AMD 65 nm next month

[GenTarkin]

As far as hitting higher clock speed, I hope they can too...but remember AMD is currently having difficulties with voltages on their 65nm...according to the few articles I read on inquirer. Then AMD said they had those problems minimized but not fixed as they stated they would probably have to release the 65nm .05v higher than planned. If thats true doesnt that mean initially possibly hardly no OC headroom?

[alayashu]

but remember AMD is currently having difficulties with voltages on their 65nm...according to the few articles I read on inquirer. Then AMD said they had those problems minimized but not fixed as they stated they would probably have to release the 65nm .05v higher than planned.

not everything that flies is comestible.

[VVJ]

Not long ago one from my Chinese beta testers sent me this snapshot

[Concorde Rules]

God I hope its not faster LOL just bought a E6600! :p

I doubt it will beat Intel, but it will close the gap, which is good I suppose.

[XeRo]

What would a normal vCore for a 65nm processor be?

[FghtinIrshNvrDi]

probably around 1.25 just like the picture posted...

Ryan

[Mikael]

Hmmm... That pic indicates a doubling of the L1 instruction/data cache. It would probably be a pretty cheap/easy way to increase performance slightly, but I don't really believe it yet.

How reliable is the source of this pic?

[biohead]

hm I thought it wouold be more like 1.15v. But I have the feelig that even at that vCore it can run above rated speeds.

[mesyn191]

Looks fake.

256KB L1$? Hyperthreading? That would require a complete arch. overhaul to add those features and its unclear that it'd do anything to improve performace as L1$ latency would go up significantly. Its unclear wether hyperthreading would do much to improve performance on the K8 too.

[informal]

Guys...L1 is 256KB as 128*2(two cores).L2 is 2*512KB...Duh

On the other hand,who says it's Hyperthreading?Maybe it's something else.
And 1.25Vcore is very well in line for dual core CPU.And that is an Eng Sample ,not retail CPU.
So,it seems AMD is pretty much on track with 65nm.

[mesyn191]

Guys...L1 is 256KB as 128*2(two cores).L2 is 2*512KB...Duh

Very wierd, it shouldn't measure them this way since they're completely seperate even if they are on the same die...

On the other hand,who says it's Hyperthreading?Maybe it's something else.
And 1.25Vcore is very well in line for dual core CPU.And that is an Eng Sample ,not retail CPU.
So,it seems AMD is pretty much on track with 65nm.

The voltage is certainly possible, but hyperthreading? It'd take a complete redesign on the same order of difficulty and complexity as it took to make K8L to add that feature, I doubt they have the time or resources to do that. In fact it wouldn't be a K8 anymore, it'd be a K8.5 or K8++ or something.

[The Ghost]

hm I thought it wouold be more like 1.15v. But I have the feelig that even at that vCore it can run above rated speeds.

i think that there is a limit to how low they can go on the voltage , it is said that it takes 1.1v minimum for the bandgap using the material that they are using

[dinos22]

lol at people hung up on hyperthreading..........RELAX it could be a software glitch.......we've seen it many times before (different software showing odd results that is)

[ozzimark]

lol at people hung up on hyperthreading..........RELAX it could be a software glitch.......we've seen it many times before (different software showing odd results that is)

yes, hyperthreading most likely will no be happening.

[mesyn191]

lol at people hung up on hyperthreading..........RELAX it could be a software glitch.......we've seen it many times before (different software showing odd results that is)

OK didn't know that, thanks.

[syne_24]

God I hope its not faster LOL just bought a E6600! :p

I doubt it will beat Intel, but it will close the gap, which is good I suppose.

lol I hope it is faster just so intel can have some pressure to make their next chip. Dont want either one of them slacking off. Of course I'll be crawling back if indeed amd 65nm regains the lead.

[Serge84]

http://129.15.202.185/athlon_rev_g/6...hancements.jpg


That looks like the K8L picture. I believe the 65nm chips are just shrinks until K8L arrives.

http://upload.wikimedia.org/wikipedi...erhound_QC.png

This is K8L.

On other thoughts...

Did we forget 939's get over 3ghz all the time. Even 3.5ghz is almost normal for the extreme ppl anyways. The windsors are limited for some reason. The older CPU's that have all the materials in them they use to have are actouly better then the cleaned up cores with no lead in them. Could this be the reason why they are such bad OCers?

I would like a FX-60 on a old 939 DFI board really if this is how its going to be until K8L. But there is a possiblity of 65nm being that good like it was from 130nm to 90nm. We should see atleast a 600mhz difference in OCing like they said. If when the end comes when they perfect stepings. We may see cpus even better then rev E. Nothing can beat a rev E in AMD's weapons arsonal. FX-60 highes clock 3.9ghz. FX-57 highest clock almost 4.2ghz. Off by about 15mhz.

That is what legends are made of!

[metro.cl]

both this cpus you show are old by now there is a new revision coming fue tu a bug/problem in this revision.

[kaBuZi]

http://www.xbitlabs.com/news/cpu/dis...911064959.html

no goodies for socket 939.

[VVJ]

Mikael

That pic indicates a doubling of the L1 instruction/data cache.

Because that CPU is dual-core: L1-I: 2x64KB=128KB, L1-D: 2x64KB=128KB.

mesyn191

Looks fake.

LOL! Kid, I do not post fakes :-) I have told this many times at XS.

256KB L1$?

Yes, because CBId reports the full L1 and L2 cache size like that. Take a look at the following snapshots:
http://cbid.amdclub.ru/html/a64x24400E6.html
http://cbid.amdclub.ru/html/a64x23800E4.html

[informal]

Yeah I've told them already ,but they won't listen .
It's full(summed up) value of cache for both cores.

It takes THE MAN himself to clarify the obvious thing.Thx VVJ

[breakfromyou]

Very wierd, it shouldn't measure them this way since they're completely seperate even if they are on the same die...

The voltage is certainly possible, but hyperthreading? It'd take a complete redesign on the same order of difficulty and complexity as it took to make K8L to add that feature, I doubt they have the time or resources to do that. In fact it wouldn't be a K8 anymore, it'd be a K8.5 or K8++ or something.

K8L = K8.50...and of course the hyperthreading part is a software glitch. i hope the best for AMD. does anybody know if AMD is going to use SiGe with the 65nm processors or K8L?

[Turtle 1]

Sorry no AMD65 next month. be luck if it gets here this year.

http://www.theinquirer.net/default.aspx?article=34282


So it seems AMD is not on track ,

[vitaminc]

K8L = K8.50...and of course the hyperthreading part is a software glitch. i hope the best for AMD. does anybody know if AMD is going to use SiGe with the 65nm processors or K8L?

I doubt they will. For some reason the AMD fans wants AMD to pull every new tricks in their next product release, in this case 65nm straight die shrink of AM2 K8.

Here's a tried and true combination in the Semiconductor world:
New design + new process + new material == failure

AMD won't be an exception and they will also fail miserably if they choose to do so.

[The Ghost]

Sorry no AMD65 next month. be luck if it gets here this year.

http://www.theinquirer.net/default.aspx?article=34282


So it seems AMD is not on track ,

sorry , you are wrong again

http://www.tbreak.com/reviews/article.php?id=467

We also spoke about the move to .65nm manufacturing process with the Opteron now moved to it and the Athlon64 line-up following in the next few months.