Damn i wonder what revision of the Badaxe that is, if its the 304 or the 306.
Printable View
Damn i wonder what revision of the Badaxe that is, if its the 304 or the 306.
So kentsfield is the successor to conroe, and will it work on lga755 and use ddr2?
ddr3 probably later when intel releases new chipset for ddr3
Yeah and apparently the conroe borads will work with it as well..just the word around not fully confirmed.Quote:
Originally Posted by andyisc00l
thats brilliant, and kentsfield isn't just a change to quad core is it? its also a change in architecture right?Quote:
Originally Posted by MnM
It's basically just 2 conroes on 1 pcb ;)Quote:
Originally Posted by andyisc00l
I like the way you say "just"..:D as if conroe isn't enough of a monster, now they make it a double!:p:Quote:
Originally Posted by s7e9h3n
Pat Gelsinger said Conroe doesn't have HT.
HT will resurface with the Nehalem generation of cores , mid 2008.
and you know that how?Quote:
Originally Posted by savantu
Because Pat Gelsinger said that current Conroes do not have HT. And HT is on the internal roadmaps for Nehalem that have been leaked. Simple as that.Quote:
Originally Posted by Fuji
Now we'll see what turns up in reality. :fact:
ooh looking good, can't wait
Screw HT, it's a waste IMO.
Huh ? If you'd bother to understand the concept I'm pretty sure you'd change your opinion.Quote:
Originally Posted by Nanometer
HT was good when it first started, but now it's useless vs real core.
I don't see how a 4 core HT CPU (8 logical core) could be better than 4 core when 98% of the software/games won't even use the first 2 core completely.
Sure on paper ,it's nice. But in reality, dual core is still far from his full potential.
Most people I know don't use virtulization. Why include virtulization enhancing features on a chip?Quote:
Originally Posted by Vric
Nehelam is still going to be dual core on most desktop chips, if I recall correctly. HT allows better use of the execution resources. If a process is truly multithreaded, it will benefit. Assuming it doesn't add unreasonable complexity/power useage to the chip, then why not have it added?
OT: Nehelam backwards is MALE HEN :p:Quote:
Originally Posted by ethernal
Why are ppl talking about Nehelam? Penryn is the next core....45nm shrink of Conroe... :)
Ok....Penryn backwards is Nyrnep :wth:Quote:
Originally Posted by freecableguy
FCG, will you be able to post benchmarks or are you limted by NDA?
(its okay if you are. i completely understand)
Because People are typically more interested in dramatic architecture shifts like Conroe, and not optical shrinks with more cache or not like Cedar Mill and Brisbane.Quote:
Originally Posted by freecableguy
I am only half heartedly interested in Penryn, Intel's 45nm process is a ways off for now Q3 2007 I bet at the earliest. I haven't quite gotten over Conroe yet, it hasn't even arrived for god sakes. :D
What I am most interested is Bloomfield which is probably going to be the most "advanced" core based on Core Architecture after that we should be looking Nehalem derived products. Apparently is a monolithic core with an unkown amount of shared cache :slobber:, I am hoping for 8-12MB or it can you imagine all that cache dedicated to a Single Threaded program. ;)
Though Allendale and Conroe optical shrinks Wolfdale and Ridgefield are of some interests, as is Penryn which is the optical shrink of Merom, all these cores will be basically be like their older versions on 65nm with some minor tweaks and more cache basically and on the newer 45nm node with perhaps some higher FSB speed.
That extra 2% of SW might account for 75% of the SW revenue in the server world in rest assured that SMT is used there ( From Xeon MP to Power 5 ).Quote:
Originally Posted by Vric
And that isn't on paper .It's real life.
Nehalem will be quad-core or more.Quote:
Originally Posted by ethernal
Pat Gelsinger said :
- at 90nm we had a little DC ( 10% )
- at 65 we have most DC (70%) and a little QC ( 10% )
- at 45nm we have most QC ( 70% ) and a little OC ( 10% )
The bad thing about SMT is that it is truly difficult to do ( why do you think AMD didn't bother ? ) .For the P4 for example it used an extra 5% die space but accounted for 90% of the problems at debuging.. It was dropped from Core familly because of time to market and thermal issues.
45nm woohoo.. to think our grand kids will have cheap <45nm based cpu systems in the future LOLQuote:
Originally Posted by freecableguy
will there be a tecnique after 45nm ?(i mean one that was mentioned till today in some roadmap or so ?)
you mean for smaller process? yes its called intel tri-gate it is based on 3d transistors and should work for 32nm and 22nm :woot:Quote:
Originally Posted by realsmasher