backtop


Print 67 comment(s) - last by Amiga500.. on Apr 25 at 9:31 AM

Giuseppe Amato gives another overview on the high points of AMD's next-generation CPU architecture

This article was first published in German on K-Hardware.de.

Yesterday, AMD held a press presentation in Munich, Germany to update journalists about its upcoming K10 processor. AMD's Giuseppe Amato, Technical Director Sales and Marketing EMEA, had a few minutes to talk about the architecture at length. This architecture, previously dubbed K8L by Henri Richard -- now publically called K10 -- is scheduled to be AMD's first monolithic quad-core design.

The integrated memory controller (IMC) will get a few new features in the K10 core. When utilizing multiple memory modules, along with proper BIOS implementation and mainboard routing, the IMC can access memory in 64-bit channels (72-bit if you use ECC). This way it is possible to read and write data simultaneously, or improve efficiency for irregular access patterns which increasingly occur in a quad-core environment. This feature is available on AM2+ and F+ boards; on "old“ socket AM2 and F boards the usual 128-bit dual-channel mode is available.

Due to split power planes, the IMC can be clocked down independently of the CPU cores, along with reduced voltage. This also enables CPU overclocking without touching the memory frequency, something that may appeal to enthusiasts. These features are again dependent on Socket AM2+ and F+ platforms.

Amato explained how the quad-core design benefits from the internal crossbar switch the backbone of communication inside the K10 CPU. With Intel's current quad-core design there are cases where data needs to travel over the FSB -- in AMDs case all inter-CPU communication takes place on-die.

The crossbar switch of the K10 core is already prepared for up to 8 cores, Amato boasted. Amato wouldn't give even a vague timeframe for market availability of such a CPU, though he indicated the company is prepared for whatever the market demands. Amato made clear that octo-core is far away in the future – Shanghai will not get 8 cores.

K10 will introduce a shared L3 cache while the individual cores have dedicated L1 and L2 caches. As long as requested data lies in L1, it can be directly loaded. This also works if the data lies in the L1 cache of another core, in which case the communication works via the crossbar switch. In case requested data resides in the L2 cache, it will be loaded to L1 and then invalidated in L2 as AMD has an exclusive cache design. The L3 Cache, however, is not exclusive, but allows for a shared bit to be set. If a core loads data marked as shared, it will reside in the L3 cache and can be fetched by other cores as well.

Amato also mentioned an array of power saving measures which, in sum, allow AMD to deliver a quad-core CPU in the same thermal envelope as today’s dual-core CPUs.

K10 adds the capability of independently clocking all the CPU cores. In current K8 processors (and Intel's Core 2 generation), all cores are clocked at the same level all the time -- the P-state can only be changed synchronously. In case of a compute-intensive single-threaded process, all cores must run on the highest level P-state. On K10-based CPUs, the idle cores could be switched to the lowest P-state, while others are in different states, depending on load.

This feature could possibly be abused by overclockers to overclock a single core above the specified levels. Amato clarified that AMD doesn't endorse overclocking, but acknowledges there are people interested in that. In a warranty case, AMD could detect PLL programmings out of spec which would deny the warranty. The new cores, however, have new thermal sensors, to improve overheating protection.

Amato closed the session by mentioning Shanghai as a successor to Barcelona in the server space for 2008. Shanghai will be an improved quad-core architecture, which is supposed to be socket-compatible with current Socket F platforms. Roadmaps available to DailyTech revealed Shanghai is a 45nm quad-core CPU featuring 6MB of L3 Cache.



Comments     Threshold


This article is over a month old, voting and posting comments is disabled

RE: A little pleased
By Targon on 4/23/2007 7:09:40 AM , Rating: 2
There is something to be said in IT to have a limited number of different machine types in service at any one time. While I disagree with the idea that all machines in an environment should be identical, when it comes to fixing problems, it really does help in many situations.

With that said, having several different types of machines in use really isn't a bad thing in IT, as long as you don't go with the, "All our machines are identical except these two machines" type situation. If you order 25 machines a month, then going with some Intel and some AMD based machines is fine.

Over the long haul, AMD does tend to let IT upgrade the processors for a longer period of time with a given socket. Socket 754 was never planned to last for a long time, socket 939 wasn't too bad for how long you could upgrade processors. Socket AM2 processors will work in systems with an AM2+ socket as an upgrade path, and socket AM2+ processors will work in socket AM2 motherboards, just without being able to have each core clock independently. When socket AM2+ processors are released, I have heard the plan is to let them work on socket AM3, though I could be wrong about that.

Now, back to the topic at hand...

Vendors upgrade their models on a very regular basis, dropping the old, and adding the new. Sure, you can still buy older model parts for quite a while if you hunt around the Internet, but when it comes down to it, if you buy 25 machines a month, every six months to a year will probably result in the motherboard or some other component being different. Along the way you have seen hard drives phase over from PATA to SATA, and even the CD/DVD drives have started the transition across the board. Power supplies have changed to reflect the transition of the market as well. Dell tends to push whatever chipset Intel wants them to in Intel based machines as another example. Have you looked at the drivers page for a Dell machine with the same model number? For the same model, you can have different chipsets and different features, so even if you order the same model machine, there WILL be differences over time.

So, the key for larger IT departments is to go with batches of machines that are roughly the same. If you go 50/50 Intel/AMD, then having both types of platforms in house will still be easy to manage. Just avoid VIA or SiS chipsets, and you should be fine.


RE: A little pleased
By PrezWeezy on 4/23/2007 1:25:50 PM , Rating: 2
I agree. I wasn't saying that my suggestion was a purely prcatical one, I was simply stating the "ideal" way to run. In reality you can't have the same maching for everyone all the time. We happen to build our machines so we can stay with one model for a longer period of time. The more uniform you can make your topology, the better for managing. I don't want to have to deal with 120 different images. I have 10. I don't use the same machine for everyone but I like to be similar if possible. And if it comes to needing to upgrade the processor, we just replace the machine and move the old one down to someone who has an even older machine.

Bob661:
As for everyone needing the same machine you are correct that not everyone needs the same machine to run their particular apps. But I can build a computer with an MSI board that has on board video, 1 gig of RAM, everyone uses DVD burners now, an 80 gig hard drive, and a C2D E6300. If someone needs more processing power I can simply put in an E6600. For someone who needs more video we can put in an add on video card. So not everyone has the same machine but everyone has a platform on which to build. Just like Ford doesn't use a different platform for the Explorer, Mountainier, and Navigator. It's the same car, just dfferent classes, therefor they use the same frame and chasis and just add options for what the customer wants, and wants to pay for.


"This is about the Internet.  Everything on the Internet is encrypted. This is not a BlackBerry-only issue. If they can't deal with the Internet, they should shut it off." -- RIM co-CEO Michael Lazaridis

Related Articles
AMD 2007 Server Roadmap
December 14, 2006, 9:47 AM













botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki