first of all there's no such things as "BE" "GE" suffix/prefix crap... there's just one and only brand here - Black Edition and as usual just one model per core, and by core I mean Brisbane (before Windsor), Toliman, Agena... and of course Kuma ;) second... well it's all wrong what you've wrote ;)Quote:
Originally Posted by FlanK3r
All along, AMD been saying that for dual core, the name stays at Athlon and NOT phenom.Quote:
Originally Posted by KeZzZu
Do we know the release date on the 125w 9950BE?
According to Inq , X2 6500+ is a 2.3GHz 2MB L3 K10.The TDP is huge , 95w.
Any bets against 6400+?
http://www.theinquirer.net/gb/inquir...athlon-branded
doesn't make much sense at all..
we have 45w 2.3Ghz Kuma rumours in one thread, then 95watt rumors in this one..
Add to that, we have 45watt 2.6Ghz (5000+), to 89watt 3.1Ghz (6000+) K8 available now.. Given a 2.3Ghz K10 is going to equal (on average) a 65watt 5600+, I don't quite see the point unless the TDP was much lower..
They HAVE to be disabled Agena's.. there's no reason for AMD to release physically larger than brisbane,native, kuma cores at clocks that make them slower whilst at the same time hotter.. Only @ 45watt would I belive it since they can charge the "EE" premium
Energy Efficient CPU's arrive from wafers done with different "recipe", this why those CPU's are more expensive... Kumas are arriving from standard wafers, and this 95W TDP is nominal wattage and in reality those CPU's on default will draw much less power.
But they're intend to be O'Ced on SB750 mobos, and under O'Ced conditions will reach that power, but at the same time will surpass in performance 6400+ Windsors, that are no more produced... in fact complete 90nm production is basically finished.
Thanks to ACC those CPU's will easily reach and in majority of cases surpass 3 GHz mark under air...For a reference G2 Brisbanes are topping 3.2 GHz under air, and this Kuma will definitely be faster.
No doubt that their low price will push sales of SB750 mobos making them the most desirable Athlon X2 CPUs.
I dunno, I still think they are disabled agenas and they don't clock well because they are the :banana::banana::banana::banana: bin. As mAJORD said, there would be no point in releasing a kuma cpu if it ran hotter, consumed more power, had a larger die and had only a slightly better ipc to make up for the poor clocking.
Newegg lists the 9950 as 140w.
Of course it is. The current 9950 is 140w. In a week or two we'll see a NEW STEPPING that only draws 125w for the same performance. I can't wait to see it overclock.Quote:
Originally Posted by squilliam
But the 140W 9950 uses less power than the 125W 9850 so u cant belive in this number.
on topic: im still confused about the 6500. if it is a 2.3 k10, i hope sb750 clocks it up to 3.2 - 3.3 on average, which would make it faster than the average overclock of a 6400+
I was right, here is confirmation that the dual core is nothing more than a disabled agena
http://www.hardware-infos.com/news.php?news=2357
Sorry to burst your bubble :),but did you even read that article?Quote:
Originally Posted by AliG
They link to PCGH website which used the X4 with disabled cores to simulate the performance of the X2 Kuma Black Edition..So no,we don't know if it is X4 with 2 disabled cores...
look at msi site, its listed as supported cpu
kuma B3 95W 6500+
http://global.msi.com.tw/index.php?f...&cat3_no=#menu
Not 6500+, just 6500! Its a Kuma and no Brisbane.
Quote:
Originally Posted by Nedjo
What?? No their not. K10s all have the same "recipe" as you call it. I work in semiconductor manufacturing and there is no such thing as a separate line for energy efficient cpus. Same line, same process. The way it works is that you have a test and sort area in a fab that bins cpu into different categories, one being how leaky(current) they are. The least leaky ones can be binned into energy efficient cpus or extreme edition cpus, then the less ones are binned down the line by decreasing frequency.
Perhaps you can share a bit of info here:) Without going into any detail that could cost you your job, How common is it for chips that have been binned for high end to be used in a lesser range? As an example, Would intel bin loads of extreme edition capable chips and end up using them as Q6600 due to lack of demand?Quote:
Originally Posted by qurious63ss
But wouldnt the 'B3' add-on actually clearify this is Agena? If it was a new chip really than I think it's a strange way to introduce a stepping:confused:Quote:
Originally Posted by duploxxx
Agreed... very weird. Two possibilities, a) Kuma is an Agena with 2 cores disabled or B) they taped out new stepping which is a Agena cut in half.Quote:
Originally Posted by Rammsteiner
Option a is expensive (high costs), option b is cheaper and less costs. But one would think that option B would be accompanied by a step revision. Hmmmm. weird.
Yes, but not likely. Most likely it would be a Q6700. Remember you have specs for each SKU so even with the Q6600 you can have chips that have different performance but would still be binned Q6600 as long as they stay within the power spec at a given frequency.Quote:
Originally Posted by gallag
ahhhh, now it makes sense! good to know ;)Quote:
Originally Posted by qurious63ss
I guess all of the questions and speculations will be answered they will be announcing it todayQuote:
(product details & positioning available on September 1, with price announcement)
do you have insight on how AMD's propriety APM tech works?Quote:
Originally Posted by qurious63ss
Wouldnt it just be on teh same wafer and the extact same process fabriaction. Since you would either need new wafers or a new mask and new electrical verification? I doubt that would be economically viable for AMD. Not even Intel do that for low power parts.Quote:
Originally Posted by Nedjo
No I don't, but if it's anything like what we do then they are basically tightening the limits (i.e. within wafer uniformity for dep or oxide steps) on their process by changing gas ratios, time, temp, or possibly different preventive maintenance on a tool. Small tweaks like this help with reducing transistor leakage.Quote:
Originally Posted by Nedjo