Awesome, I voted the same as IDC without reading, I feel good about myself.
Does anyone have ideas on how faster clockspeeds/Higher Ghz would impact xtor size and thus die size for Bulldozer?
Do faster xtors need to be larger?
Thank you for the vote. For anyone else contemplating on voting please notice I set the poll to anonymous.
I don't think you can figure out anything useful with this line of reasoning. However, I find the relationship between sizing and area to be an interesting topic, so if you ask questions, I will try to answer them.switching speed has a rollover point versus xtor length so there will be a minimum length for which the switching time is also a minimum (what we laymen think of as a transistor "width" from all those fancy SEM cross-sections is actually the length of a xtor, the width runs perpendicular).Does anyone have ideas on how faster clockspeeds/Higher Ghz would impact xtor size and thus die size for Bulldozer?
Do faster xtors need to be larger?
The question you are asking though is really quite complicated if you aren't familiar with circuit design.
I'd love to see Ctho9305 or TuxDave break it down for the general audience, take it away fella's! :thumbsup:
The process is called chip-stacking and it happens already in the flash industry and set to become the standard required for producing the ram densities needed for DDR4 in another 2 yrs.
With regard to the stacked DDR4 mentioned, does anyone know how much of an issue heat dissipation becomes? Would it be enough to force lower memory speeds in order to maintain reasonable component life?
~300 mm^2
I think some people are forgetting about all the "uncore" stuff that is needed (big L3 cache, IMC, northbridge, hypertransport links), and that the modules are each likely to come with 2MB of L2.
Either that, or they don't realize this is a 4-module/8-core die.