backtop

Print E-mail del.icio.us 53 comment(s) - last by AlmostExAMD.. on May 24 at 8:17 AM
Details of the K8L architecture and AMD quad-core roadmap inside

AMD's Spring Forum just finished up and industry insiders got a chance to sit down with DailyTech and discuss the major cornerstone of the forum, K8L.  K8L is AMD's next generation processor technology, which Henri Richard first revealed to Chris Hall at Digitimes back in March.  Richard himself described K8L as evolutionary, rather than revolutionary, and it appears as though that comment was spot on the mark from the details of K8L we've seen. 

Chuck Moore, a senior AMD Fellow, gave a presentation at AMD's Spring Forum revealing some of K8L's features. Aside from being a quad-core-friendly architecture, there are three key features that will separate K8L from K8: cache, memory and HyperTransport. 

Moore revealed that K8L will be the first AMD processor to have L3 cache since the K6 CPU.  Each core has an independent L2 cache, but the entire processor shares an L3 cache pool.  There's no word yet on exactly how much cache the K8L can hold, though the K8L will be a 65nm SOI process so AMD engineers have a bit more die real estate to play around with. 

K8L will support DDR2 and DDR3 when it becomes available, although it’s still anyone's guess as to whether or not the market will actually adopt DDR3 quickly enough to warrant using DDR3 aggressively at the core launch.  Tom Trill, Samsung's Director of DRAM Marketing, was extremely hesitant to claim DDR3 would make it to the desktop without "significant" performance gains over DDR2 -- lest anyone repeat many of the mistakes made when the industry migrated from DDR1 to DDR2.

HyperTransport 3 will be a key element of K8L.  HyperTransport 3, which was just ratified a few weeks ago, increases the frequency of the current HyperTransport bus from 1.4GHz to 2.6GHz, or from 2.8GT/s to 5.2GT/s.  Current AMD Opteron processors only support HT links operating at 1GHz, though the HT 2.0 specification allows these links to run as fast as 1.4GHz.  Non-K8 quad core processors will almost certainly take advantage of this additional headroom as data across these links gets more crowded.  However, K8L processors will have the advantage of using the full 5.2GT/s per link defined in HyperTransport 3. 

K8L processors are expected to be very co-processor friendly, allowing for additional HT and HTX interconnects specifically for math or cryptography acceleration. Current Opteron 2xx and 8xx processors use three HyperTransport links per die, but AMD's documentation did not reveal how many HyperTransport links K8L would utilize.  Recently, AMD's Phil Hester claimed embedded on-chip coprocessors were part of the company's long term plan just a few months ago.  While we may not see embedded co-processors with the K8L, it does look like the architecture is gearing towards supporting co-processors in a big way.

Behind closed doors, insiders revealed to DailyTech a few tidbits of the long term quad core roadmap.  AMD will introduce no less than four quad-core families over the next two years, with the first being Deerhound.  Deerhound, we are told, will be a Socket F server processor expected to ship late next year on the K8 -- not K8L -- architecture.   Deerhound did not appear to support FB-DIMM. 

In early 2008, AMD's corporate roadmap claims a quad core desktop CPU will make an appearance, dubbed Greyhound.  Greyhound is slated to become the first quad-core AMD chip to use the HyperTransport 3 bus, and the memory controller is slated to support DDR2 and DDR3.  Unlike Deerhound, Greyhound will use the K8L architecture, and all the goodies that come with it, including the 5.2GT/s HyperTransport support.  Unless AMD's plans change drastically between now and 2008, the processors will require a new socket.

There are very few details of the other two processor families, but hopefully future forums will get a chance to touch on some of those developments as they firm up.  However, we do know that server processor cores are expected to migrate to K8L after Deerhound. 

K8L has a few other highlights, including dynamic powering of sections of the processor -- a page taken straight out of Intel's Yonah playbook.  The p-states, as far as we can tell, will be separate for each core and the memory controller.  The K8L core will also have a 32-byte prefetch (versus 16 right now), 48-bit addressing with 1GB pages and the ability to process 128-bit SSE instructions in a single cycle, another very Intel-esque feature. 

Expect to see more tidbits over the next few days as roadmaps and transcripts are still being released. 

Comments     Threshold


This article is over a month old, voting and posting comments is disabled
Timetable
By nrb on 5/18/06, Rating: 0
RE: Timetable
By Spoonbender on 5/18/2006 1:03:57 PM , Rating: 3
Why is that a problem? Intel survived for three *years* without being able to close the performance gap. A company doesn't have to be permanent #1 to survive. AMD still has a great product, and Intel is still unable to compete in higher end machines (4-way cpu's where the FSB just can't deliver enough bandwidth, for example).

In the desktop market, yeah, Intel will have an advantage for a couple of months. But hey, if AMD *really* need to compete, they can easily crank up the voltage a bit, and deliver some faster chips, at a higher TDP. That way, they might be able to at least come close to closing the performance gap. They've also got the 65nk transition coming up soon, which will help them squeeze a few more percent out as well.

And who cares if Intel makes a quad-core product first? It won't perform very well unless they manage to crank up the bus bandwidth *a lot*.

And probably most importantly, AMD has finally managed to convince a lot of people that AMD is a great brand. That means a lot of people will buy AMD products, even when they're not #1.

That's what Intel has survived on for the last 3 years. The same effect should be able to keep AMD going for half a year, don't you think? ;)


RE: Timetable
By MrKaz on 5/18/2006 1:20:37 PM , Rating: 2
Perfect reply!


RE: Timetable
By RyanVM on 5/18/2006 3:04:18 PM , Rating: 2
Intel's also got a lot more spare cash floating around than AMD. They afford to be behind for awhile. AMD can't.


RE: Timetable
By Devil Bunny on 5/18/2006 10:43:10 PM , Rating: 2
This happens all the time in other Industry's. Look at the oil industry for example, you have atleast 4 major oil companys competing right now, and only one can be number one. Also look at the graphics card area, the Nvidia 5 series sucked horribly, so people bougth ATI, and Nvidia is still around today. Just becuase a company isnt number one doesn't mean that they will fail, and the CPU market is just the same.


RE: Timetable
By Griswold on 5/18/2006 1:15:58 PM , Rating: 2
No, I dont think it looks bad for AMD. Intel thinks that core 2 will make up for 20% of their sales by the end of the year. Thats not too impressive. It just tells us that they know, that the whole world wont jump all over core 2 at launch. They may have a superior product available, but that doesnt mean AMD will go bancrupt the next day. You need two things to be successful: a good product and people who buy it.

As for quad cores on desktop.

How many of these desktop quad core (they will be the price premium edition) do you think will be sold? How long did it take for dual core to become a somewhat mainstream product?

Judging this book by how soon quad cores for desktop appear is bad at best.

Besides that, AMDs roadmap policy is just crap. Snippets here, some pieces there, all that spiced up with speculative stuff sites like hkepc pull out of their arse. Thats really not a good foundation to base any speculations on. Just another field AMD has to improve alot on, it goes hand in hand with their crappy marketing.




RE: Timetable
By defter on 5/18/2006 2:28:36 PM , Rating: 1
quote:
No, I dont think it looks bad for AMD. Intel thinks that core 2 will make up for 20% of their sales by the end of the year. Thats not too impressive.


That's VERY impressive. Do you have any idea what was percentage of K7 unit sales two quarters after launch? Or what about K8? I can guarantee that those numbers were significantly below 20%.

One thing people forget when talking about marketshares is that laptops account for almost 50% of the market. In the laptop market Intel has already excellent product and there isn't really much hurry to replace majority of Yonah's with Merom.

When you remove Celerons and mobile parts from Intel sales, 20% of the total sales suddenly will look a quite big number...


RE: Timetable
By Griswold on 5/18/2006 5:40:29 PM , Rating: 2
Actually, if I remember correctly, this 20% figure was just for desktop parts. That leaves 80% for P4 sales. Intel still has a huge P4 stockpile.

And just for the record, I doubt that laptops make up for 50% of the market. There is alot of growth there, but its hardly near the absolute sales of desktop machines yet.


RE: Timetable
By JumpingJack on 5/21/2006 5:23:37 PM , Rating: 2
Actually that is not true, laptop sales are neareing 50% -- it has been mobile that has kept Intel going and more or less hiding the impact of share loss:

http://www.computerworld.com/hardwaretopics/hardwa...


RE: Timetable
By JumpingJack on 5/21/2006 5:26:34 PM , Rating: 2
Here is another, laptop sales surpass 50% in retail in 2005.... I wonder is this is a factor contributing to the fact that AMD ate some into the DT retail as Intel pretty much controls mobile.
http://www.firingsquad.com/news/newsarticle.asp?se...


RE: Timetable
By photoguy99 on 5/18/2006 4:31:03 PM , Rating: 2
quote:
Judging this book by how soon quad cores for desktop appear is bad at best.


Maybe, but quad cores a year early on the server would be a devastating blow to AMD.

The advantage scales so well and almost all server applications benefit from it.

Having dual core servers first was one reason Intel lost share in that segment.



RE: Timetable
By Griswold on 5/18/2006 6:03:20 PM , Rating: 2
Deerhound will appear 2007 as mentioned above. Wont be based on K8L though. Intels Cloverton (reminds me of the early P-D rush job with 2 cores glued together back in 2005) will be around sometime in 1H 2007, probably rather sooner than later, but definitely not 1 year ahead of AMD.

Two more things:

- Enterprise customers arent early adopters. Both Woodcrest and Cloverton wont go "from 0 to hero". These people want validated hardware, not the newest thing on the market. A K8 quad on the other hand is a known factor.

- It remains to be seen how well cloverton really scales, especially in multi-socket systems.


RE: Timetable
By zsdersw on 5/19/2006 7:17:46 AM , Rating: 2
If, as you say, they don't want "the newest thing on the market", which I suspect is probably true, it serves to hinder early adoption of new platforms and products from both AMD and Intel.. not just Intel.


RE: Timetable
By bob661 on 5/19/2006 11:04:59 AM , Rating: 2
quote:
If, as you say, they don't want "the newest thing on the market", which I suspect is probably true, it serves to hinder early adoption of new platforms and products from both AMD and Intel.. not just Intel.
He's correct. The enterprise sector doesn't jump on the latest hardware unless there's an immediate need for the new tech. And even if they do jump on the new stuff, there will be quite a bit of testing before it's entered into service.


RE: Timetable
By bob661 on 5/19/2006 11:07:33 AM , Rating: 2
This applies to both Intel and AMD. Also, it doesn't seem to hinder either company. If there have been a lot of companies asking for Opterons, then more than likely there will be immediate purchases of the Dell Opteron servers.


RE: Timetable
By lethalchronic on 5/19/2006 1:34:24 PM , Rating: 2
Quote- "How many of these desktop quad core (they will be the price premium edition) do you think will be sold? How long did it take for dual core to become a somewhat mainstream product?"

I think mainstream here is a complete understatement here. I think it is easy to confuse our market (Daily Tech readers) with what is truly mainstream. Getting to that mainstream point in the hardcore market or tech savvy market is one thing.

My point is dual core has a long way before the title mainstream can be applied. We will come to that point when OEMs start pushing it seriously. The mainstream market has always been (unfortunately) tied with the OEMs.


AMD the thing some of you missed
By proamerica on 5/19/2006 12:12:57 PM , Rating: 2
One thing you guys are missing I think is that AMD is going to try and do some serious Ghz scaling on AM2. I've seen some rampant speculation about K8 on 65nm being able to hit 4Ghz. Right now the fastest available dual core is the 2.6Ghz FX60, if they can get that up to 3.6Ghz on 65nm then maybe they will have an answer to Conroe before K8L.

Of course, who the heck knows whats true about the future, but with AMD claiming they are bringing back the "Mhz war" and that they intend to win it, the idea of K8 hitting 3.5Ghz and above is not beyond the realm of possibility.




By saratoga on 5/19/2006 4:03:24 PM , Rating: 2
quote:
One thing you guys are missing I think is that AMD is going to try and do some serious Ghz scaling on AM2. I've seen some rampant speculation about K8 on 65nm being able to hit 4Ghz.


Thats retarded. Its a shallow pipe, and they're running into it's limits. Just look at the 90nm shrink where they went from 2.6GHz to about 3GHz on 90nm. Realistically, the K8 is unlikely to scale much higher then it is now, regardless of die size.


By Spoonbender on 5/19/2006 5:49:33 PM , Rating: 2
"Rampant speculation"? Yes, I can't believe we didn't take that into account... ;)

Forget about seeing 4GHz A64's. It will scale beyond the current 3GHz, and apparently their 65nm process will even scale relatively well. But *not* 33% above the original max. And not above what the P4, an architecture *designed* for MHz scaling, could do.
3.5GHz might not be impossible though.