So, this will effectively double the processor's performance... I guess?

I question how much utility you could gain from this.

First off there's the obvious fact that this is pretty much a stop-gap to wring more performance out of single-threaded apps that can't easily be multi-threaded...

But unless there are an absolute gang of independant instructions waiting to be executed on an Athlon, then this will probably not mean much... When it is utilised it could mean as much as 100% improvement. But more likely 50-75%...

But I'm starting to think about Amdahl's Law, and that this can only help when you have that many independant instructions waiting to be executed, which might happen, say, only 25% of the time. That means that we're looking at a 15% or so improvement.

I'd make a joke about Hyper-threading making a 15% performance hit somewhere here and so RHT being an appropriate name, but meh...

This will probably be very useful on the K8L however, what with the checking dependancies on retire rather than on load. Then you'll have a lot more independant instructions flying about, and big gains could be seen :-D

An interesting technology, anyways...
-Mole

Concatenated Cores.

I just want to see Intel try to claim they own the word Core.

Bah. AM2 only? I knew it was too good to be true for my "lowly" 939 X2.

Dammit!

Nifty stuff, though. If true. If it works as advertised, I'd say a 25-50% gain in performance (in single-threaded apps; probably most noticeable in games) from a simple software update is none too shabby and, as far as I'm aware, unprecedented in the CPU market.

Also, it makes sense that AMD would delay the announcement from the launch of AM2 to the launch of Conroe to get as much effect from it as they possibly can. They are going to need something to steal at least a bit of Intel's thunder, and this may just do the trick.

Still, I hope the AM2-only rumor is false. But since I bought a 939 X2, I know it won't be ;)

I wonder what the marketing name is going to be. You KNOW it's not going to be "reverse-hyperthreading". How about "Singularity"?

Hi....
Have you thougth of the possibilities that AMD actually know that they can get 50% or more in games apps already and they also expect that intel can also do it by reverse engeniering to achieve the same results by the time all the bugs have been all cleared out of the 2 cores as one approach.

Enter: 4X4 wide market availabily.

By the time Intel could cache up with the single slot solution AMD could still have a HUGE performance advantage over single socket single/dual core solutions available

1 X2 =130-170% 2 X2=250-310%
1Core Duo 130-170%

Since they are pushing the 4X4 first, I am lead to believe that "if" they already know they can get 50% or more from this 2 in 1 approach they are mostly doing it to get at least one year of advantage over intel since they will have to do the core_update-custom_osdriver-Dual_socket_platform_launch all togueter.

If it is really possible to get a 50% out of a single core(2.0Ghz=3Ghz) and maybe 80% of a dual socket/dual core (2.0Ghz=3.6Ghz) the same ppl that could have upgraded to intel core duo will just get a 4x4 with just one 300 bucks cpu
http://www.newegg.com/Product/Product.asp?Item=N82E16819103735
and upgrade the rig later with same model cpu for 150-250 bucks when required...ie UT2007 or anything like it.

This way AMD could aim to rule the high end and mid range with one stone and they can keep their current price structure.

I was expexting to upgrade my existing socket 754 rig with only a socket 939 because I did not see that much difference in between DDR and DDR2 to justify a whole jump from 754 to AMD2 since I could get 2GB of High spec (2-2-2-5) ram now and then use it in dual channel in a socket 939 rig later and get equivalent performance with reduced cost. Now I think I would hold the memory purchase to see it if it would be worth to wait just a little longer.

The K8L upgrade will be available for the Opteron family in about a year if it is not postponed more that 3 months. I also wonder if the AMD2 family will get the K8L upgrade within a year after the release for the Opteron.

If "only" a bios upgrade and maybe a driver update enables AM2 K8L cores to be used on what will be the current 4x4 systems is currently in the plans for AMD..... Ahhh!! How much fun it could be.

i'm not sure how much performance unifying execution across both cores will bring
on one hand it will technically double the number of execution units available per process
on the other hand, is the K8 core execution really getting a backlog?
i had always though it made good allocation of instruction level parallelism

i definitely know we won't see a doubling in performance though
there would be some tricks in terms of pipelining in passing the results of an instruction just out of an alu onto something just going into an alu
i don't see happening with 2 separate cores unless there are crossbars between "everything"

k8 is a 3-issue core if i'm not mistaken
so this will effectively give 6-issue wide
it will need to rely on the fact that there are processes can be broken down to a 'better than 3 instructions at a time' level
this will help clear one thing up for me though
if there is no significant performance boost, it would suggest a 4-issue wide core maybe not be as much of an advantage as first thought

i also don't know why it's "reverse hyper-threading"
by the sound of it they could implement a hyper threading thing in there
"hey we've got 2 execution units spare. what do we do with them?"
"run this thread in there. it's from the same process and using the same memory locality"

Yeah, I heard about this a few months ago ... with regard to the K10.

stop this speculation, you are making a mockery of AMD.

I hope this will work for socket 939 chips!

This seems all based on a couple of patents. I really think this is another form of FUD. They are probably intoducing a few new instructions in their next revision for better multi-processor threading support. I wouldn't go beyond that.

Any word on Opterons?

DRTFA

I remember a few months back anandtech saying in a rather tantalyzing way that AMD had a surprise coming in the next few months. I wonder if this is it.

If there is a 50% increase in single threaded performance (like most old and current games), I can see many gamers (not neccesarily hard core) hold on to their old systems longer, or change their minds about upgrading to a conroe system (along with other types of users...).

Is this supposed to be only for AM2 or for S939 X2's as well?

This was awfully quick..... I remember just hearing about R-HT about a month ago from AMD and now they are already done ? Already in released hardware ? Wow.

So many announcements from both parties, I will believe them when I see them, to many announcments not enough products...