The Same Ol’ FX
A Familiar Architecture
Now that the excitement of introducing dual-core has died down, revisiting older designs isn’t nearly as enthralling. Not all is the same with AMD’s FX-57, though. This is the first FX family member with roots in the San Diego core, manufactured on a 90nm process and equipped with the E-stepping features (SSE3, improved memory controller, lower operating voltage). As a result of the node transition, die size shrinks from 193 square millimeters to a diminutive 115. And that’s after a small bump in transistor count, too, from 105.9 million to 114 million.
San Diego is characterized by a 1MB L2 cache, like the 130nm ClawHammer that preceded it. The core’s technological equivalent is Venice, which has already been available for some time with 512KB of L2. Although the FX-57 operates 200 MHz faster than the FX-55, jumping over to this new core helps keep thermal output manageable. Maximum thermal power is still rated at 104W, meaning that any cooling solution designed for the FX-55 works here as well.
The rest of the core’s vital specifications are the same. You get 128KB of L1 cache, divided between 64KB of data and 64KB of instruction memory. It still interfaces with a Socket 939 interface, and continues to support DDR memory at speeds of up to 400 MHz, yielding up to 6.4 GBps of memory bandwidth. The HyperTransport bus continues to operate at 1 GHz (2 GHz bi-directionally).
Then there are the value-added extras. Of course, 64-bit extensions are inherent to the architecture and now exploitable through Microsoft’s Windows XP Professional x64 Edition. Enhanced Virus Protection is another, accessible through Windows XP Service Pack 2 and used to cut back on certain types of malicious code. Finally, Cool’n’Quiet technology is now a standard feature on the Athlon 64 FX. Used to cut back on power consumption and heat output by reducing clock frequency, the feature can be turned off through your BIOS in order to maintain full performance.