AMD Zambezi news, info, fans !

[Formula350]

If they did a 125W 4140P it could probably reach 4G default as the BD silicon core-gates itself. They should try doing a "E8400" along with the top-end, as 4 int modules will be all that most people need.

I think they should make something like an FX 4130P (I assume P = Performance) that has 8 modules, but only uses 1/2 of each to take advantage of all the potential bandwidth heh (John mentioned that on the 20Q blog, that a thread on each module would yield more throughput then two threads pers, but would produce more heat since modules couldn't shut down.)

I'd consider buying that, if they'd overclock into the 4.5GHz area almost guaranteed

[Manicdan]

I think they should make something like an FX 4130P (I assume P = Performance) that has 8 modules, but only uses 1/2 of each to take advantage of all the potential bandwidth heh (John mentioned that on the 20Q blog, that a thread on each module would yield more throughput then two threads pers, but would produce more heat since modules couldn't shut down.)

I'd consider buying that, if they'd overclock into the 4.5GHz area almost guaranteed

this might be a catch 22 though,

if you have 4 threads on 2 modules, you get 10% less perf, but you might get 15% better power savings compared to having 4 threads on 4 modules. that extra power savings turning into higher clocks and might get you 10% faster speeds, resulting in the same exact performance in the end.

so the question will be, does 4ghz on the FX-8110 equal 4.4ghz on the FX-4110, and which one will consume more power when running 4 threads

[Olivon]

Where's AMD Bulldozer ? - Sweclockers

Although the boards are ready for launch, the engineers have not access to real production samples of Bulldozer-based processors. Instead they use scaled-down and locked "engineering samples" and the older Phenom II to function test the new models.

The situation has even gone so far that motherboard manufacturers are reluctant to launch the new models in June and is considering cuts in supply. Although it is still using the old Phenom II socket AM3, but this combination is more difficult to market.

Without a precise planning from AMD's site now requires manufacturers to the new processors will take until August or even September. It is dangerously close to Intel's launch of Sandy Bridge-E which has the potential to steal the limelight, at least among enthusiasts who put performance first.

[Brother Esau]

this is why I was forced to build a Intel System for my server, Hardware failure dictated that I could wait no more for these phantom processors. Let alone the fact that the server side of Bulldozer will no see the light of day until Dec...maybe!

[demonkevy666]

this is why I was forced to build a Intel System for my server, Hardware failure dictated that I could wait no more for these phantom processors. Let alone the fact that the server side of Bulldozer will no see the light of day until Dec...maybe!

Enron Of cpu's....

IMO
it's because the boards aren't there is BS.
The board them self can take AM3 cpu's In fact they where slated to be out by March If I recall.
There isn't a problem with the cpus it's the licensing SLI doing it making boards late.
I'd like to also know why everyone is claiming the I7 2600K as high end ? When it's mid-range for intel. :/

[Brother Esau]

Enron Of cpu's....

:/

]

I know, my server feels dirty and ashamed I hate Intel with a passion! But what are you going to do when single socket AMD Server / Workstation solutions are gimped?

[RussC]

What a shock, if true.

RussC

[freeloader]

I'm pretty sure someone else has said it, but if AMD has to significantly delay the launch of BD, wouldn't they have to inform their shareholders?

As far as I know, they have not issued any warnings of a delay to anyone, yet.

[DarthShader]

Sounds like AMD secrecy is bordering on paranoid.

[Brother Esau]

Frankly, I am dissapointed on the fact that I had to resort to a Intel Build for my server. But, considering the fact that I have been waiting 4 years for this Unicorn it was more of a reality decision versus a paranoia one, just could not wait any longer do to hardware failure

[eXa]

this might be a catch 22 though,

if you have 4 threads on 2 modules, you get 10% less perf, but you might get 15% better power savings compared to having 4 threads on 4 modules. that extra power savings turning into higher clocks and might get you 10% faster speeds, resulting in the same exact performance in the end.

so the question will be, does 4ghz on the FX-8110 equal 4.4ghz on the FX-4110, and which one will consume more power when running 4 threads

Also, there is supposedly some situations where having the work done on 2 modules rather than 1 can hurt performance. Shared resources can be an advantage sometimes.

[keithlm]

Also, there is supposedly some situations where having the work done on 2 modules rather than 1 can hurt performance. Shared resources can be an advantage sometimes.

We don't really know yet whether your statement is true or not. We can theorize that in certain situations it could be true. But in real world applications it might not happen. (Unless you have a source that has broken NDA?)

PLUS: That kind of situation would probably happen much less often with the BD architecture than it would with the competition's architecture.

We'll all just have to wait and see.

I think it will be interesting to see what benchmarks that have been extremely important to the competition for the last few years might suddenly become much less important. (Or what older versions will stop being quoted.)

[Formula350]

this might be a catch 22 though,

if you have 4 threads on 2 modules, you get 10% less perf, but you might get 15% better power savings compared to having 4 threads on 4 modules. that extra power savings turning into higher clocks and might get you 10% faster speeds, resulting in the same exact performance in the end.

so the question will be, does 4ghz on the FX-8110 equal 4.4ghz on the FX-4110, and which one will consume more power when running 4 threads

Yea, that's a good point. I didn't really consider the Turbo bonus any :\ Out of sight, out of mind; I have no Turbo, therefore I don't even factor it into situations heh

Also, there is supposedly some situations where having the work done on 2 modules rather than 1 can hurt performance. Shared resources can be an advantage sometimes.

Yea, but John said it requires added programing or specific coding in order to take full advantage of the shared resources design :\ I had mentioned that earlier, where he went on to say that M$ was working on implementing that for future OSes. I figure it's going to be in Win8, but naturally I'm hoping for a Win7 patch to utilize it.

We don't really know yet whether your statement is true or not. We can theorize that in certain situations it could be true. But in real world applications it might not happen. (Unless you have a source that has broken NDA?)

See response above...