[Kuroimaho]

Xbitlabs was the first to jump the gun on these new cpus. Link

Athlon II X2 250 overclocked to 3.9GHz with the Vcore increased by 0.175V above the nominal and remained absolutely stable throughout the tests.

The 550BE processor remained stable at 3.98GHz frequency with the core voltage increased by 0.15V.



Firingsquad.com

Tomshardware

Techspot

icrontic

Legionhardware

[Bima Sylirian]

Great! 550 is also black edition!
Gotta forget E5200 and get with one of these.
Hope the pricing will be great (just as usual)!

[Kuroimaho]

I also worried they won't make it a BE as it's "only" a dual, so a pleasant surprise.

[Bima Sylirian]

Well yeah... I've never heard 550 will be a black edition until I read this article.
Now it's just how they compared to E7000 & E8000 series in gaming.
Callisto might not as good as high-end Wolfdale in application because AMD strikes that one with Heka. But at least Callisto should be able to compete in gaming. And I also hope there will be no issue with backward compatibility.
They should be able in the market in June according to the latest news I've read.
Damn AMD for keep delaying their product!

[nzaneb]

I like that they utilized the L2 from the disabled cores for Regor. It really helps it to be a contender, even without the added benefit of L3. I can't wait for these to hit the shelves Nice link Kuro.

[cdawall]

NB clocks suck wonder if thats the chip or they just left them alone to oc?

[Archer0915]

When will they be on the egg? and what are the odd's of getting a cherry like these? I predict aversge OC of ~3.8-9.

Originally Posted by Xbit

The major difference between Phenom II X2 and Athlon II X2 is evidently in the configuration of their cache-memory system. Phenom II X2 has a 6MB L3 cache typical of most Phenom II processors. Athlon II X2 has no L3 cache, but instead – what a surprise! - boasts twice as large L2 cache. The modular structure of the Deneb core has definitely worked very well for AMD here. It allowed AMD engineers to combine the L2 cache memory of the two working processor cores with the L2 cache memory of the disabled cores. We tried to explain on the next picture how they transform a fully-functional quad-core Deneb die into the two dual-core modifications aka Callisto (Phenom II X2) and Regor (Athlon II X2).

I like being correct way back here AMD Delays Release of Quad-Core Athlon II.
Gotcha BAMA

Originally Posted by Archer0915

For the looks of it I just have to wonder if AMD is going for a one size fits all die?? It seems that the 1 meg L2 duals must be a modified core or are the starting a modular design and................never mind I don't think they are smart enough for that.

Originally Posted by Kuroimaho

Actually that might be real, not really modular design but still a good idea. Because the cache is between the cores, on the native duals they leave the cache of the two removed core so they double the cache for those duals.

From ground up modular design will be Bulldozer but the mobile line might come first those are much more in need of a refresh than their desktops.

I am quite certain I will pick up one, for the price it will be fun.

Another thing I just read that the Istanbul will be an upgraded design of K10.5 with better power savings not just 2 additional cores, it might trickle down to these new cpus which could also explain the delay.

Originally Posted by Archer0915

A modular design would be cost effective and MORE VERSATILE than just cutting off cores the possibilities are limitless with one base die design. This disabled crap may be a good thing in the end as it may lead to a flex design concept in which cores can be cut and caches combined or split or halved ............ you get the picture. A way to use all of what is there neede to be implemented ?reserve registers? redirect commands? offloading through HW to a sud processing unit? the registers are there tha pipes are open if the core is not up to spec hardwire the CPU to offload to a slower locked core when load reaches XX% of core max load.

Originally Posted by Kuroimaho

The L2 less duals are native. No disabled crap, that's why they can launch them sooner than the quads and the triples.
Exactly as you would like it, they took the quad kept the L2 of the 3rd and fourth core, but removed the cores (will not take up die space) so this way the dual will have twice the cache of the L3less triplcecore and quadcore. This is how I imagine they will make the modular designin the future, they skew the die space so one cpu ends up with less modules while the other more and this can be adjusted wiith relative ease compared to current cache and core salvaging.
That's why I said it is nicely done.

Bama either the yield is pathetic or the demand is there either way I guess this saves money I agree with the logic of your initial synopsis but I figured that logic would not rule.

Originally Posted by AlabamaCajun

Like an X3.5

Cost is the major factor and transfer speeds are the other. Could they produce a modular core die, sure but not with an evil beast breathing down your neck. On that X3.5, you have a point, why not just release the X3s with the lame core running in low gear.

The whole pardigm needs to shift and get away from X86 and go with one of the cell designs. Sure a few programs run on quads but how many titles are available in scaleable threads? Also consider the trend to "boxes" that has just about decimated the PC game market.

AMD will soon drop the TDP on the quads and up the speed, we know that's coming. I'll get one of the newer low power AM3s and a mobo later when the timing is right. I may also get a dual if it will run in a AM2+ board.

[Zurvan]

How on earth does Xbit know Regor is Deneb derivative ?

[Kuroimaho]

Originally Posted by Bima Sylirian

Now it's just how they compared to E7000 & E8000 series in gaming.

As I sold my 8200ES and we know murphy's law, I know what to expect.

Originally Posted by nzaneb

I like that they utilized the L2 from the disabled cores for Regor. It really helps it to be a contender, even without the added benefit of L3.

Yes actually it is faster than I expected, and with the high multis unlocking on the BE only befits those with older mobos who could not OC it through HT.

Originally Posted by Archer0915

When will they be on the egg? and what are the odd's of getting a cherry like these? I predict aversge OC of ~3.8-9.

I guess around what Kuma's cost them. Launch about CT so early June.

Originally Posted by Archer0915

I like being correct way back here AMD Delays Release of Quad-Core Athlon II.
Gotcha BAMA

Come on take a look at the picture, it's ridiculous they use like half of the die with the 2X0, especially if we consider they harvest via triplescores, and cache via the 8X0 cpus.

Then they launch Caspian same time as these, that's a dualcore, 2MB L2 sounds familiar isn't it ?

How on earth does Xbit know Regor is Deneb derivative ?

Well they delayed the quads without L3 so they can not harvest these from those.

The only other way would be to use caspian for these which will also debut on Computex as noted above. The caspian doesn't have DDR3 controller, so up to which do you think is more extreme having a disabled DDR3 controller on a mobile cpu or more than half of the die on a desktop cpu. But why would they only release DDR3 caspian's next year only ?

[Archer0915]

Originally Posted by Kuroimaho

I guess around what Kuma's cost them. Launch about CT so early June.

Come on take a look at the picture, it's ridiculous they use like half of the die with the 2X0, especially if we consider they harvest via triplescores, and cache via the 8X0 cpus.

Then they launch Caspian same time as these, that's a dualcore, 2MB L2 sounds familiar isn't it ?

Originally Posted by Archer0915

Bama either the yield is pathetic or the demand is there either way I guess this saves money I agree with the logic of your initial synopsis but I figured that logic would not rule.

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
YUP BAD YIELD

Thanks for the price and time line guesses.

[utech]

x2 550BE and 250 sounds good and looks good.

I should wait them rather than 720BE now.

By the way, as we can see, DDR3 doesnt give AMD a big boost on performance, and DDR2 still have long life in the market now, and X2 may targeted at someone who just want to upgrade their old X2 but does not want to change the motherboard.

If you can get an AMD new X2 (can OC to nearly 4G) running on your old motherboard and only cost you around $150, then compare with the upgrade to DDR3's cost. You will easily made the decision.

[ChanceCoats123]

That is a great looking chip!

[QuietIce]

Originally Posted by cdawall

NB clocks suck wonder if thats the chip or they just left them alone to oc?

What? 2400+ and 2600 sucks ...?!?!?

[KTE]

Originally Posted by Archer0915

I like being correct way back here AMD Delays Release of Quad-Core Athlon II.
Gotcha BAMA

You were both guessing

Although his guessed assertions look to be wrong... we don't own crystal balls, simple as that.

Or do they? I have no idea how this Regor is the same new native dual-core, caches not even tweaked for the core and with the smaller die size. Xbit is showing it to be a salvaged Deneb which means die size will be the same as Deneb, nothing like all the gossip before this. There's no new mask here.

But Xbit won't be saying that for no reason, they'll have reviewers NDA docs explaining it all including the images to use by now.

I can estimate Chew/AMD OC guys posting some OC as soon as they're allowed, which is usually the night before NDA expiry.

[Zurvan]

Originally Posted by Kuroimaho

Well they delayed the quads without L3 so they can not harvest these from those.

Regor was supposed to be NATIVE dual core, not harvested Propus.

L2 is cache to individual core. I am really amazed if the design is modular enough to allow L2 from one core be added to another, without any design changes. Since they claim its harvested Deneb, how is that even possible when Deneb fixes 512kB per core.
Even if thats possible, why do Heka and Callisto have only 512kB L2 per core ?

[Kuroimaho]

A lot of things were supposed to be out by now as well propos included so there must be some trouble on their front so Xbitlabs might be right, but as I wrote it above it is launched the same time as of the mobile platform with caspian a dual with 1MB L2 / core I rather believe it is the same die as the mobile than harvested deneb.

Concerning prices at least here it should launch 13K Yen and 11K Yen, so under 130 and 110$ I find them a bit high going triple would be a dozen $ over the 550 and quad only 50$ more.

[Archer0915]

Originally Posted by KTE

You were both guessing

Although his guessed assertions look to be wrong... we don't own crystal balls, simple as that.

Or do they? I have no idea how this Regor is the same new native dual-core, caches not even tweaked for the core and with the smaller die size. Xbit is showing it to be a salvaged Deneb which means die size will be the same as Deneb, nothing like all the gossip before this. There's no new mask here.

But Xbit won't be saying that for no reason, they'll have reviewers NDA docs explaining it all including the images to use by now.

I can estimate Chew/AMD OC guys posting some OC as soon as they're allowed, which is usually the night before NDA expiry.

HEY DON'T BUST MY BUBBLE I was feeling good about that But in all reality I don't think it was modular by design but it just worked out that way.

[cdawall]

Originally Posted by QuietIce

What? 2400+ and 2600 sucks ...?!?!?

all of the quads i have gotten will do 2900-3100 stable on air with 1.4v NB, now the quads i have used have all been ES chips so that may make a difference (3x945 C1's and a 955 C2)

however my 720BE's (3x) all did 2600-2800 on air 1.5v NB

[QuietIce]

That explains it. I'm not sure I'd want to put that much voltage into it 24/7 ...

[cdawall]

Originally Posted by QuietIce

That explains it. I'm not sure I'd want to put that much voltage into it 24/7 ...

i haven't killed anything so far