New rumors about next-gen Xbox CPU specs hit the web

While Sony has now revealed some of the first details surrounding its Playstation 4 game console, Microsoft is still not saying anything officially about its plans for the successor to the Xbox 360. That hasn't stopped a plethora of rumors about the console from leaking, however, and the most recent claims involve the next Xbox's processor.

The report comes from VGleaks.com, which claims AMD is creating the processor for Microsoft. The same site recently posted up info on what the console will have in terms of Kinect support.

The CPU inside the console, code named Durango, has two modules, each with four x86 cores inside, according to the article. Each of those cores reportedly runs on a single thread at 1.6Ghz. The article adds that the cores each feature "a 32 KB instruction cache (I-cache) and a 32 KB data cache (D-cache), and the 4 cores in each module share a 2 MB level 2 (L2) cache. In total, the modules have 8 hardware threads and 4 MB of L2."

One thing that did stand out was that while each core in the Durango CPU reportedly has a clock speed of 1.6 Ghz, that's actually half the speed of the cores inside the current Xbox 360. The Xbox 360, which launched in 2005, has a PowerPC-based processor with three cores and a clock speed of 3.2 Ghz.

However, even with that lower clock speed, the Durango CPU should perform much faster than the chip inside the Xbox 360. That's thanks to a number of other hardware changes inside the chip, including the Durango CPU cores getting dual x64 instruction decoders in order to decode two instructions per cycle.

Source: VGleaks.com | Image via VGleaks.com

Report a problem with article
Previous Story

Microsoft offers "enhanced" IE10 with Bing and MSN [Update]

Next Story

Samsung holding public Galaxy S IV event in NYC's Times Square March 14th

90 Comments

Commenting is disabled on this article.

Could the systems be, gasp, identical in hardware?

At least AMD found something to keep them afloat. I'm happy for them.

Now to actually make a CPU that fights off Intel's offerings on equal footing so prices will finally go down.

You CANNOT compared GHz between different generations of CPUs. That is not a good way to judge whether one CPU is faster than another. What you need to know is its FLOPS. GHz means nothing in terms of performance.

You kids these days, AMD64 were 64-bit extensions, EXTENSIONS added to native x86 CPUs. Surely that has progressed over the years but they are still based on the x86 architecture but can execute AMD64 instructions due to those EXTENSIONS.
Pure 64-bit processors cannot execute 32-bit code. The architecture wars were like this:
Intel: We don't think x64 is necessary for consumers at the moment. Lets focus in large business IA-64
AMD: Here are the AMD64 extensions to the existing x86 architecture. Intel you can have them if you like.
Microsoft: Lets make Windows 64-bit based on AMD64. Lets also have an IA-64 version for the freaks

I love how people still think clock speed has everything to do with processing power... already have people saying why are the new console processors slower then the old ones.... *hits head on desk*

Not to mention all connecting buses optimized for gaming. Now MS building in Kinect and wanting to be more of a multimedia hub may give it a slight hit on pure gaming performance. Who knows. I'm sure they will both perform similarly, just like this generation does, more or less.

And that's the game Samsung is playing with the mobile phones; more RAM more Cores, higher clock speed. Everyone thinks more is always better in every scenario.

Looks like the processor might be faster than the PS4's. This has 4MB of L2 Cache where the PS4 probably only has 1/2MB L2 Cache. I'm not an expert but the Xbox having more L2 Cache might be compensation for slower RAM.

I would put faith in the engineers here. It's easy to throw around all kinds of numbers about cache but it's very non-intuitive as to how various designs perform in different situations.

Essentially the same. For games that appear on both consoles, I doubt any developer is going to spend a lot of time optimizing for what are two nearly similar systems. For the exclusives, well, there's no comparison anyway.

Tony. said,
Can someone compare this to the PS4 CPU? Pros and Cons?

This console's CPU isn't even confirmed so it cannot be compared.

according to already stated rumors: same CPU at the core, one has Sony customizations, the other one has MS own changes on it. Performance should be virtually the same, unless one of these companies made a breakthrough discovery while optimizing those chips.

"including the Durango CPU cores getting dual x64 instruction decoders in order to decode two instructions per cycle"

Does anyone know how this feature compares with hyper-threading? Could HT complement this?

pmdci said,
"including the Durango CPU cores getting dual x64 instruction decoders in order to decode two instructions per cycle"

Does anyone know how this feature compares with hyper-threading? Could HT complement this?

With the sparse details given it sounds like AMD's implementation of HT.

HT is not only about dual-decoders. HT duplicates the whole pipeline and registers. The only thing that HT does not duplicate is the actual core...

MFH said,
HT is not only about dual-decoders. HT duplicates the whole pipeline and registers. The only thing that HT does not duplicate is the actual core...

This sir knows his stuff. HT is different than what this CPU is doing (and is better).

pmdci said,
"including the Durango CPU cores getting dual x64 instruction decoders in order to decode two instructions per cycle"

Does anyone know how this feature compares with hyper-threading? Could HT complement this?

HT is parallelism, it is a complete different topic.

Let's say that the next command is a 64bits operation and every commands takes 1 second to run.
"sudo make me a sandwich" (1 second total)

in 32bits, the same should expressed as 2 operations:
"sudo make me"
"a sandwich". (2 seconds).

However, this CPU is doing to do the opposite, to use a 64bits cpu to do 2 32 bits operation at the same time (2 times faster). The trick, is to use short commands.

For example:
"sudo sandwich and a cola" (1 second).

It can realize two short commands at the same time.

Anaron said,
The PS4 may have better specs but the next-gen Xbox will have a D3D11.1-compatible GPU. That might bridge the gap a bit.
The PS4's GPU has this as well so...

Aaron Olive said,
The PS4's GPU has this as well so...
It would be rather surprising to find DirectX itself being used on a non-Microsoft product?

Aaron Olive said,
The PS4's GPU has this as well so...

Capable, most certainly, usable? Indefinitely not. Have fun in the modified OpenGL world!

Hopefully we will get "buy once, run anywhere" with the next xBox. I just bought the Star Wars pinball bundle for Windows 8 and I wish it would have included running on the xBox and Windows Phone. I'd pay a 30-50% premium for this.

rojorojo said,
It's gonna run the windows kernel so the app platform can be unified with Windows 8/RT/Phone

xbox has always ran the windows kernel

McKay said,
The 360 runs on the Windows CE Kernel, this next one might run on NT.

no they don't, they run on a modified Windows NT kernel, Windows CE was based on the 9x kernel line.. the first xbox was based on a modified windows 2000 (NT) kernel, 360 was an offshoot of the same kernel with new extensions on it

FoxieFoxie said,
I only care abotu RAM and GPU, looks like PS4 will; have my money tyis t9me

So you don't care about the games or its features? Curious what will you be doing with the PS4 ?

FoxieFoxie said,
I only care abotu RAM and GPU, looks like PS4 will; have my money tyis t9me

I hear that. good for you. well, I only care about what the console can do it doesn't matter if something else is out there that's a dat bit more powerful.
so far the rumors say the the next xbox has wifi direct....I think this is cool so my big ass TV can be on my wall with only the power cord visible (we'll see if it's true).
I'm not much of a fan about exclusives....if the game is good I'll play it.
will I still need my cable box or I can just get my cable service through the xbox?
for me it's not just about games anymore. I'm spending too much money to only have a single purpose machine sitting in my house.

FoxieFoxie said,
I only care abotu RAM and GPU, looks like PS4 will; have my money tyis t9me

I used to be like you until I wound up with a tubo graphix 16, an Atari Jaguar and a couple 3DO systems all collecting dust.

Now I realize there is more than horespower to make a console successful.

FoxieFoxie said,
I only care abotu RAM and GPU, looks like PS4 will; have my money tyis t9me

Specs aren't everything. If they're that important to you then just drop consoles and go PC only. As long as MS doesn't do anything drastic or completely drop the ball then they should be ok. New system looks promising but we'll know in April eh?

We'll know more in April it seems, though I'd like to point out that even if AMD is making it MS runs it's own hardware engineering group that makes cpu and gpu designs which they then own most of the IP on. That's the case with the chips in the Xbox 360 so although the next Xbox will run on x86 I bet it's a tweaked design from MS with some AMD bit's mixed in of course. The same for the GPU I expect. Doing it this way they have more control over the hardware through the life of the system and can control costs better.

GP007 said,
We'll know more in April it seems, though I'd like to point out that even if AMD is making it MS runs it's own hardware engineering group that makes cpu and gpu designs which they then own most of the IP on. That's the case with the chips in the Xbox 360 so although the next Xbox will run on x86 I bet it's a tweaked design from MS with some AMD bit's mixed in of course. The same for the GPU I expect. Doing it this way they have more control over the hardware through the life of the system and can control costs better.

Actually MS just buys the intellectual property for the design, they don't do the initial design work. All the improvements afterwards are by MS.

Zkal said,

Actually MS just buys the intellectual property for the design, they don't do the initial design work. All the improvements afterwards are by MS.

yeah,they license the IP from amd,and they get full digital schematics of the chip with the masks. Then they are free to modify it then send it off to a fab house,although amd might require themselves to fab it.

just like arm chip makers.they license arm chip, gpu ,USB controller, display controller, then they create a layout with all these connected,then send it over to TSMC for example to manufacture it.

Similar to the recent 64 bit procs from Intel and AMD it will be an x86-64 processor. So 64 bit extensions to the x86 instruction set

ramesees said,
Similar to the recent 64 bit procs from Intel and AMD it will be an x86-64 processor. So 64 bit extensions to the x86 instruction set

Not particularly bothered about that, of course it's needed for the extra ram etc

but what I really like is the the idea that we'll be getting games that are more geared towards multithreaded use. It'll be another bonus for PC gamers who've been shovelled console ports over the years.

Maybe it'll be worth owning quad core for games ...

Hmm, could someone enlighten me cause I have no idea how this work. It's a x86 Processor but it can decode x64 instructions? How would that work? I am curious about these tech stuff.

This also sounds like it's pretty much the same CPU thats in the PS4 - AMD, x86-64, 8 cores, 2 modules with 4 cores each. It's likely based on AMD's Jaguar architecture which the PS4 also uses.

AMD's Jaguar is a low-end low power x86-64 CPU thats coming to PC this year, but with 4 cores. Even with 8 cores in the consoles it wont compare to a mid-range PC CPU. These consoles are very weak in the CPU department.

I'd expect the PS4 to have a better GPU than the next Xbox simply because MS are going to integrate or include Kinect with their next console, which raises the cost. So MS will have to compensate by having a lower end GPU and RAM or the console will just cost too much, and all leaked specs are showing this, which have so far been totally accurate for the PS4 (minus the RAM which was 4GB originally, but seems Sony have doubled it at the last minute to compete with the next Xbox which will have 8GB).

Edited by 1Pixel, Feb 28 2013, 2:33pm :

Celestial Being said,
Hmm, could someone enlighten me cause I have no idea how this work. It's a x86 Processor but it can decode x64 instructions? How would that work? I am curious about these tech stuff.

Keep in mind that x86-64 isn't true 64bit computing. It is 32bit computing with memory address and register extensions for 64bit. This is the main reason x86-64 is the most common "64 bit" technology variant in use. Since the chip is still a 32bit chip at its core there is no need to emulate 32bit. Allowing the chip to perform just as fast in both modes..

So, a true 64bit chip wouldn't be able to decode X86 without some sort of emulation layer (see Itanium), but this chip can because it is a 32bit chip with the ability to address 64bit in terms of memory and registers.

LogicalApex said,

Keep in mind that x86-64 isn't true 64bit computing. It is 32bit computing with memory address and register extensions for 64bit. This is the main reason x86-64 is the most common "64 bit" technology variant in use. Since the chip is still a 32bit chip at its core there is no need to emulate 32bit. Allowing the chip to perform just as fast in both modes..

So, a true 64bit chip wouldn't be able to decode X86 without some sort of emulation layer (see Itanium), but this chip can because it is a 32bit chip with the ability to address 64bit in terms of memory and registers.


You have no clue what you are talking about, do you?

http://en.wikipedia.org/wiki/X86

It's also worth noting that in this case, the 64-bitness is somewhat irrelevant. The Xbox 360 PowerPC processors were a "true" 64 bit processor. AMD, on the other hand, created the x86-64 architecture and have never (to my knowledge) released a 64 bit only processor.

Basically, if AMD was the best choice, you were getting x86-64. This choice always had more to do with the GPU than the processor.

LogicalApex said,

Keep in mind that x86-64 isn't true 64bit computing. It is 32bit computing with memory address and register extensions for 64bit. This is the main reason x86-64 is the most common "64 bit" technology variant in use. Since the chip is still a 32bit chip at its core there is no need to emulate 32bit. Allowing the chip to perform just as fast in both modes..

So, a true 64bit chip wouldn't be able to decode X86 without some sort of emulation layer (see Itanium), but this chip can because it is a 32bit chip with the ability to address 64bit in terms of memory and registers.

Don't anyone listen to this guy. He has no idea what he's talking about.

vcfan said,

I always see this guy spreading misinformation. Like last week he was informing that wp8 and metro run on c# and .net only.

Err... I'd like a citation to where I said Metro is only C# and .NET...

Either way, I've gotten the message in regards to contributions.

LogicalApex said,

There is a reason it still has X86 in its name.

Yes, the reason is that it's based on X86 - not to mention that X86 started in 16bit (8086). Will you now claim that these CPUs are still 16bit at their core - they can to this day execute 16bit assembly without a problem? (not in Windows 64bit as Microsoft prevents that in Windows...)

Leopard Seal said,

Don't anyone listen to this guy. He has no idea what he's talking about.

technically he is right.......

X64 is just an extension of x86.... address space and registers where widened, new instructions where added along with a "64bit mode"... and technically a x86 32-bit chip is just an extended 16-bit 286 chip... intel always adds on to it's chips by extensions...

the only "true" current 64-bit chip intel made was the Itanium chip, and it didn't support non-64bit code as all the 32bit instruction layers didn't exist, IA-64 was a true 64-bit architecture... but AMD pretty much forced AMD64 onto Intel for the consumer markets

Leopard Seal said,

Don't anyone listen to this guy. He has no idea what he's talking about.

I don't understand what was said incorrectly. x86-64 isn't "true" 64-bit computing in the purist sense.

neufuse said,

the only "true" current 64-bit chip intel made was the Itanium chip,

What about the Intel i860?

AWilliams87 said,

I don't understand what was said incorrectly. x86-64 isn't "true" 64-bit computing in the purist sense.

Yes it is. By every possible definition, x86-64 CPUs are fully 64-bit and executing 64-bit code on them is "just as 64-bit" as anything. They aren't ONLY 64-bit because they also support 32-bit (and 16-bit I think still) operation. That is a feature, not a limitation in any way (and you could easily design a 64-bit only x86-64/AMD64 chip).

Thank you Neufuse. IA64 Itanium anyone?

neufuse said,

technically he is right.......

X64 is just an extension of x86.... address space and registers where widened, new instructions where added along with a "64bit mode"... and technically a x86 32-bit chip is just an extended 16-bit 286 chip... intel always adds on to it's chips by extensions...

the only "true" current 64-bit chip intel made was the Itanium chip, and it didn't support non-64bit code as all the 32bit instruction layers didn't exist, IA-64 was a true 64-bit architecture... but AMD pretty much forced AMD64 onto Intel for the consumer markets

Brandon Live said,

Yes it is. By every possible definition, x86-64 CPUs are fully 64-bit and executing 64-bit code on them is "just as 64-bit" as anything. They aren't ONLY 64-bit because they also support 32-bit (and 16-bit I think still) operation. That is a feature, not a limitation in any way (and you could easily design a 64-bit only x86-64/AMD64 chip).

Then maybe someone should send the Intel Engineers a memo that they don't know what their own CPU architectures are...

IntelĀ® 64 Architecture refers to systems based on IA-32 architecture processors which have 64-bit architectural extensions, for example, IntelĀ® CoreTM2 processor family)...

Source: http://software.intel.com/en-u...-64-ia-64-architecture-mean

You can't have an X86-64 CPU without 32bit (X86) as X86-64 is, as I said earlier, extensions to X86 to allow 64bit addressing and registers, among other things. It is not a complete architecture without 32bit support.

Why is this stuff so hard for some people to grasp is beyond me at this point.

Edited by LogicalApex, Feb 28 2013, 6:27pm :

At this point this is just bickering on words. x86-64 fully supports 64-bit computing and fully supports 32-bit computing. It can be considered a "true" 64-bit chip in the sense that it fully supports 64-bit computing, and it can also be said not to be a "true" 64-bit chip in the sense that it doesn't only support 64-bit and is based on an existing 32-bit design. So the debate isn't about the nature of the chip but what does it mean to be a "true" 64-bit chip, which is fairly pointless.

May I present a question the answer of which will also give the correct answer to this petty quarrel.

Can you run real mode (16-bit) DOS on Ivy Bridge/Piledriver?

>>So the debate isn't about the nature of the chip but what does it mean to be a "true" 64-bit chip, which is fairly pointless.<<

I would respond, research the Itanium Architecture. And use this as a definition: http://en.wikipedia.org/wiki/64-bit_computing

does the processor in question meet this criteria?

Without further qualification, a 64-bit computer architecture generally has integer and addressing registers that are 64 bits wide, allowing direct support for 64-bit data types and addresses.

neufuse said,

you do notice one keyword there right? "current"?....


The Itanium is not really a current product, it's more like a Zombie that hasn't vanished from the market yet...

Phouchg said,
May I present a question the answer of which will also give the correct answer to this petty quarrel.

Can you run real mode (16-bit) DOS on Ivy Bridge/Piledriver?


Yes it can - (not on Windows 64bit though)

MorganX said,
http://en.wikipedia.org/wiki/64-bit_computing

does the processor in question meet this criteria?

Without further qualification, a 64-bit computer architecture generally has integer and addressing registers that are 64 bits wide, allowing direct support for 64-bit data types and addresses.

The x86-64 has integers and addressing registers that are 64-bit wide. When you add 2 64-bit integers on x86-64 it's one add operation operating on two 64-bit operands and it drops the result in a 64-bit register.

Dr_Asik said,
The x86-64 has integers and addressing registers that are 64-bit wide. When you add 2 64-bit integers on x86-64 it's one add operation operating on two 64-bit operands and it drops the result in a 64-bit register.

Then without further qualification, we can say it is a 64-bit architecture that probably won't be running all 64-bit code?

Your previous response, that both sides are right depending on the criteria is probably best.

LogicalApex said,

Then maybe someone should send the Intel Engineers a memo that they don't know what their own CPU architectures are...

Source: http://software.intel.com/en-u...-64-ia-64-architecture-mean

You can't have an X86-64 CPU without 32bit (X86) as X86-64 is, as I said earlier, extensions to X86 to allow 64bit addressing and registers, among other things. It is not a complete architecture without 32bit support.

Why is this stuff so hard for some people to grasp is beyond me at this point.

I was thinking the same thing :-)

x86 is an instruction set. It evolved from 16-bit designs and then 32-bit versions came along. Now there's a 64-bit version. Current chips implement all three, but there's nothing stopping you from making a 64-bit only x86-64 chip. No one bothers, because there isn't really a market for that (nor a particular advantage so far as I can tell).

It isn't like your x86-64 CPU is running 32-bit most of the time and in exceptional cases "does something 64-bit." If you're running a 64-bit OS, it is running fully 64-bit, just as 64-bit as an Itanium chip. If you go run a 32-bit program, its 32-bit mode kicks in. But if you don't do that, then your CPU is only ever a 64-bit CPU. Just like 64-bit ARM CPUs are "fully" 64-bit, even though some ARM CPUs are 32-bit. You can essentially separate the instruction set (x86 or ARM) from the "bitness" of the CPU implementing it.

Also, Intel did not design it. There's a reason we on the Windows dev team generally refer to it as AMD64 (our build flavors these days are x86, amd64, and arm).

Caveat: x86-64 is more than just a 64-bit implementation of x86. It took the opportunity to make other changes (new instructions, more registers, etc). When you run 64-bit code on it, it actually works quite differently from 32-bit x86 code (i.e. calling conventions are markedly different - can take some getting used to for a developer debugging it).

MorganX said,
Then without further qualification, we can say it is a 64-bit architecture that probably won't be running all 64-bit code?
It's a multi-purpose architecture that fully supports both 64-bit and 32-bit computing. For all intents and purposes it's as good a 64-bit CPU as any other, but in a sense you could call it not a "true" 64-bit architecture I guess, although without further qualification that'd likely elicit the kind of feedback we've seen on this thread.

Edited by Andre S., Feb 28 2013, 8:56pm :

MFH said,

Yes it can - (not on Windows 64bit though)

Indeed. In fact, they even start in this mode to show us the lovely BIOS, which is entirely 16-bit stuff. Before it loads EFI, of course, which is out of the scope of this crappy debate.

Now - if it can run 16-bit code, is it 64-bit, 32-bit or 16-bit processor? Or either and all at once?

So it happens that x86-32 architecture is an extension of 16-bit counterpart. AL and AH makes AX, AX extends to EAX, EAX extends to RAX. Ain't it fun, when logic breaks and only apex feeling remains?

Phouchg said,

Indeed. In fact, they even start in this mode to show us the lovely BIOS, which is entirely 16-bit stuff. Before it loads EFI, of course, which is out of the scope of this crappy debate.

Now - if it can run 16-bit code, is it 64-bit, 32-bit or 16-bit processor? Or either and all at once?

So it happens that x86-32 architecture is an extension of 16-bit counterpart. AL and AH makes AX, AX extends to EAX, EAX extends to RAX. Ain't it fun, when logic breaks and only apex feeling remains?

Beat me to it! One of the reasons why IA-64 never took off is because the bootloader environment would require EFI and GUID Partitions (this is why Intel designed it) because on a normal X86/X86-64 processor you can switch between 16 bit, 32 bit and for x86-64, 64 bit (full 64 bit boot is being worked on and expected to be completed for the next version of windows). This meant the PB environment had to be different.

Anyway that, as you said, is out of the scope of this argument. As you said, is an x86-64 bit processor a 16, 32 or 64 bit processor? It is none of them? As far as I'm concerned it's all of them depending on what they are doing. So in summary, just because a processor may ONLY be able to run 64bit registered instructions does not mean that one capable of running 16, 32 and 64 bit instructions can't be one.

So yeah, Intel builds it's (successful) processors off previous designs which keeps compatibility. x86-64 is a 64 bit capable processor like it not there is no such thing as a "true" 64 bit processor I mean it either supports it or it doesn't.

Celestial Being said,
Hmm, could someone enlighten me cause I have no idea how this work. It's a x86 Processor but it can decode x64 instructions? How would that work? I am curious about these tech stuff.

Ok, I'm going to be talking a bit out of my ass, but I'm going to try to go into a bit more detail.

Modern processors (well, actually this has been true for many many years) are hybrid in design. At the core, they tend to be more similar to what people would refer to as RISC-like processors. These processors execute what is commonly referred to as microcode (it is much more complicated than I'm making it out to be.) These RISC-like processors can tend to be very different from generation to generation in the same CPU line and the internal microcode can also change greatly. But all of this is hidden from us.

Built around this RISC-like core are the CISC instruction decoders and pipelines that you refer to. These decoders read the x86 or x64 instructions and convert them to the internal microcode. Those internal instructions are then fed to the core RISC-like processor. The important thing to remember is that the internal microcode looks nothing like the CISC instruction sets we are use to. It can be said that the processor is a very fast emulator. It is just that the emulator is specifically design to "emulate" specific instruction sets very fast (i.e. x86 and x64). Another indication of this disjoint between the "emulated" instruction set and the internal RISC-like core is that the internal core often has many more registers that have been abstracted away from such "emulated" registers as EAX, EBX, etc. One of the reasons this is done to allow some of the more advanced performance improvements.

As far as the notion of if AMD64 is x86 processor plus more stuff is basically just incorrect. Does it support x86? Yes, and it does it very well. Does it support x64? Yes (and if you look at the x64 instruction set, it is vastly different from the x86 instruction set).

The problem revolves around the notion of what is a microprocessor. If you take the myopic view of a microprocessor is defined as "what instruction set it implements", then yes, AMD64 is x86+x64. But that just doesn't tell the whole story.

Imagine that you live in a neighborhood where it is required that every house looks exactly the same as the previous house (the instruction set) but you are allowed to add a few frills on the outside to make your house worth more (this would be such things as enhanced instructions such as SIMD that Intel and AMD add to make their processors distinctive and "better"). The good news about your house is that your absolutely nuts neighborhood association doesn't care one bit how you build or decorate the inside of your house. (This would be the internal decoding layers, pipelines, RISC-like core). So even though every house looks mostly the same from the outside, what is on the inside is totally different.

So yes, you can argue every house is the same while being able to argue that every house is different.

So back to the notion of is it and x86 processor or a true 64 bit processor. There is no doubt that it is a true 64 bit processor. For marketing reasons, however, it wears a cloak of x86 compatibility. Sure, there are probably some extra stuff they had to add to this true 64 bit processor to better support the x86 instruction set, but that doesn't mean it isn't a 64 bit process. If anything AMD64 is x64+x86, not x86+x64.

@torlac Brings back memory of Mac vs PC's old days. It was RISC vs CISC, yet few knew the Pentium had a RISC core even back then.