: News
: Reviews
: Editorials
: About
: Contact
: Advertising
: Privacy
: Links
: Forums




AMD Athlon 64 3200+ Socket 939 90 nm


Smaller, Cooler, Faster...

by Josh Walrath



            Intel was one of the first to venture into the 90 nm waters with the release of the Prescott core Pentium 4 processor in January, 2004.  The complaints about massive heat production and power consumption soon followed in reviews and articles, and many were pointing to Intelís new 90 nm process as the culprit.  This was perhaps a bit disingenuous on these writersí parts, but it is not totally incorrect.  The 90 nm process does exhibit a higher percentage of leakage than past processes, but this is a percentage and not an actual power figure.  The largest problem with the Prescott is its design.  To run at that high of a speed, with that many transistors, there has to be a lot of power applied to the design.  The more power that has to be applied to make the transistors switch faster, the higher the leakage will be.  It is simple physics, and not some nefarious design flaw in the 90 nm process.  To prove this point simply look at the Dothan core for the Pentium M (the 90 nm, 2 MB L2 cache Centrino processor).  This is a very cool running and energy efficient processor that is made side by side with the Prescott.

            This is not to say that the jump to the 90 nm node was not without its fair share of problems.  Many remember the birthing pains with the jump to 130 nm, but the transition to 90 nm was even harder.  Time and money are the necessary cures for any process jump, and throwing enough engineers at the problem tends to make things work better.  Intel had a lot of engineers working on the 90 nm node, and they were the first out of the gate with a mass produced product (though IBM had 90 nm PowerPC parts at this time, they were not produced in such large quantities as what Intel did with Prescott).  Intel engineers still do a lot of work to refine their process, but changes with the Prescott core have also been made to lower the power and heat problems.

            The release of the Prescott and all of its problems led many to believe that any 90 nm product from any manufacturer will experience these same problems.  This is a very poor assumption (as proven by the Dothan core).  Many also believed that if Intel had as many problems as it did, it would follow suite that AMD would experience the same problems.  Another poor assumption.  Intelís problems are Intelís alone.  AMD has had its fair share of problems with their migration to 90 nm, but the very fact that the Athlon 64 core is much different from Prescott has helped them along.  The Athlon 64 is a very well balanced design that does not push the clock speed envelope as the Prescott does.  Because AMD is not forced to make the transistors switch as fast as possible, it is inherently more power efficient and cool running.  In fact, the design has more in common with Dothan than it does with Prescott.

While this isn't the exact processor I have (obviously), it is essentially the same.

            Still, the big question in everyoneís mind is if the 90 nm Athlon 64 will be as hot as the 130 nm version, or in fact hotter.  AMD has added something extra into the mix that Intel doesnít have, and that is SOI (Silicon On Insulator).  This extra wrinkle may make all the difference in the world.  We have seen the effects of SOI on AMD products, as the Athlon XPís on their base 130 nm process were fairly hot running and maxed out at 2.2 GHz.  The Athlon 64, a much more complex product, is able to do 2.6 GHz on 130 nm SOI (with some strained silicon thrown in for good measure).  The Athlon 64 also runs significantly cooler than the older Athlon XP.  So will SOI make a difference as well?

            To put these things to the test, AMD was kind enough to ship me a 90 nm Athlon 64 3200+ Socket 939 processor.  Externally there is nothing different with the 90 nm product, as it is covered by the same heatspreader the 130 nm versions use.  If the heatspreader were removed, then the differences would be immediately obvious.  The older Newcastle core (130 nm 512 K L2 cache) is around 144 mm square, while the new Winchester core (90 nm 512 K L2 cache) is around 84 mm square.  The 90 nm node allows AMD to make far more Athlon 64ís than the 130 nm process.  Just for reference, the Clawhammer core (130 nm 1 MB L2 cache) was a massive 193 mm square, but the new Venus/Troy/Athens cores (90 nm 1 MB L2 cache) will only be 114 mm square.


Next: Test Setup


If you have found this article interesting or a great help, please donate to this site.


Copyright 1999-2004 PenStar Systems, LLC.