Sign in with
Sign up | Sign in
Your question

AMD 65nm in production now(5 months early)

Last response: in CPUs
Share
August 18, 2006 10:32:09 PM

http://www.fabtech.org/content/view/1757/2/

I guess they want to keep it quiet so they can sell out of all the 90nm first....
August 18, 2006 11:13:40 PM

lol,,, some rabbid intel fan boy will soon jump all over this saying its not a reputable site & its pure propoganda,,, & they will have to see it on 4 more sites & hold on to one before they even think about the possibility of any truth to it.
I always suspected AMD & Dell had some thing going on in the back ground for Dell to have stuck out is head & SLAP intel in its face.
I wonder what the performance is like tho,,, conroe has really made its mark performance wise.

sic em boyz.
August 18, 2006 11:17:40 PM

What I don't understand about the AMD roadmap is they say that the first 65nm will not be any faster, and are given the same names(x2 4200,4600,etc).... Then later they can make 40% faster models(or something like that)....

How can you go smaller without getting faster? Make the pathways skinnier, but give them farther to travel?!?!
Related resources
August 18, 2006 11:37:21 PM

Quote:
What I don't understand about the AMD roadmap is they say that the first 65nm will not be any faster, and are given the same names(x2 4200,4600,etc).... Then later they can make 40% faster models(or something like this)....

How can you go smaller without getting faster? Make the pathways skinnier, but give them farther to travel?!?!


Node shrinks do not give out free performance. And just because they're laying down 65nm (which isn't actually 65nm, it's 110nm litho that is subsequently shrank to 65 through a chemical process.) does not mean that the chips will even bin out like they think.

Intel has been sampling working 45nm Nehalem parts for a couple months now. Core 3 is gonna blow conroe outta the water.
a b à CPUs
August 18, 2006 11:43:56 PM

Quote:
http://www.fabtech.org/content/view/1757/2/

I guess they want to keep it quiet so they can sell out of all the 90nm first....


Besides the point, Intels already done 65nm for months now, and when they do scale AMDs down to 65nm, Intel still has the clock for clock advantage, 65nm will allow AMD to achieve higher clock speeds sure, but if they cant scale higher then Intels Core 2 Duo (atleast 20% higher clock speeds) then its a lost cause for now until a new design (K8L) comes out (wheres the benchmarks?).
a c 111 à CPUs
a b À AMD
August 19, 2006 12:07:34 AM

AMD does this....
look at thoroughbred vs palomino, .13 down from .18... the XP 1800 on both perform the same.

There was room for speed upgrades. and they did use it after. but a die shrink does not always make a speed change.

They will use it later for that. For now they can make .65 x2's(cooler ones) for cheaper..Great for OEM's....

............................................

Core 4 will kick core 3's ass :) 
August 19, 2006 12:37:34 AM

Quote:
What I don't understand about the AMD roadmap is they say that the first 65nm will not be any faster, and are given the same names(x2 4200,4600,etc).... Then later they can make 40% faster models(or something like that)....

How can you go smaller without getting faster? Make the pathways skinnier, but give them farther to travel?!?!


common miss conception, you think this way because ussally a die shrink also includes some new tech, i think intel has done it everytime... don't quote me on that. or atleast more recent 1s.
August 19, 2006 12:39:12 AM

65nm production was always scheduled to begin mid-2006 so this is nothing new and they are right on schedule. They are supposed to reach 50% production levels in Q1 2007, which is still a while away, but official product launch is December 2006. Although it is a possibility, I really don't see Dell launching on 65nm in September since there is no way they can have enough chips since Fab 36 is probably only at 10%-20% capacity right now with yields lower than 90nm as they tweak and mature the process.
August 19, 2006 1:25:33 AM

Further proof gman01 and allhell have no idea.
August 19, 2006 1:32:47 AM

Quote:
http://www.fabtech.org/content/view/1757/2/

I guess they want to keep it quiet so they can sell out of all the 90nm first....


According to my reliable source AMD is in mid transition to 65nm, and according to my reliable source AMD won't have production till mid 2007 most likely quarter 3 of 2007. But that’s just what my reliable source said.

Moo.
My Reliable Source Tells Me Your A Tard Too!
August 19, 2006 1:35:27 AM

Sounds like AMD is right on schedule, not early.

Quote:
According to the schedule, 65nm dual-core Brisbane is planned to release its Prototype in Aug, EVT (Enginerring Verification Testing)and early DVT(Design Verification Testing) version in Nov., and starting the production in Dec.


http://www.hkepc.com/bbs/itnews.php?tid=627532&starttime=0&endtime=0
August 19, 2006 1:43:17 AM

Yes you are right ,, I dont know what the heck you are talking about now?
August 19, 2006 1:45:04 AM

Your stupid uninformed posts.
August 19, 2006 1:47:30 AM

Because I said I wondered about the performance of the 65nm chips?
August 19, 2006 1:49:13 AM

Sigh, heres your dumbass post again.

Quote:
lol,,, some rabbid intel fan boy will soon jump all over this saying its not a reputable site & its pure propoganda,,, & they will have to see it on 4 more sites & hold on to one before they even think about the possibility of any truth to it.
August 19, 2006 1:51:56 AM

Quote:
Yes you are right ,, I dont know what the heck you are talking about now?


Yeah, I know what's your impression. He got his panties moisted when his "master" spreaded the FUD that "intels 90nm is better than AMD's". LOL :D 
If you look at the new Athlon X2 EEs, they consume less heat than intel's 65nm conrunts. :) 
August 19, 2006 1:53:28 AM

& it just burns you up that I put that there to shut you all up first lol..
let my lame post alone cry baby...
August 19, 2006 1:53:32 AM

Quote:
Yeah, I know what's your impression. He was the same one saying that "intels 90nm is better than AMD's". LOL Very Happy


When?

Quote:
If you look at the new Athlon X2 EEs, they consume less heat than intel's 65nm conrunts. Smile


Wow. Hand picked, undervolted and underclocked CPUs use less power, who'd of thought!
August 19, 2006 1:55:15 AM

Quote:
& it just burns you up that I put that there to shut you all up first lol..
let my lame post alone cry baby...


You making an ass of yourself doesn't burn me up.
August 19, 2006 1:57:30 AM

Oh well he usually lumps me into his stupid labelling and whatnot, hes still a dipsh!t either way.
August 19, 2006 1:59:36 AM

And where are these products anyways? I dont see them available anywhere.
August 19, 2006 2:01:08 AM

You are such a tired & lame little kitty,,, you should go to sleep before they put you to sleep.
August 19, 2006 2:02:36 AM

Are all your posts here going to be stupid?
August 19, 2006 2:04:18 AM

I'll do anything to pi$$ you off,,, hows that little kitty burning in hell?
August 19, 2006 2:07:52 AM

I guess that answers my question.
August 19, 2006 2:10:17 AM

Interesting article.

Completely off topic, jack you need an avatar so i can easily skip stupid posts and go straight to yours, something that easily stands out. :) 
August 19, 2006 2:11:58 AM

Quote:
Interesting article.

Completely off topic, jack you need an avatar so i can easily skip stupid posts and go straight to yours, something that easily stands out. :) 


Word.
August 19, 2006 2:13:21 AM

Since you like him so much why dont you give him a kitty also & you can play with each other... & you can add your other mate... as a matter of fact you all shall be refered to as one of the cat trio from now on.
August 19, 2006 2:15:21 AM

Quote:
Word.

brain dead.
August 19, 2006 2:16:54 AM

Quote:
Word.

brain dead.

Moo.
Whats Wrong Baby Need Your Bottle!
August 19, 2006 2:23:09 AM

OK, I'll bite...

Quote:
...Also, if you read up on scaling, a straight up scaling does not guarantee 'faster', it guarantee's lower power but NOT faster. In fact, if you apply 0.7 (the normal scaling factor) to all transistor parametrics, including field strength (i.e. voltage), without doing anything else fancy, theoretically a 65 nm transistor would perform equal to a 90 nm transistor.
Jack


If this scaling down guarantees lower power, then why is the Pressler 65 so much hotter at the same speed as a Northwood 90? Didn't the smaller scale create huge leakage issues and thus require more power?
August 19, 2006 2:24:29 AM

Northwood 90nm? Where? Hell, I'd want one. Sure beats my Prescott256.
August 19, 2006 2:40:26 AM

Quote:
http://www.fabtech.org/content/view/1757/2/

I guess they want to keep it quiet so they can sell out of all the 90nm first....


OMG! are you the Gman from You know "numa numa dance"?OMG ya videos ROX3D they cracked me up everytiem i saw new mix of your video
August 19, 2006 2:41:04 AM

OK, so maybe it's 120 vs. 90 instead. The issue is still the same... 8)
August 19, 2006 2:44:20 AM

Thank you.
August 19, 2006 3:23:00 AM

Quote:
Interesting article.

Completely off topic, jack you need an avatar so i can easily skip stupid posts and go straight to yours, something that easily stands out. :) 


I suggest a large pock-marked phallus.
August 19, 2006 3:24:11 AM

Right, interesting jack. So then we can expect the first new 65nm AMD's to be less power consumption..less heat..same basic work output. If we're lucky they might overclock better too right? I loved my 1800+, it zoomed.

On another note, if you change your icon to a kitty, I hope your thoughtful responses don't disappear and become mindless one line banter and fanboy ( i hate that word, it's so overused) bashing like action and spud. You always seem to share something relevant and informative which i commend :)  Others should follow your lead
August 19, 2006 4:31:15 AM

Ditto.
Action Man lately has really turned into a complete ass.
August 19, 2006 7:28:23 AM

Quote:
I'll do anything to pi$$ you off,,, hows that little kitty burning in hell?
Hey dude STHU! That NEDM cat realy was burned alive.
August 19, 2006 7:48:31 AM

Actually thats happy cat not the cat that was burned alive.
August 19, 2006 8:01:40 AM

Well I know that but I was refering to the cat that started NEDM. AKA NEDM cat.
August 19, 2006 8:05:15 AM

Quote:
http://www.fabtech.org/content/view/1757/2/

I guess they want to keep it quiet so they can sell out of all the 90nm first....


AMD still have 90nm facilities in Fab 30 and partner with Chartered Semi-Conductor for 90nm chips.

AMD's first commercial 65nm production is expected in Q4.
August 19, 2006 9:35:46 AM

Quote:
Interesting article.

Completely off topic, jack you need an avatar so i can easily skip stupid posts and go straight to yours, something that easily stands out. :) 


I suggest a large pock-marked phallus.

You mean like your yearbook picture?
August 19, 2006 10:13:21 AM

Hm, even if 65nm means no performance gains, it *should* mean lower power needs, correct?

This *could* mean, depending on the numbers, that AMD could take the performance/watt crown back, if not the overall perfoamance crown.

This would fit I think, as AMD seem to want to target the mobile segment, and alot of the ATi deal was about competing with platforms like Centrino, where power usage is very important. A lower thermal design also leads to a higher CPU density in servers, so could be an advantage there.

On the other hand people like me dont give a damn about performance/watt and just want performance regardless of cost, and will stick with Core 2 Duo weather or not AMD take the other crown back.
August 19, 2006 10:21:38 AM

Quote:

Second Edit: I went back and re-read my first post to Newf --- I wanted to add my opinion to what we know of Prescott's neburst design. Netburst was born in 2000 with Willamette and was a mild pipe lenthening from Northwood -- Presscot just went hog wild. When I look at the chip specs and the fact that the ALU's are doubleclocked, I cannot help but wonder if the first spin of the design just sucked so bad for performance and as such rather than FIX the design they thought --- heck, let's just double clock a few execution units until it gets faster --- if this is true (and it may or may not be) then it was a complete and total disregard for power --- and represents probably the worst technical decision in the history of Intel -- I would even go insofar as to say incompetent design talent, no wonder Core 2 uArch was handed off to the Israeli design team --- heck Banias and Dothan were darn good chips and made on the same process.


LOL, ancient history for some, recent history for others (moi). My main computer has a Willamette CPU at it's core, that's the one that got replaced by this current system underneath. I have to say, it was not THAT bad - at idle, you couldn't hear the fan, that's good right? *grins* Of course at max power I compared that computer to a jet engine. It's kinda like boy-racers doing up their car with bigger exhausts, I didn't need to do that with mine, the roar of those fans would have put to shame any competitor :lol:  Was this 130nm or 110nm? I can't be bothered booting it up to check at this present moment in time.

But you are completely right, Dothan was lovely. I still have a 1.5Ghz 2Mb level 2 cache Dothan in a laptop. It always made me laugh that a slower CPU from the same company performed 33%-50% better than the 1.6GHz P4. The shrink to 90nm wasn't really relevant, it was everything else that they did with the chip. Taking into account frquency differences, it is/was very competitive with my 3500+ laptop.

*general*

It will be good if AMD get 65nm off the ground sooner rather than later, so they can start getting the process right and yields up - they could possibly use this to make cheaper, higher level cache cpu's. Even though they are way behind in the performance war right now, low prices sell to the general market. I can't imagine they will have it at this stage though, by Q1 2007 at the earliest.

So what's up with these people like allhell, when they state they are just trying to wind up on purpose? Surely that's worthy of reporting and either warnings being issued or more? If there aren't enough moderators to keep these boards clean then I want to voluntear! :p 
August 19, 2006 10:32:47 AM

Quote:
Are all your posts here going to be stupid?
His sig is as close to common-sense as we're gonna see from him. He said he was 50 years old in a "forum members info" thread, but obviously that was a lie too. I think he's 15 not 50.
August 19, 2006 1:41:09 PM

Quote:
http://www.fabtech.org/content/view/1757/2/

I guess they want to keep it quiet so they can sell out of all the 90nm first....


According to my reliable source AMD is in mid transition to 65nm, and according to my reliable source AMD won't have production till mid 2007 most likely quarter 3 of 2007. But that’s just what my reliable source said.

Moo.
My Reliable Source Tells Me Your A Tard Too!

Sounds like you are talking about the k8l server chip schedule not the .65nm schedule
August 19, 2006 1:47:23 PM

Quote:
What I don't understand about the AMD roadmap is they say that the first 65nm will not be any faster, and are given the same names(x2 4200,4600,etc).... Then later they can make 40% faster models(or something like that)....

How can you go smaller without getting faster? Make the pathways skinnier, but give them farther to travel?!?!


It's because they cannot get the performance up. Right now looking at the numbers, they will get performance about 10-12 % higher than 90 nm on average, if 90 nm peaks at best yield at 2.4 GHz, then when mature the 65 nm will peak at 2.6 GHz.

Also, if you read up on scaling, a straight up scaling does not guarantee 'faster', it guarantee's lower power but NOT faster. In fact, if you apply 0.7 (the normal scaling factor) to all transistor parametrics, including field strength (i.e. voltage), without doing anything else fancy, theoretically a 65 nm transistor would perform equal to a 90 nm transistor.

Finally, to get field strength through the gate to be 'larger' than the 0.7 scaling factor, companies often use gate oxide thickness as the major knob. Gate ox thickness was made thinner going from 250 nm to 180 nm to 130 nm to 90 nm, but at 90 nm it stopped. Both AMD and Intel run 90 nm gate oxide thickness at 1.2 nm, it physically cannot be made any thinner.

Hence, AMD is trying to 'stress' engineer, they have not done a very good job at it, as the data shows. This is not their fault as when you get smaller the impact of stress is much less. So, if you read their list of 'enablers' for the 65 nm node it is quite long --- 4 stressors and still cannot get performance above 2.4 GHz at launch.... (this is assuming of course that roadmap is accurate, between last Dec and final launch they may have found a magic knob).

Edit, and if this rumor is true, they are not quite 5 months early, a small release to one supplier is not a launch and second, they have stated Dec 2006 for many months now as limited release of 65 nm parts, sept. is 3 (three) months early and 1 year behind Intel.

Jack

Thank you for that answer....
August 19, 2006 1:51:08 PM

Quote:
Further proof gman01 and allhell have no idea.


Sorry I had to ask a question, and admit I don't know everything about the cpu scaling....

If any of us knew everything - We wouldn't be here wasting our time educating ourselves on this forum....
!