SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Technology Stocks : Intel Corporation (INTC) -- Ignore unavailable to you. Want to Upgrade?


To: pgerassi who wrote (144603)10/2/2001 6:04:24 PM
From: Noel  Read Replies (1) | Respond to of 186894
 
Pete, Re: Did you ever take a look at how they did it?

I did look at the papers on this subject in detail and you are correct in that they do have what looks like two pipelines operating 180 degrees out of phase.

Okay, now here is the mindboggling part: Each of those pipelines is operating at 4 GHz so you actually end up getting an 8 GHz effect!!!

I am glad the marketing people did not use the 8 GHz number.

You make very good arguments about how you can make something appear to operate at infinite frequency by just making parts of it staggered by infinitesimal fractions of a clock cycle.

Just think of it, by your lights, a section that runs at 1000MHz and one that runs at 1000MHz 0.1 cycle out of phase could be thought as a single 10GHZ pipeline because one uops could be followed by another after only 0.1 of a base cycle.

But what you cannot do in the case you have described above is feedback the output of such a beast back into itself and make it operate at 10 GHz. It is only 10 GHz in the forward direction and not for feedback.

So most knowledgeable circuit types usually consider a part as operating at a particular frequency if it is capable of receiving data at a certain rate and then one is able to wire the output back to the input and still function at that same data rate. (You probably know that this kind of functionality is actually required for iterative operations.) Something like a pipeline oscillator. This is the best way to look at the real operating frequency.

From that point of view this ALU does operate at 4 GHz. It receives its data at 4 GHz, does its operation in 1/(4 GHz), and can use that output back at its input, alongwith any newly received data, to generate the next bit of data at 1/(4GHz) later.