part and another could order a 3/1 part and they could accomodate. If they could keep that part cost effective, they could potentially lock down the future of the set top box. Shame that they missed this last round of consoles. Would have been a very tempting part for MS and Sony.
Clear up misinformation by posting more, WOW! Too bad you missed link to the agreements posted here last year hehehehe!
I agree that Intel & IBM don't share any process knowledge directly - in fact Intel is the most secretive of the chipmakers in my experience.
But where the technology exchange occurs is through the vendors, who do most of the process development work these days. If IBM or Intel needs to know how to etch or deposit 45 nm lines, they have to ask their etch and CVD vendors (AMAT, TEL, Novellus etc) to do it, they can't figure it out themselves.
Since these same vendors also supply AMD, IBM etc, much of that knowledge filters over, either legally or illegally. That's why it's tough to maintain a significant lead in semiconductor technology.
Intel is first to 45 nm, yes, but AMD are very clever, so don't count them out. People were doing that for years when their stock was $5/share, and they missed out on a lot of money.
Sony is so fond of the 45nm process it developed with IBM/AMD/Toshiba that it plans to get out of manufacturing like TI and use a fab-lite model.
Keeping the pace with Intel and Samsung is increasingly difficult and I wouldn't be surprised that others bail out from the race to post 45nm process tech.
I thought we were in agreeance for the fact that Penryn was the successor to Merom, therefore a mobile dualcore cpu?
And yes, I used agreeance in a sentence. +10 points for me
I know this is old, but Huh? Are you being sarcastic or are you serious? I'm pretty sure you're joking. But anyway I was talking to Orayn who said
"Just a reminder, we won't be talking about Penryn the same way we talked about Conroe; Penryn is the next generation's Merom. I'm not sure which desktop chip is the second-gen C2D... Also, I'd bet that after Penryn, we'll move on to Core 3."
http://www.reghardware.co.uk/2007/01/17/intel_tapes_out_penryn/
coldpower27
[H]ard|Gawd
- Joined
- Jan 27, 2005
- Messages
- 1,154
The Penryn family just refers to processor based on Core micro-architecture on the 45nm process. Penryn itself is the codename for the mobile variant.
There are multiple codenames depending on where your talking about.
Harpertown Quad Core Server DP
Yorkfield Quad Core Desktop
Wolfdale Dual Core Desktop
Penryn Dual Core Mobile
There are multiple codenames depending on where your talking about.
Harpertown Quad Core Server DP
Yorkfield Quad Core Desktop
Wolfdale Dual Core Desktop
Penryn Dual Core Mobile