AMD to start Bulldozer AM3+ production by March 2011, launch in April 2011

[FlanK3r]

yes, ES at 32nm SB are, but one man said me, for production at 32nm is LGA2011 cancelled....(and at October 2010 roadmap I seen not info about LGA2011 :-) )

[nunchucka]

why is it always the same on this forum that i come into a thread titled amd bulldozer, and leave because all thats in it is about intels next ... what is it with the mods here?

[freeloader]

why is it always the same on this forum that i come into a thread titled amd bulldozer, and leave because all thats in it is about intels next ... what is it with the mods here?

Fugger + Intel perhaps? Don't know, but it would be nice for Intel users to stay out of AMD threads and vice versa.

[Sn0wm@n]

yeahh more about bulldozer ... less about sandy bridge ....

[FlanK3r]

When do you escape some result with ES Zambezi?

[informal]

It's going to be a very tight seal on BD info until launch in Q2.

[Johnny87au]

It's going to be a very tight seal on BD info until launch in Q2.

Just like there release of the 6xxx and probably the same with Antilles

[generics_user]

Just like there release of the 6xxx and probably the same with Antilles

considering the performance preview of bobcat i expect a preview in march (maybe earlier) together witht he 900 series chipset

[Oupavoc]

any leaked prices?

[FlanK3r]

lol, prices and now? about 5 months before launch?:-D

[mstp2009]

All I care about is 1) IPC, 2) Overclockability, and 3) thermals.

Until that is known, all this is fluff.

[FlanK3r]

add 2) seems good from design...But, time show more...

[Solus Corvus]

All I care about is 1) IPC, 2) Overclockability, and 3) thermals.

Until that is known, all this is fluff.

The realworldtech article gives us some hints to some of these questions. The indicators are for medium IPC and high frequency. Of course thermals probably won't be known until launch.

[-Boris-]

All I care about is 1) IPC, 2) Overclockability, and 3) thermals.

Until that is known, all this is fluff.

IPC is useless without frequency. What would you like the most? 1: 500MHz at 125W with a performance equal to a X6 1055T? Or 2: 8000MHz at 95W with a performance 40% higher than i7 980X?
You want performance, thermals and overclockability. IPC and frequency are secondary.

[nn_step]

IPC is useless without frequency. What would you like the most? 1: 500MHz at 125W with a performance equal to a X6 1055T? Or 2: 8000MHz at 95W with a performance 40% higher than i7 980X?
You want performance, thermals and overclockability. IPC and frequency are secondary.

IPC * clockspeed = performance [except in cases explicitly independent of system performance]

What is wanted is an optimal mix of performance, power, and efficiency.

[Motiv]

lol, prices and now? about 5 months before launch?:-D

I predict the prices will be inline, with intels equivalent products. THere may possibly be a premium on server chips, if the thermals for multi-core processing are under control.

damn I'm good.

[-Boris-]

IPC * clockspeed = performance [except in cases explicitly independent of system performance]

What is wanted is an optimal mix of performance, power, and efficiency.

As I said:
"IPC is useless without frequency."

And how do you define efficiency? Isn't a good efficiency an optimal mix of performance and power? Well and maybe die size.

[demonkevy666]

The realworldtech article gives us some hints to some of these questions. The indicators are for medium IPC and high frequency. Of course thermals probably won't be known until launch.

wait then what is you base for IPC ? cpu wise it must have a control you getting that from?

??

[Solus Corvus]

wait then what is you base for IPC ? cpu wise it must have a control you getting that from?

??

I mean probably higher then P2 but lower than SB. They look to be making up for it with clocks and cores.

[FlawleZ]

I mean probably higher then P2 but lower than SB. They look to be making up for it with clocks and cores.

Yeah, I have a feeling that IPC will be very similar to Nehalem.

[nn_step]

As I said:
"IPC is useless without frequency."

And how do you define efficiency? Isn't a good efficiency an optimal mix of performance and power? Well and maybe die size.

efficiency is how well the processor fits the task at hand. For example, massively parallel tasks tend to be far better suited towards simple in-order cores. Were as single threaded but compute heavy tasks tend to be far better suited towards super heavy OoO cores.

[zerazax]

No, there's different kinds of efficiency

From the product perspective, it's how much performance can you get for the given constraints

[haylui]

I'm not sure what IP stands for in this case, but if Brazos has IP in november and is supposedly launching in January then wouldn't June be a better guess for Zambezi?

Initial Production?

[MAS]

AMD keep silence. Four months before launch - no tests. The most likely scenario is that zambezi has single-thread performance level compared to nehalem (clock-for-clock), in other words zambezi is faster than K10.5 by the same per cent K10.5 is faster than K8 (10 per cent)

The main advantage of bulldozer is therefore its eight (pseudo-)cores.

[Olivon]

AMD keep silence. Four months before launch - no tests. The most likely scenario is that zambezi has single-thread performance level compared to nehalem (clock-for-clock), in other words zambezi is faster than K10.5 by the same per cent K10.5 is faster than K8 (10 per cent)

The main advantage of bulldozer is therefore its eight (pseudo-)cores.

No tests, no previews = Sandy bridge road cleaning ...