AMD 65nm now available

[Viditor]

Inq Article

"THE WAIT FOR AMD 65 nanometre CPUs is over. At least, this is what on-line stores indicate today"
The CPUs are marked Revision F2, and the maximum TDP varies between 65W and 89W - in the real world, 55-62W and 72-81W is expected - depending on the number of cores, of course. The advertised features are Cool'n'Quiet, NX-Flag, SSE3 and Pacifica.

There are also 65nm EEs available, albeit in far lower volumes than their non-economic counter-parts.

We're expecting the first overclocking reports on these CPUs next week, so if you're unsure how much headroom the first generation 65nm has, it won't hurt to wait a bit longer"

 

[myocardia]

I don't see what good it's going to do them, besides the obvious lower production costs, unless they'll include all of the core enhancements that you've been saying they'll include, which they don't seem to.

 

[Viditor]

Originally posted by: myocardia
I don't see what good it's going to do them, besides the obvious lower production costs, unless they'll include all of the core enhancements that you've been saying they'll include, which they don't seem to.

Nope, the core enhancements (K8L or Rev B) are in Q2...
However, it's quite possible that the lower end X2s will be able to overclock near C2D levels (making them a MUCH more attractive but for many here).
However, the most important part is that they will be cheaper to make. That's why I have been chuckling to myself when people tell me how expensive the FX chips for 4x4 will be. AMD has stated all along that they will be less than $1000 for 2 of them (on par with a single Kentsfield). The only way to do that cost effectively is with 65nm...

 

[The-Noid]

http://valid.x86-secret.com/show_oc?id=137503

The problem again with Kentsfield is it is as easy as stink to overclock and keep stable. Same as C2D. AMD will not be attractive until K8L comes out. Cost, Heat, Speed. If you charge the most for a premium product it has to perform the best. The 4x4 with <$1000 still need a $300-$400 motherboard and 4 video cards. A Kentsfield Xeon 3220 (2x E6400, $513 in 1000) is going to beat all of the 4x4 in raw processessing, let alone the $1,000 version.

This is without overclocking, that on the 4x4 is going to require you to turn on a Nuclear plant by your house. At the lowest these chips are going to be 2x 90w TDP at stock. thats 60-70w more than a Q6400.

 

[nyker96]

I wonder if these things can be overclocked much higher than X2's 3Ghz wall.

 

[harpoon84]

Originally posted by: Viditor

Originally posted by: myocardia
I don't see what good it's going to do them, besides the obvious lower production costs, unless they'll include all of the core enhancements that you've been saying they'll include, which they don't seem to.

Nope, the core enhancements (K8L or Rev B) are in Q2...
However, it's quite possible that the lower end X2s will be able to overclock near C2D levels (making them a MUCH more attractive but for many here).
However, the most important part is that they will be cheaper to make. That's why I have been chuckling to myself when people tell me how expensive the FX chips for 4x4 will be. AMD has stated all along that they will be less than $1000 for 2 of them (on par with a single Kentsfield). The only way to do that cost effectively is with 65nm...

'Near C2D levels'? A 3GHz C2D = 3.6GHz on X2. A 3.5GHz C2D = 4.2GHz X2. And that's with 2MB L2 Allendales. A 3.5GHz Conroe would be equivalent to a 4.4GHz X2... good luck getting that on air anytime soon.

Yes, if somehow the 65nm X2s can overclock to 3.5 - 4GHz it will mean they can compete with overclocked C2Ds. Somehow, I'm not holding my breath. AMD can't even clock 65nm above 90nm speeds after all. If they could, the FX-7x @ 2.6 - 3GHz would not be on 90nm and chewing up to 125W per CPU.

65nm K8 will make AMD more money by cutting costs, but it'll hardly dent the C2D stranglehold on the performance and enthusiast market. It'll have about as much effect as Intel's 65nm Netburst effort against AMD X2... not much at all.

 

[Viditor]

Originally posted by: harpoon84

Originally posted by: Viditor

Originally posted by: myocardia
I don't see what good it's going to do them, besides the obvious lower production costs, unless they'll include all of the core enhancements that you've been saying they'll include, which they don't seem to.

Nope, the core enhancements (K8L or Rev B) are in Q2...
However, it's quite possible that the lower end X2s will be able to overclock near C2D levels (making them a MUCH more attractive but for many here).
However, the most important part is that they will be cheaper to make. That's why I have been chuckling to myself when people tell me how expensive the FX chips for 4x4 will be. AMD has stated all along that they will be less than $1000 for 2 of them (on par with a single Kentsfield). The only way to do that cost effectively is with 65nm...

'Near C2D levels'? A 3GHz C2D = 3.6GHz on X2. A 3.5GHz C2D = 4.2GHz X2. And that's with 2MB L2 Allendales. A 3.5GHz Conroe would be equivalent to a 4.4GHz X2... good luck getting that on air anytime soon.

I was talking about clockspeed of course...an X2 4200+ clocked to 3.5 for example would be a very nice chip, and would be easily competitive with the 6300/6400 C2Ds. And that's certainly not out of the realm of possibilities! I think your scale of comparison is heavily skewed...3 GHz Allendale is not going to keep up with a 3.6 GHz X2.

Yes, if somehow the 65nm X2s can overclock to 3.5 - 4GHz it will mean they can compete with overclocked C2Ds. Somehow, I'm not holding my breath. AMD can't even clock 65nm above 90nm speeds after all. If they could, the FX-7x @ 2.6 - 3GHz would not be on 90nm and chewing up to 125W per CPU.

Why would you say that they can't clock 65nm faster than 90nm? That really makes no sense (typo?). If you are referring to the fact that they aren't releasing 65nm at a faster clock than 90nm initially, this is bog standard practise for both Intel and AMD...I can't think of an exception to this.

65nm K8 will make AMD more money by cutting costs, but it'll hardly dent the C2D stranglehold on the performance and enthusiast market. It'll have about as much effect as Intel's 65nm Netburst effort against AMD X2... not much at all.

I agree that the C2D core will still remain much better at equal overclocks, but you're wrong about it not having an effect. The OC market looks for the best bang for the buck, not just the fastest chip...
I personally don't have a clue how well it's going to perform on OC, but as AMD has already released a 3 GHz on 90nm, a 33% OC after a node change to hit 4GHz seems very possible.

My point was that you should stop making up your mind before we actually see the thing...

 

[Lonyo]

A 3800+ is cheaper than an E6300, so if the 65nm ones do overclock well, then for the very lowest end dual core CPU's, an X2 may become a more attractive proposition than it currently is.

 

[Cheex]

I was always told to never believe The Inquirer.....

 

[Bateluer]

So when are we going to see these 65nm AMD chips in retail?

 

[coldpower27]

Originally posted by: Viditor

Originally posted by: myocardia
I don't see what good it's going to do them, besides the obvious lower production costs, unless they'll include all of the core enhancements that you've been saying they'll include, which they don't seem to.

Nope, the core enhancements (K8L or Rev B) are in Q2...
However, it's quite possible that the lower end X2s will be able to overclock near C2D levels (making them a MUCH more attractive but for many here).
However, the most important part is that they will be cheaper to make. That's why I have been chuckling to myself when people tell me how expensive the FX chips for 4x4 will be. AMD has stated all along that they will be less than $1000 for 2 of them (on par with a single Kentsfield). The only way to do that cost effectively is with 65nm...

Considering that the 65nm Brisbane are only going to be used for the low end to mid range for the moment how is that gonna help the FX's?

$1000 USD is what I have been hearing, for 2 Athlon FX 70's, and 1500USD for the top end FX-74's though we will have to see what the final pricing will beo nce they arrive, since all we have is hearsay for now.

The FX's will be on the 90nm node when they launch. as AMD can't reach high clockspeeds with 65nm yet and I am not aware of any 2x1MB model for the 65nm node. As it looks like AMD is adopting 0.5x multipliers again.

It's simply because the yields are low in those clock ranges that they are charging so high, not really because it's epxensive to make, AMD has been making 183mm2 die's for the mid range at 152USD or so for the past while. So charging 1000 USD or more for a pair of 230mm2 dies is not a problem for the high end.

 

[Accord99]

If you check the store links, it looks more like a mistake since it still references them as 90nm and Rev. F

 

[Cheex]

Originally posted by: Accord99
If you check the store links, it looks more like a mistake since it still references them as 90nm and Rev. F

See....

Originally posted by: Cheex
I was always told to never believe The Inquirer.....

There you have it.

 

[harpoon84]

Originally posted by: Viditor
I was talking about clockspeed of course...an X2 4200+ clocked to 3.5 for example would be a very nice chip, and would be easily competitive with the 6300/6400 C2Ds. And that's certainly not out of the realm of possibilities! I think your scale of comparison is heavily skewed...3 GHz Allendale is not going to keep up with a 3.6 GHz X2.

Then I'm afraid you have unrealistic expectations. A 3GHz Allendale will easily keep pace with a 3.6GHz X2. Seeing how a 2.94GHz E6300 outpaces a 3GHz X2 by an average of 20% <a target=_blank class=ftalternatingbar...es/cpu/display/core2duo-e6300.html</a>">Xbitlabs comparison</a> I don't see how you came to your conclusion. Allendale is at least 20% faster clock for clock against X2, and Conroe is around 25% faster clock for clock.

Why would you say that they can't clock 65nm faster than 90nm? That really makes no sense (typo?). If you are referring to the fact that they aren't releasing 65nm at a faster clock than 90nm initially, this is bog standard practise for both Intel and AMD...I can't think of an exception to this.

I meant that people shouldn't expect miracles from a die shrink, especially at launch. The primary benefit from a die shrink is to cut manufacturing costs and reduce power consumption

Look how much 90nm helped compared to 130nm for K8. Correct me if I am wrong, but 130nm K8s overclocked to around 2.4GHz on average. 90nm K8s are overclocking to around 2.8GHz on average. That's about a 400MHz boost from a die shrink, so unless AMD worked some miracles I don't expect to see 3.5 - 4GHz overclocks from a 65nm X2 anytime soon. Of course, I'd love to be proven wrong, increased competition between AMD and Intel can only benefit us as consumers, but realistically that's not going to happen until K8L.

I agree that the C2D core will still remain much better at equal overclocks, but you're wrong about it not having an effect. The OC market looks for the best bang for the buck, not just the fastest chip...
I personally don't have a clue how well it's going to perform on OC, but as AMD has already released a 3 GHz on 90nm, a 33% OC after a node change to hit 4GHz seems very possible.

My point was that you should stop making up your mind before we actually see the thing...

The only 'effect' 65nm is likely to have is on AMDs bank balance, which is a good thing of course. A healthy AMD = healthy competition.

The OC market is after best value of course, but I don't see what anything AMD has to offer that can beat a $150 E4300 in that respect. The lower FSB reduces the need for a costly mobo and high speed RAM, thus reducing platform costs to comparable levels to AMD's offerings.

I haven't 'made up my mind' regarding 65nm K8, I just don't share the same rosy outlook you have for them.

I do hope they will clock up to 4GHz as you mentioned. That would be friggin awesome and open the enthusiast market right up. I just don't think that's very likely, at least not in the short term.

BTW may I mention that the link TheINQ provided shows a 90nm X2 4200+, not a 65nm... go figure.

 

[CKXP]

it's the inquirer :disgust: and looks to be a typo on the e-tailer's part....AFAIK there is no 4200+ "Brisbane" only 4000,4400,4800, and 5000

 

[harpoon84]

The page has been removed from TheINQ index, although the link is still active. Obviously Theo stuffed up and is now trying to cover his tracks. :roll:

Looking closer at that German E-tailer, it clearly states Rev F which is 90nm...

TheINQ - king of reporting BS.

 

[Kur]

Originally posted by: Cheex
I was always told to never believe The Enquirer.....

Fixed

 

[Brunnis]

65nm is hardly going to provide near 4GHz overclocks without significant effort. The move from 130nm to 90nm gave us about 15% increased frequency headroom and there's really not much reason to suspect that initial 65nm parts will offer much more. K8 was probably not designed with very high clock frequency in mind and a redesign of the core (K8L?) is probably needed to get to the levels that some people are hoping for.

Don't get me wrong, I have an AM2 system that could do with a nice A64 X2, so I'd love for 65nm K8 to be a real killer. Just trying to keep my expectations realistic.

 

[nyker96]

I would say if the new 65nm X2 3800+ can break 3 Ghz, say 3.5, I think it can be quite a price/perf winner. It will be equal to E6300 in many respect and since it's 30 bucks cheaper when they get the supplies issue straightened out then it's a pretty decent buy for the money. But then again if intel comes out with 150 dollar E4300s, this could once again tip the scale toward Intel on the low end.

 

[A554SS1N]

Originally posted by: nyker96
I would say if the new 65nm X2 3800+ can break 3 Ghz, say 3.5, I think it can be quite a price/perf winner. It will be equal to E6300 in many respect and since it's 30 bucks cheaper when they get the supplies issue straightened out then it's a pretty decent buy for the money. But then again if intel comes out with 150 dollar E4300s, this could once again tip the scale toward Intel on the low end.

If AMD's chips could go that high, I doubt they'd be calling them X2 3800+ as they'd be bound to want to make more profit on it by selling it as a higher model.