Take the 2-minute tour ×
Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.

I'm studying Moore's law. I'm curious why many people say Moore's law ends, so we switch to multicore CPU. From all materials I can find, I understand this by such logic: increasing clock rate -> increasing transistor count -> increasing power consumption -> don't increase clock rate, instead, increase the number of core -> power consumption deceases

Now, a question arises: how to increase the number of cores? do we need to increase the number of transistors? if so, the power consumption problem still occurs.

Can anyone answer this question?

share|improve this question

closed as off topic by Linker3000, 8088, Nifle, Sathya Sep 8 '11 at 4:54

Questions on Super User are expected to relate to computer software or computer hardware within the scope defined by the community. Consider editing the question or leaving comments for improvement if you believe the question can be reworded to fit within the scope. Read more about reopening questions here.If this question can be reworded to fit the rules in the help center, please edit the question.

    
Perhaps two 1.5Ghz cores are potentially faster than one 3.0Ghz core? –  Hand-E-Food Sep 5 '11 at 23:21
add comment

2 Answers

up vote 0 down vote accepted

Power is not really the main motivation for multicore processing, it more has to do with performance for the tasks typical for a given system.

The big thing is that increasing clock speed only really helps if you have tasks that require huge blocks of computation. In this case, high frequency processors with deep pipelines really begin to pay off.

However, most tasks that run on a normal workstation need processing power only intermittently in response to events, but there are many hundreds of these tasks. In this case, it becomes advantageous to have better thread-level parallelism, which can be attained by having multiple processor cores.

The reason why multicore processors have really begun to take off in recent years is an effect Moore's law, not a cause of its continued validity. Chip producers can fit more an more transistors on a chip, but at this point, there's not a whole lot left to do to improve single core performance for typical workstation tasks. After all, adding more transistors doesn't really correspond to increased clock speed. So what do they do with all those transistors? Add more cores and increase multitasking performance.

It's not really a new phenomenon. Multiprocessor systems have been around for quite some time. It's just that now, with increased transistor counts, there's little reason to put cores on separate chips.

When you start talking about having dozens of cores, then significant power savings is possible, because you can turn off the cores you don't need when the system is under low load.

share|improve this answer
add comment

The problem with increasing the clock rate on a single core is, well, physics. Eventually, you are testing the heat and conductive limits of all known elements.

Split the same power consumption over two cores means you can split the heat dissipation.

So increased power consumption over one core != increased power consumption over two cores.

Or think of it this way. In order to get one person to carry 300 lbs, you'll need to not only beef up their muscles, but their frame as well. The scaling fallacy tells you that this isn't a mere linear increase, but rather becomes a cubic increase in most factors.

Divide that same 300 lbs among two people and you solve a lot of issues that were a factor before. Balance, proper blood flow to the muscles, skeletal frame strength etc etc.

So yes, increased power consumption still occurs. But the bigger problem, heat, was solved.

So taking the same number of transistors off one core and splitting it into two cores or more is well worth the cost of added overhead.

share|improve this answer
add comment

Not the answer you're looking for? Browse other questions tagged or ask your own question.