AMD will let out "anti Hyper-Threading" in this year: the secret weapon Socket AM2

Status
Not open for further replies.
Anyone willing to place bets on this? ah crap.. probably not allowed according to forum rules, is it? :'(
Time to go to the real gamblers: stock market.

Anyway: I have no idea what AMD is doing, so I am taking a wild guess. Since this information is from some 'random' overclocking site that I have never heard of or seen other references too, i will take it with a grain of salt and say that I do not believe that this will happen.
 
I think is what it's saying is that the AMD cores(including X2) can dynamically go from being single core, dual core, or quad core setups at random based on what they're trying to do and how many true cores were included.

A single core would be 32 bits of processing. While if you make 2 cores look like one you have 64bits of processing. Quad would be 128bits. I'm guessing it's mainly a way to accelerate SSE instructions. Since 128 bits of true full speed processing would be great for vectors.
 
I've heard about this before, a long time ago actually. I think they said something about two cores working together on one process...sounds good. It can bring real use and support for dual cores.
 
Anarchist4000 said:
I think is what it's saying is that the AMD cores(including X2) can dynamically go from being single core, dual core, or quad core setups at random based on what they're trying to do and how many true cores were included.

A single core would be 32 bits of processing. While if you make 2 cores look like one you have 64bits of processing. Quad would be 128bits. I'm guessing it's mainly a way to accelerate SSE instructions. Since 128 bits of true full speed processing would be great for vectors.
Click to expand...

Lay down the crack pipe man... :eek:
 
With Dual processors and dual graphic cards, do we need 1000W PSU? Electrical company will love it.
 
Flyboat said:
With Dual processors and dual graphic cards, do we need 1000W PSU? Electrical company will love it.
Click to expand...

... getting their paychecks fatter, no?
 
Anarchist4000 said:
I think is what it's saying is that the AMD cores(including X2) can dynamically go from being single core, dual core, or quad core setups at random based on what they're trying to do and how many true cores were included.

A single core would be 32 bits of processing. While if you make 2 cores look like one you have 64bits of processing. Quad would be 128bits. I'm guessing it's mainly a way to accelerate SSE instructions. Since 128 bits of true full speed processing would be great for vectors.
Click to expand...


What?!?
 
What?!?
Click to expand...

Make one big core instead of a couple smaller ones whenever it is beneficial to do so.

When dealing with 3D space you're typically using vectors(3-4 floats depending on what you're doing). A typical 32 bit processor likes doing math on single 32 bit floats. With SSE you start running the same instructions across multiple floats. So with quad core(128 bit) you could do an entire vector at the same time without doing it in 4 steps.
 
ok...so supposedly it's "already" in the A64 X2's...but only for AM2? Why would the socket matter? I thought that the X2-939 and X2-AM2 used the same CPU "cores" whilst using a different mem controller.
 
nst6563 said:
ok...so supposedly it's "already" in the A64 X2's...but only for AM2? Why would the socket matter? I thought that the X2-939 and X2-AM2 used the same CPU "cores" whilst using a different mem controller.
Click to expand...
There are other differences, such as virtualization..
 
Intel has something like this as well don't they? I remember reading something about it...I will look around for the link tomorrow
 
Damn.. If this is legit, AMD's launch of 4x4 is not so bad of an idea.. 4 cores working as two.. might cost a little more, but those who can will have a crazy little system. And most people will be able to if AMD allows any Dual core A64 processor into their 4x4 product. (i wonder if all four cores can work as one.. both processors can acess the same memorybanks in 4x4.)
 
Anarchist4000 said:
Make one big core instead of a couple smaller ones whenever it is beneficial to do so.

When dealing with 3D space you're typically using vectors(3-4 floats depending on what you're doing). A typical 32 bit processor likes doing math on single 32 bit floats. With SSE you start running the same instructions across multiple floats. So with quad core(128 bit) you could do an entire vector at the same time without doing it in 4 steps.
Click to expand...

The whole point of SSE is that a vector can be done in one step. On a single core, on a single exectution unit for that matter. Conroe will in fact handle multiple full SSE vectors in a single clock cycle. Breaking an SSE vector up across multiple cores is an extrodinarly inefficient way to doing things.

Anti-hypertheading is just distributing instructions from 1 thread to both cores in an attempt to exploit some degree of parallel exectution with what is compiled as a sequential code path.
 
FreiDOg said:
The whole point of SSE is that a vector can be done in one step. On a single core, on a single exectution unit for that matter. Conroe will in fact handle multiple full SSE vectors in a single clock cycle. Breaking an SSE vector up across multiple cores is an extrodinarly inefficient way to doing things.

Anti-hypertheading is just distributing instructions from 1 thread to both cores in an attempt to exploit some degree of parallel exectution with what is compiled as a sequential code path.
Click to expand...

yeah
 
drizzt81 said:
Anyway: I have no idea what AMD is doing, so I am taking a wild guess. Since this information is from some 'random' overclocking site that I have never heard of or seen other references too, i will take it with a grain of salt and say that I do not believe that this will happen.
Click to expand...

Amen to that - as much as I'd like to see it happen. I highly doubt AMD wouldn't even hint at something this big that during their last analyst day - especially with these conroe numbers floating around. Well, I'll sit here and wait...
 
FreiDOg said:
Anti-hypertheading is just distributing instructions from 1 thread to both cores in an attempt to exploit some degree of parallel exectution with what is compiled as a sequential code path.
Click to expand...

If it's true I'll eat my Athlon!

You heard it here first!
 
Big Fat Duck said:
sorta like SLi, except for processor cores?
Click to expand...

Well, considering the ATI/AMD rumors, CrossFire would be more appropriate. :)
 
I'm pretty sure AMD's "Secret Weapon" is the price cuts they just announced recently. I don't think their reverse HT technology is close to done yet.
 
enelson125 said:
I'm pretty sure AMD's "Secret Weapon" is the price cuts they just announced recently. I don't think their reverse HT technology is close to done yet.
Click to expand...
QFT.

When AMD mentioned it earlier in the year, it was as a possible feature for the post-K8L core. i.e. AMD's next generation core coming in 2008.

overclockers.ru is just posting the rumor that was circulating a couple of weeks ago as fact now. I would believe that if reverse HT has instruction extensions that AMD might be releasing software specifications for it this year. Otherwise this is just completely bogus.
 
pxc said:
QFT.

When AMD mentioned it earlier in the year, it was as a possible feature for the post-K8L core. i.e. AMD's next generation core coming in 2008.

overclockers.ru is just posting the rumor that was circulating a couple of weeks ago as fact now. I would believe that if reverse HT has instruction extensions that AMD might be releasing software specifications for it this year. Otherwise this is just completely bogus.
Click to expand...

for it to realy work right it needs to be passive and that could mean its in K8L or even the curent AM2 cores but just need the driver to make it work so who know

but thats just a guess
 
all i got to say is you amd fans are truely desperate to hold unto any sliver of hope that the amd wont get trambled by intel for the next coming months. amd mentioned doing this earlier this year and in their implementation of this new technology they would need to do more than just change the way the cpu worked. they would also need to change the way a motherboard worked as well. im shocked that you guys are actually believeing the rumors of some backwaters pc site that you probably never heard of up untill that link you just saw.
 
wow...now there's a flamebait post if I ever saw one.

what's wrong with people talking about rumors? Some believe it, some don't. I don't see a need to call an entire group of people desperate. Hell, if that's the case I guess "all you intel fans" have been "desperate" for something like Conroe to come along for the past few years? yeah....whatever. flames go both ways and doesn't make the topic of conversation go anywhere but lock and b7.
 
DemonDiablo said:
all i got to say is you amd fans are truely desperate to hold unto any sliver of hope that the amd wont get trambled by intel for the next coming months. amd mentioned doing this earlier this year and in their implementation of this new technology they would need to do more than just change the way the cpu worked. they would also need to change the way a motherboard worked as well.
Click to expand...
reference?
DemonDiablo said:
im shocked that you guys are actually believeing the rumors of some backwaters pc site that you probably never heard of up untill that link you just saw.
Click to expand...
if you had actually read the thread, you'd have realized that most people are extremely sceptical about this issue. nice flame-bait nonetheless. Next time, come with facts, if you do not mind.
 
not even close to flamebait. thread is overly useless as it is. if it got locked it would be for the better rather than spreading more fud as fact.

drizzt:

reference to what? what i said about amd's plans on this subject before? check the hardocp's archives. the article was up on it maybe back in april. i see far more people in this thread getting ready to praise this horrible rumor as fact than fiction.

i guess i just snapped b/c i see the intel guys doing the exact samething with HT on conroes...yea ok there goes another rumor thats turning into pointless fact. to many people running around with tinfoil hats on imo.

nst:


the second you saw me as some intel fan boy, is when you completely failed to "figure me out". could i have put my thoughts in nicer "prettier" words sure i could i have, but some people need a good slap across the face to wake them up to the reality of the world.

i dont see problems with disscussing rumors that have some legitimate facts around them but this? this isnt even a rumor is just a steaming pile of bull. at least link to a more credible site on the topic of "RHT" for some serious debate imo.

but hey feel free to to see my thread as flamebait and feel free to discuss this topic to your hearts content.


Mod Edit: Tinfoil hat flamebait threads? No ... that hasn't been happening much of late at all in here .

Everyone please recall: http://www.hardforum.com/showthread.php?t=760666

Additionally, Genmay and the " Soapbox " may be found here: http://www.hardforum.com/announcement.php?f=6&announcementid=50 ... join it's fun. Thank-you ~ Papa-Ming
 
Status
Not open for further replies.
Back
Top