65nm Socket939 A64 CPUs

[dinos22]

Are they going to happen or not?

or is AMD going to leave them at 90nm and to die of slow death

[MrRevhead]

Doubt it dude, they are projecting EOL for 939 within the next 6-9 months, so dont see why they would suddenly convert to use 65nm dies when phasing a socket type out.

[ozzimark]

Doubt it dude, they are projecting EOL for 939 within the next 6-9 months, so dont see why they would suddenly convert to use 65nm dies when phasing a socket type out.
replace current production with 65nm production.. saves the trouble of running two process sizes at once, and when ramped up, 65mn should be cheaper to produce.. smaller dies, so more chips per wafer ;)

[MrRevhead]

replace current production with 65nm production.. saves the trouble of running two process sizes at once, and when ramped up, 65mn should be cheaper to produce.. smaller dies, so more chips per wafer ;)That logic does make some sense, but when AMD arent planning on implementing 65nm technology on their CPU's for at least another 6 months, why would they make such a move for something that will be just about taken off the market? It more likely we will see what you suggested in the Skt 754 line-up which is planned to have a slightly longer life (EOL = Q3/4 2007).
Wouldnt AMD want to clear stock rather than continually replenish it, increasing their chances of writing off excess stock? Looks like 90nm will still be around till about 2nd/3rd quater 2007, produced in both 200mm & 300mm wafer form, so its no great loss to AMD.

In the near term, AMD is transitioning from 200mm wafers for its 90nm SOI process to 300mm wafers. This will help serve to get AMD up to speed with 300mm wafer before volume production of 65nm devices begins. As 200mm wafer production is phased out, 65nm processes will be ramping into full swing.....

In moving to 65nm, AMD has adopted a strategy of constant improvement of its current transistor technology to ensure a smooth transition. The last generation 90nm parts will overlap the first generation 65nm parts, as many of the key features of their structures will be shared.Original Source (http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2768)

http://img228.imageshack.us/img228/5044/65nmroadmap6pq.th.png (http://img228.imageshack.us/my.php?image=65nmroadmap6pq.png) http://img115.imageshack.us/img115/4711/fab385ql.th.png (http://img115.imageshack.us/my.php?image=fab385ql.png)

[dinos22]

Doubt it dude, they are projecting EOL for 939 within the next 6-9 months, so dont see why they would suddenly convert to use 65nm dies when phasing a socket type out.
if they are phasing it out than it looks terminal hey......they'll probably use the left over capacity to product the last of the S939 stuff before they convert the fab to 65nm process

that was shorter than expected for S939 :rolleyes:

[MrRevhead]

that was shorter than expected for S939 :rolleyes:Yeah, same man. I was hoping it would of hanged around for longer with 754 being given higher preference to be phased out first, but obviously not according to DailyTech (http://dailytech.com/article.aspx?newsid=1554).

[Magnj]

:( Don't die s939. I'm not ready to give up my rig

[Raybo]

:( Don't die s939. I'm not ready to give up my rig

I love suprises,don't give up hope yet.

:toast:

[RockinHack]

S939&S754 will die in 2007 Q1,when Socket F released...

[savantu]

The crossover between 90nm and 65nm will happen for AMD in Q3 2007.

[largon]

RockinHack,
Why would Socket F (which is a server socket) have any effect on S754/S939?
btw. Socket F should be out 1st of August 06 (http://www.xtremesystems.org/forums/showthread.php?t=103422).

[RockinHack]

RockinHack,
Why would Socket F (which is a server socket) have any effect on S754/S939?
btw. Socket F should be out 1st of August 06 (http://www.xtremesystems.org/forums/showthread.php?t=103422).
Socket F isnt a server socket.I meant about socket 1207,with integrated pci-e controllers...

Because,amd has an announce 2 months ago....

They said about,S939 semprons did not choice of users,so they contiuned with S754 Semprons,because S754 boards are very cheap,and one more thing that,available turion processed notebooks have S754..So S754 and S939 will die same day(offical)....

S939 systems,I dont think about they will release 65 Nm Athlon 64 processors..Because,they must interested in "65 Nm quad-cores"...

Amd's product capability,cannot enough for this,remember that,when we hear about S939 cpu's,amd stopped to produce S754 130 Nm athlon 64 cpu's.

Maybe this time,when K8L launched,amd will stop again S939 90 Nm cpu's,and release S939 Semprons cheaper than A64's...

Although,amd didnt produce different production technology cpu's(For example,180 Nm Cpu's and 130 Nm's didnt produced together,180 nm term closed,130 nm term opened)etc...

[largon]

Socket F isnt a server socket.I meant about socket 1207,with integrated pci-e controllers...Errr...
Socket F = LGA1207 = Opteron 2000-series DDR2 socket

And, these procs don't feature integrated PCIe controllers, they'r just Rev.E Opterons with DDR2 support and a new socket. The platform will later support QC and possibly DDR3 once Rev.H (K8L) is available.
65 Nm -- 130 Nm -- 90 NmNewton meter cpus? :p:

[\Karting_freak]

Newton meter cpus? :p:
nano meter

[arisythila]

I still see new Socket A CPU's. I doubt they will stop all production of the S939 for years to come. They will convert to 65nm aswell. IT doesnt make sense to make 90nm S939's and 65nm AM2's. They will convert to 65nm S939, only logical way to do business.. Cheaper for them to produce aswell.

~Mike

[arisythila]

nano meter

N + m, since its a capitol N its Newton, and lower case m is meter.

Nm is Newton Meter

nm is nanometer.

~Mike

[FghtinIrshNvrDi]

Newton Meter's???

Automotive Junkie #1: "Dude, my dual core has 2 cores at 90 newton meters a piece! That'll totally spank yours. I have a 180 Newton meters of torque.."

Automotive Junkie #2: "Man, you're right. I got jewed. I only have a pair of 65 newton meter cores... 130 newton meters of torque just couldn't cut it."

Ryan

[K.I.T.T.]

Newton metre = Nm unit of force

nm = nano metre = 1 x (10^-9) metres unit of length

:fact:

[Jaco]

in that case , i want a dual core 130Nm cpu :D

[albumleaf]

I still see new Socket A CPU's. I doubt they will stop all production of the S939 for years to come. They will convert to 65nm aswell. IT doesnt make sense to make 90nm S939's and 65nm AM2's. They will convert to 65nm S939, only logical way to do business.. Cheaper for them to produce aswell.

~Mike

Um, socket A production stopped long ago, what you're seeing is new old stock processors. And you're retarded, why would it be any cheaper to even bother to move to 65nm for 939 when they're dumping the socket anyway.

~LOOK AT ME I'M SIGNING MY POST.

[celemine1Gig]

Um, socket A production stopped long ago, what you're seeing is new old stock processors. And you're retarded, why would it be any cheaper to even bother to move to 65nm for 939 when they're dumping the socket anyway.

~LOOK AT ME I'M SIGNING MY POST.

May I ask you who the hell rained on your parade? A bit more politeness and respect couldn't harm. :nono:

BTW, I do believe that we'll see some 65nm S939 CPUs sooner or later. Maybe rather later, but I doubt that they will keep producing 90nm when they have a chance to switch to 65nm, in order to be more cost effective. But it'll probably take some time to get to the point.

[perkam]

For all those as clueless as the thread starter:

http://www.xtremesystems.org/forums/showthread.php?t=102628

65NM AM2 CPUs with Enhanced clocks are called "Rev G" CPUs coming out q4 2006. The current AM2 processors are "Rev F" processors.

They will be AM2 ONLY. Rev G processors will be the first significant change in A64 core since K8's original introduction. The second significant change is the K8L, a brand new architecture coming late 2007 to have some kentsfield for breakfast with a side order of conroe.

Perkam

[savantu]

...The second significant change is the K8L, 65nm Quad Core pwnage coming late 2007 to have some kentsfield for breakfast with a side order of conroe.

Perkam

Aaaa..the 1st QC AMD for desktops will appear in H1 2008 , the Greyhound core.

By that time Kentsfield is replaced by Bloomfield , 45nm , QC , shared L2 8/16MB ...I'm willing to bet that Bloomfield will have Greyhound for breakfast. :stick:

[zabomb4163]

The second significant change is the K8L, 65nm Quad Core pwnage coming late 2007 to have some kentsfield for breakfast with a side order of conroe.

Perkam

keep in mind kentsfield will be 45nm at that point. (hint: 4ghz on air)

[perkam]

Aaaa..the 1st QC AMD for desktops will appear in H1 2008 , the Greyhound core.

By that time Kentsfield is replaced by Bloomfield , 45nm , QC , shared L2 8/16MB ...I'm willing to bet that Bloomfield will have Greyhound for breakfast. :stick:Really. K post edited.

Then what how exactly is K8L different from Rev G ??

Perkam

[savantu]

Really. K post edited.

Then what how exactly is K8L different from Rev G ??

Perkam

Rev G is 65nm DC with an extra FPU unit.

K8L build on that with double the width of the floating point execution units, and increases the data width feeding the FPU's by a factor of 2x.

Add to that 4 cores , 512KB L2/core and shared 2 MB L3.Other features in the picture ( some might also be present in Rev G altough I doubt it ) :

http://www.pureoverclock.com/images/review/k8l_closeup.gif

[largon]

Aaaa..the 1st QC AMD for desktops will appear in H1 2008 , the Greyhound core.Desktop K8L QC will be here sooner than 08. See the attached slide from AMD Analyst Day 2006, from 2 weeks ago.

-> desktop QC K8L available @ 3Q07 -> whole new µ-arch in 1H08

The new arch should be quite something... AMD estimates QC+65nm will bring +60% more perf/watt - the new arch perform 150% better. :eek:
Rev G is 65nm DC with an extra FPU unit.AMD has said Rev.G is an optical shrink with practically identical logic compared to Rev.F.
No extra FPU (let alone L3, OoO, etc.) for Rev.G...
:(

[savantu]

Desktop K8L QC will be here sooner than 08. See the attached slide from AMD Analyst Day 2006, from 2 weeks ago.

-> desktop QC K8L available @ 3Q07 -> whole new µ-arch in 1H08

Desktop as in workstations as for the the new microarch , K8L is the new microarch , the advances you see there are due to smaller process.


The new arch should be quite something... AMD estimates QC+65nm will bring +60% more perf/watt - the new arch perform 150% better. :eek:AMD has said Rev.G is an optical shrink with practically identical logic compared to Rev.F.
No extra FPU (let alone L3, OoO, etc.) for Rev.G...
:(

The 150% is from comparing 45nm with todays 90nm.

I also thought Rev G is an optical shrink but more and more I keep seeing an extra FPU.Where did AMD say Rev G is only a shrink ?

[toddm27]

I don't think 65 nm is in the plans for 939, just am2

[dinos22]

i think we've established now that 65nm will not happen for S939.........>>> THREAD closed :)








For all those as clueless as the thread starter:

http://www.xtremesystems.org/forums/showthread.php?t=102628

65NM AM2 CPUs with Enhanced clocks are called "Rev G" CPUs coming out q4 2006. The current AM2 processors are "Rev F" processors.

They will be AM2 ONLY. Rev G processors will be the first significant change in A64 core since K8's original introduction. The second significant change is the K8L, a brand new architecture coming late 2007 to have some kentsfield for breakfast with a side order of conroe.

Perkam
lol at the clueless comment :p: :)

[perkam]

i think we've established now that 65nm will not happen for S939.........>>> THREAD closed :)








lol at the clueless comment :p: :)Well the thread is for discussion of these types :)

The clueless comment was not so much for not knowing but for starting a new thread when I've linked that thread a myriad of times.

Perkam

[largon]

K8L is the new microarchHector Ruiz doesn't agree on that:

As [Intel] closes the technology gap, it will be a much tighter race, but we're going to introduce a really new architecture that will work well with our partners for the best performance. We're going to start sampling it at the end of 2007 and roll it out in 2008. (http://www.businessweek.com/technology/content/jun2006/tc20060605_019097.htm?chan=topStories_ssi_5)
Where did AMD say Rev G is only a shrink ?Dirk Meyer said it @
http://babelfish.altavista.com/babelfish/urltrurl?lp=ja_en&trurl=http%3a%2f%2fpc.watch.impress.co.jp%2fdocs%2 f2006%2f0531%2fkaigai273.htm

[savantu]

Hector Ruiz doesn't agree on that:


That is K8L and a new microarch for notebooks is expected in 2008. ( could it be that one ? )

[slavearm]

I have seens tons of links pointing to Rev. G being a die shrink only. Does anyone have a "reputable" link to show otherwise?

[Serge84]

Its really not that hard to make Rev E 65nm 754 and 939 chips. Actouly since the 65nm process is cheaper since AMD is aready half way into it. Doing 2 processes would be more expencive. They will support making socket 754 and 939 cpus until the end of 2007. We thought socket 754 was dead. Well its not even got a new rev E and 90 nm tech. Same things going to happen again with 65nm's and the 754's as well as the 939's. But its only a smaller process and something like a new steaping like Rev E8 or something like that. Will not make much of a difference. Just make cooler and cheaper cpus. Well will get a advantage in OCing. But 754's and 939's are long from dead. Don't listen to the press. If we listened to them we wouldn't have beleaved a E6 venice existed for 754's on 90nm processes. =P AMD said socket 754's and 939's will be supported until 2007 so until they say otherwise we will see new chips just like those new E6 754's it was hard to beleve when they 1st came out with.

PS: Here is the real K8L Die. http://www.theinquirer.net/?article=32424

[slavearm]

That image is pretty crappy. But it is not like anyone has any proof that it is not real.

[RockinHack]

Hey guys I think,Amd wont announce S939 65 Nm Cpu's,because,when 65 Nm process starts,(In am2),they cant produce 65&90 Nm together,because S939 users wont buy any Am2 cpu if they will do it...

S754 and S939 will die in same day...Amd announced that,S939 Semprons will not produce,so they are continuing with S754 Semprons,S754 Athlons died 1 years ago...

S939 Killer will be 65 Nm Am2 cpu's I think...

And than,X2-Quad Core Cpu's will release after 8 months..It will be nice,but I dont think they can beat Kensfield...

In 2007 Q2 or Q3,we will see S1207 K8L quad-cored cpu's...It can beat Intel because of using Ddr 3 mem sticks...