AMD Bulldozer Core Patent Diagrams

I take it that they are basing the comparison on throughput, seeing as how there is less than 3% difference between the 2500K and 2600K in things that use 4 threads or less.

Perhaps "Across the board" is to strong if you define it as win in everything, but I certainly expect the 970 to be stronger in both low-thread applications and high-thread applications and win versus BD in the vast majority of applications for desktop, number crunching, rendering, encoding, compression and gaming.

 

If you phrase it like that then I partially agree, because I see it myself as reasonably possibility at this moment.
I guess top FX-8 will be somewhat competitive in MT tasks to i7 2xxxK CPU's but no so much in ST. The opposite can be true comparing to i7 9xx, e.g. loosing in MT but winning in ST thanks to aggressive Turbo.

But I rather wait and see. Better to have low expectations and eventually be nicely surprised than other way around.

 

A short errata description for the various BD re-spins.

 

Thanks fellix!

Looks like they struggle with clock propagation on this chip.
New process, new core, big headache

 

So is the L2 now fixed?

 

Oh, B0 seems quite bad. I better don't ask how A0 was looking then

 

http://www.brightsideofnews.com/new...rmance-projections-for-bulldozer-opteron.aspx

 

ouch. at least these will compete against some of the dual socket 2011, price-inflated six core and eight core Xeons.

 

rpg.314's and fellix's links/images do not work for me or where pulled out.

LE: It all works now here @ the office. Strange..

 

Is that for real ? How can BZ be that far behind ? that FX is 8 cores right ?

 

But 8 cores don't mean much if applications can't scale effectively to so many threads, like many of those listed applications.

 

That could be the case too. But the 990X bested the 2600k, I still believe that the 2600k has better single core perf but 990X is hex-core. Well hopes the 3 month delayed will turn something better.

 

http://www.xbitlabs.com/news/cpu/di...ck_Speeds_of_AMD_Opteron_Bulldozer_Chips.html

 

usually how much delta there's in desktop vs. server processors?

 

http://www.pcgamer.com/2011/07/26/amds-10-core-piledriver-chips-revealed/

 

the dual-die server processors are a good 1GHz slower than desktop processors (but with twice the cores and support for a lot more memory)

 

AMD Shows Off Opteron "Interlagos" Again: No Performance Benchmarks, No Design Wins, No Launch Date Announced.

 

That's it. Now I am firmly in the BD is fucked camp.