Core Multiplexing technology???

Printable View

Show 100 post(s) from this thread on one page

06-28-2006, 10:26 AM
theteamaqua

so how is single core different from this core multiplex technology?? they are different right?? CMT turn multi thread into 1 core or 2 core (kentsfield) right??
06-28-2006, 10:39 AM
Thorburn

Quote:

Originally Posted by theteamaqua

so how is single core different from this core multiplex technology?? they are different right?? CMT turn multi thread into 1 core or 2 core (kentsfield) right??

If its the same as what is shown in the Anandtech article then Core Multiplexing would allow an idle core to do speculative threading work instead of sitting doing nothing. If you have multi-threaded code then it goes to work on the other thread instead automatically.
Single core mode turns off the second core completely, no operations carried out whether you have multi-threaded code or not.
06-28-2006, 10:47 AM
PetNorth

Quote:

Originally Posted by yoti

Maybe here we have answer?
http://www.anandtech.com/tradeshows/...spx?i=2507&p=7

No, Mitosis, at this moment, is only a paper project, far to be implemented...
06-29-2006, 11:48 AM
Donnie27

Quote:

Originally Posted by PetNorth

No, Mitosis, at this moment, is only a paper project, far to be implemented...

HAHAHA! But RH/T is real and proven right?
06-29-2006, 11:52 AM
BrownTown

dunno about him, but they are both BS in my book
06-29-2006, 11:54 AM
i||uSi0n^

its too early to say anything until both AMD and Intel give out some info.

i||uSi0n^
06-29-2006, 12:04 PM
Donnie27

Quote:

Originally Posted by i||uSi0n^

its too early to say anything until both AMD and Intel give out some info.

i||uSi0n^

QFT!
06-29-2006, 12:55 PM
informal

As for Core2 Duo,there is no clustered mulithreading in it.Just simple no.Mitosis is a 5 year project with possible introduction in Nehalem core(which is not so distant time).

And as for AMD's clust. multithreading ...Well ,Rahul Sood just confirmed it on his blogspot(in comments section on his latest blog entry).

So expect to see it introduced on the dawn of Conroe's release.
Performance were not mentioned,but Rahul said that it "Sure helps" in the fight against Core2 Duo.We will see how much ;)
06-29-2006, 01:31 PM
PetNorth

Quote:

Originally Posted by Donnie27

HAHAHA! But RH/T is real and proven right?

I thought this thread was about CMT. Did I miss something?
06-29-2006, 02:13 PM
Tony

Quote:

Originally Posted by PetNorth

I thought this thread was about CMT. Did I miss something?

Its about Intels new tweak yes, but we are bound to have talk of AMD's responce also as in reality they are the same thing.

It makes me wonder if you guys don't want more speed reading this thread, some of you seem unwilling to embrace new tech...

Its just a shame I haven't dug any further info as of yet...
06-29-2006, 02:40 PM
PetNorth

No Tony, the question is that some of us are unwilling to embrace simple rumors, speculations, inventions or whatever, without any near or remote fact.
06-29-2006, 03:23 PM
informal

Quote:

Originally Posted by Tony

Its about Intels new tweak yes, but we are bound to have talk of AMD's responce also as in reality they are the same thing.

Amd's responce to what???:confused: To disabling one core/using all the L2 cache?Amd use exclusive caches for their multicore...

For 1000th time,there is no clustered multithreading in Core2 Duo.Amd doesn't have to respond to anything since it is not there.
06-29-2006, 06:42 PM
Donnie27

Quote:

Originally Posted by PetNorth

I thought this thread was about CMT. Did I miss something?

Explain that to the others link on post #133 LOL!
06-29-2006, 06:54 PM
Donnie27

Quote:

Originally Posted by informal

Amd's responce to what???:confused: To disabling one core/using all the L2 cache?Amd use exclusive caches for their multicore...

For 1000th time,there is no clustered multithreading in Core2 Duo.Amd doesn't have to respond to anything since it is not there.

Here, try out this response?

Quote:

Originally Posted by Hans de Vries

Originally Posted by back in April
Name: Hans de Vries (hansdevries@chip-architect.com) 4/6/06

ten9 (jhdjksa@huhu.com) on 4/6/06 wrote:
---------------------------
>Hello Hans,
>
>Do you think there is a fourth decoder like Gipsel argues
>here:
>
>http://www.aceshardware.com/forums/...60237&forumid=1
>

These are the micro-code ep(roms) for complex cisc
instructions. They operate as one single memory since
they all get the same address from the micro-sequencer
which handles the complex instructions. Going from 3 to
4 therefor doesn't say anything at all, it's just more
memory, or the same amount of memory with larger cells.
The rest of the architecture is visibly 3-way. The re-
order buffer, the integer schedulers, the integer ALU's.

>And what do you think about the floating point units?
>

It's virtualy identical to existing K8's. There are also 3 HT
units and not 4 as claimed by somebody at aces. It's not a K8L.

>Regards,
>
>Ten9
>

It seems to be an earlier prototype in which they were still
experimenting with the new much denser L2 cache. The fact
that they left open such a wide area with the size of the
old cache means they were not sure enough about it at that
time. Showing the cache now would imply that it must be OK.

More cache would be the most important thing for AMD to
counter Conroe's integer IPC.

Regards, Hans

So much for the other BS!
06-29-2006, 06:56 PM
Donnie27

Now it reamins to be seen what and how Core Multiplexing will work but at least let's wait and see!
06-29-2006, 11:06 PM
JumpingJack

Quote:

Originally Posted by PetNorth

No, Mitosis, at this moment, is only a paper project, far to be implemented...

If you are interested in more technical details of Mitosis, here are a few Intel authored paper:

http://www-cse.ucsd.edu/users/tullsen/pldi2005.pdf published in June or 2005
http://www.ece.lsu.edu/tca/papers/collins-01.pdf another 2001 I believe.
http://www.cs.ucsd.edu/~jbrown/papers/sp-cmp.pdf
http://www.iccd-conference.org/proce...4/22310360.pdf

Just a handful, but Intel has been working on this for quite some time evidently, I have found papers (PDF files not electroncially available) dating back to 1996.
06-29-2006, 11:29 PM
taemun

Quote:

Originally Posted by JumpingJack

Just a handful, but Intel has been working on this for quite some time evidently, I have found papers (PDF files not electroncially available) dating back to 1996.

:eek: thats quite a long time ago lol

I wonder that Anandtech said 5 years (in 2005)... that would mean at least 14 years of research. Seems very long. Lets hope that CMT [exists and] is an implentation of Mitosis.
06-30-2006, 12:23 AM
Tranzmit

Interesting how for years we try to go multicore and what have you and the moment we get it we all try to get it to act like a single core again.

Obviously an extreme oversimplification but yeah. I really think we're all better off getting devs writing multithread happy code and hoping that the os's and the drivers and the hardware starts communicating together better.

I seem to remember reading ssomething from intel about how they're actually moving toward a somewhat cell like approach with lots of "special purpose" mini cpus that are much better at doing what they do best and having a single general purpose unit that can allocate the workload off to the other more specialized units but without such a penalty as Cell seems to have with it's non optimised code where it just falls over.

Oh happy first post to me. Been lurking for awhile but figured i'd better start somewhere :D
06-30-2006, 03:32 AM
ted3

Quote:

Originally Posted by Tranzmit

Interesting how for years we try to go multicore and what have you and the moment we get it we all try to get it to act like a single core again.

Happy 1st post! Singlethread performance is a bit interesting for me right now, next year i probably couldnt care less. But maybe i care again when/if i get a Kentsfield? Lets say a coming game is optimized for dualcore and cant utilize the 4 cylinders of Kentsfield, then this sounds like a great way to make Kentsfield a monster Twin.
06-30-2006, 03:50 AM
PetNorth

Quote:

Originally Posted by Donnie27

Explain that to the others link on post #133 LOL!

You know, I'm only responsible for my posts ;)
06-30-2006, 04:52 AM
Vassili

Quote:

Originally Posted by ted3

Happy 1st post! Singlethread performance is a bit interesting for me right now, next year i probably couldnt care less. But maybe i care again when/if i get a Kentsfield? Lets say a coming game is optimized for dualcore and cant utilize the 4 cylinders of Kentsfield, then this sounds like a great way to make Kentsfield a monster Twin.

I am sure they don't make it dualcore optimized but multithreaded...;)
06-30-2006, 06:40 AM
ted3

Quote:

Originally Posted by Vassili

I am sure they don't make it dualcore optimized but multithreaded...;)

Wish i could be as sure about that as you are... Parts of a games code can be executed in separate threads but it demands a very solid code to hand off everything to threading. Most games could prove to complex to do that, its very likely that the main logic has to execute in one thread, what remains is not enough work to utilize 3 cores (or even one core in some cases), those threads could too often just sit there and wait for the main thread with a Kentsfield.

Anandtech interviewed Tim Sweeney about this once, the future looks bright but you could also get the impression that multicores mainly give the developers the chance to add lots more bells and whistles. Good for realism, lets hope thats not all.. Just got to wait and see, i hope you are correct Vasilli, but i have my doubts for now :)
06-30-2006, 08:30 AM
Donnie27

Quote:

Originally Posted by PetNorth

You know, I'm only responsible for my posts ;)

Yes you are, my bad.
06-30-2006, 08:36 AM
Donnie27

Quote:

Originally Posted by ted3

Wish i could be as sure about that as you are... Parts of a games code can be executed in separate threads but it demands a very solid code to hand off everything to threading. Most games could prove to complex to do that, its very likely that the main logic has to execute in one thread, what remains is not enough work to utilize 3 cores (or even one core in some cases), those threads could too often just sit there and wait for the main thread with a Kentsfield.

Anandtech interviewed Tim Sweeney about this once, the future looks bright but you could also get the impression that multicores mainly give the developers the chance to add lots more bells and whistles. Good for realism, lets hope thats not all.. Just got to wait and see, i hope you are correct Vasilli, but i have my doubts for now :)

What if that second core could help the GPU, NIC and Sound cards? Folks are selling P-GPUs now.
06-30-2006, 11:10 AM
Turtle 1

Isn't this what multiplexing is really About?

http://www.intel.com/pressroom/archi...0040218net.htm

Intel Unveils Platform Strategy For Ultra-Wideband Wireless Technology ...
... complete specification for a multi-band orthogonal frequency division multiplexing (OFDM) technology-based ... For more information on IDF and Intel technology, visit http://developer.intel.com

I believe we will see Mitosis with the Nehalem core If AMD has RH/T now as many claim . We may see Mitosis on Penryn

Show 100 post(s) from this thread on one page