I suppose that if there is little cost to doing this, then they might as well do it. But I will be surprised if it really improves performance very much.

Here's a question for people: which would you prefer -- a single core CPU that has 8 execution units and Hyperthreading OR a dual core CPU with 4 execution units per core and the ability to do reverse hyperthreading?

I would think that it easier to divide a single core into two virtual cores via hyperthreading than to do reverse hyperthreading.

But perhaps that's a false choice? Perhaps we can have both, and let the OS decide which is appropriate for a given situation? So long as the extra circuitry to support this doesn't take up too much die space, I suppose the added flexibility might be worth it...

Uh oh! Looks like Shintai is gonna cry. Either that or say this is a fantastical wonderful idea, and that Intel is on top of things and AMD is just using a lesser inferior inmplementation that won't work.

*laugh*

Wasn't this technology a mechanism to keep the processor cool during intensive single threaded operation?

Basically you switch the process between cores on each context switch, or every N context switches. Leakage increases as heat increases, so by switching to the cooler core, leakage is reduced, and hence lower power consumption for the same amount of work.

LOL you guys are just slow, we already have 4 pages of discussions on this last week. :D

The diagrams are too simplified to be taken seriously, and there's just too much speculation. Hell i hardly believe AMD is going to pull off someting worth dropping a core.

you CAN'T split instructions like that. They're not always mutually exclusive. The problems worsens with longer pipelines. you have a write back occuring on one as an instruction needs to be processed on the other demanding that memory- those cores need some modifications to permit forwarding and other pipeline based functions to occur between cores. and continuous checking of the instructions to verify which core it's optimal to place it on (it sure as hel isnt going to be even odd!) and which conflicts with the other, when to stall and how bad a stall on one core would affect the other. If it is all there so be it, but It's nothing a mobo bios can add.

GamePc has a preview oh rly!!...WTF !!!??? intel is getting owned or what ?

Could the Core 2 Duo really be reverse hyper-threaded engineered, and hence the huge performance jump that we are seeing from Intel? Breaking all the hearts of loyal AMD fans?

And if the above statement IS True! Then you AMD fans can start rubbing your hands together and drolling all over your keyboards, cuz if a 3800 X2 can do half or more better with a Bios and software update, you haven't seen anything yet!! What will a 4200 X2 do? You could have a real world 4.2ghz processor hitting on all eight cylinders and running on nitrous.

It shouldn't be too hard to check for this in the current Core2 leak/reviews. Just run something that scales very well, but has a selectable number of threads. The difference between single and dual threads on a CMT CPU should be much lower than on a non-CMT CPU.

and no Woodcrest review yet ....so kitty was right, then ?

We heard about this a few months ago.
Interesting year for CPU's indeed.

AHAHA.... AHAHAHAHAHAH... AHAHAHAHAHHAHAHA *falls off chair*

"HyperThreading is bad, Mkay? P4s are hot, Mkay? Multiple cores visible to the user are bad, Mkay?"

Mkay.

Not much of a guru on processors but this "reverse hyper-threading" was always my dream of how dual core processors would of worked in the first place. But Alas, my dream will come true.
About performance? I look at it this way, if more pipelines will help video performance, then "reverse hyper-threading" should also be the best thing to hit the planet. I mean would you pick out the video card with the least pipelines?
Ahem, also. Could the Core 2 Duo really be reverse hyper-threaded engineered, and hence the huge performance jump that we are seeing from Intel? Breaking all the hearts of loyal AMD fans?

Quite fascinating how so many people can discuss a topic so extensively, yet devoid of any real information on the subject.

lol he said a review didn't say a credible one LMAO!

Will this so called "reverse hyper-threading" be available to true duallies with a bios/software update?

Here you go

http://www.tomshardware.com/2006/06/26/xeon_woodcrest_preys_on_opte...

and where is the Woodcrest review ?....stop the damn announcement and press release....give the ppl some nice HARDWARE !!!

fp!

interesting tech for both camps, but I rather have multi-threaded code instead.