PDA

View Full Version : Intel and Reverse HT


jAkUp
06-27-06, 12:15 PM
http://www.tweaktown.com/news/5891/is_intel_also_in_on_reverse_hyper_threading/index.html

a12ctic
06-27-06, 12:46 PM
CMT is just the ability to be able to turn one of the cores off in a dual core processor...

CaptNKILL
06-27-06, 12:59 PM
CMT is just the ability to be able to turn one of the cores off in a dual core processor...
I dont think so.

There seems to be a pretty thorough discussion of it going on here:
http://www.xtremesystems.org/forums/showthread.php?t=104178

jAkUp
06-27-06, 01:09 PM
CMT is just the ability to be able to turn one of the cores off in a dual core processor...

"Core Multiplexing Technology" certainly doesn't sound like something that disables one of the cores.

http://dictionary.reference.com/browse/Multiplexing

a12ctic
06-27-06, 01:44 PM
http://www.hardware.fr/medias/photos_news/00/17/IMG0017489.gif

theres an example of it

What be the things clear: the latter information are only of the rumors that do not deserve not all the attention that them is reach. not. For speak only last one, the option Core Multiplexing is activated by default on the card mother Intel and it is indicated that his désactivation... deactivates the second core: how to think in these conditions that she have an any report with the technology type enunciated above? In this case, being given his name one can think that she related to the communication of the two cores with the chipset through the FSB that is common.

CaptNKILL
06-27-06, 01:52 PM
http://www.nvnews.net/vbulletin/attachment.php?attachmentid=19016&stc=1&d=1151430568

From the looks of it, its either making a single thread run faster (by threading it somehow), or just making it use both cores (why? I dont know). The one thing it does NOT do is disable the core.

nemecb
06-27-06, 02:32 PM
Really starting to wish someone from either company would come out and confirm or deny all of these rumors. Sounds like a sweet technology if it works, but all we have are some pretty vague rumors at this point (an online translation and some "user-created diagrams" based on someone's interpretation of an Intel BIOS from what I've seen).

TBH though the more I think about it the more I don't believe this is really going to make that much difference. No doubt some very specific applications will benefit and some other apps may be rewritten to take advantage, but it seems a little too good to be true IMHO.

CaptNKILL
06-27-06, 02:34 PM
TBH though the more I think about it the more I don't believe this is really going to make that much difference. No doubt some very specific applications will benefit and some other apps may be rewritten to take advantage, but it seems a little too good to be true IMHO.
I agree.

shabby
06-27-06, 05:48 PM
AMD needs a software Patch from MS... this could work on Intel/Conroe too.

Amd chips already have this built in? *chuckle*

Dazz
06-27-06, 05:55 PM
I am sure Intel chips will perform worse due to the intplimation of the 2 chips on one core with FSB vs AMD inderpendent HT connection.
Still it better having more cores then trying to increasing the clock speed as it's getting harder and harder to do that.

walterman
06-27-06, 09:22 PM
Well, if they can manage to spread the instructions of a single thread across several execution units, yes, it will work. (it should need extra logic).

Unified caches between the cores will help a lot here to avoid cache coherency protocols (+1 point for intel here).

ViN86
06-27-06, 11:13 PM
CMT is exactly the same as reverse hyperthreading. it turns two physical cores into one "super" core.

-=DVS=-
06-29-06, 06:11 PM
Bah so much for RHT , Conroe already super fast , with this its gonna be ultimate CPU for games :drooling:

Sound_Card
06-30-06, 12:06 AM
Conroe = :headbang: