AMD Athlon™ 64 FX-57

by John Reynolds

 

Introduction

AMDAMD has held a firm lead in gaming performance since the release of their Athlon 64 processors, and the company only strengthened that position with the release of the FX-57 earlier this summer. The unfortunate side effect of AMD’s lead over Intel, however, is that this new flagship part was introduced at a new, higher price point compared to previous Athlon 64/FX releases, and without the expected trickle-down on the pricing of existing processors already available on the market. Does the FX-57′s performance justify its price, though? We’ll take a look at how it handles SimHQ’s benchmark suite after briefly examining the changes and improvements AMD have made to this new part.

The FX-57 arrives clocked at 2.8GHz, a speed bump of 200MHz over the FX-55, with a transistor count of 114 million compared to the FX-55′s 106m. Based on AMD’s San Diego core, the FX-57 is manufactured using a 90nm SOI (Silicon on Insulator) process at the company’s Dresden fabrication plant. AMD has been gradually transitioning their desktop parts to this new process, introducing the San Diego (1 MB L2 cache) and Venice (512 KB L2 cache) cores earlier this year as the company worked to refine the 90nm process. The FX-57, however, is not just a new part with a simple speed bump derived from a new process: AMD has introduced some changes to this new processor that are worth a quick overview.

The Athlon 64 series’ on-die memory controller has in the past displayed compatibility and performance issues with mismatched memory modules and with fully populated memory slots. The San Diego core that the FX-57 is based on has an improved memory controller that allows for different sizes of installed memory modules in dual-channel mode; moreover, the controller also maintains memory speed with fully populated slots, unlike previous cores that saw performance degradation under such configurations. AMD has also added support for SSE3 instructions to the San Diego core and tweaked the part’s TDP (thermal design power) to be the same as that of the FX-55, allowing the new part to operate just fine with existing cooling solutions. Overclockers in particular might want to take note of this fact.

Months after its initial release the FX-57 remains priced at over $1,000; the FX-55 is currently priced close to $800 and AMD’s high-end dual core, the X2 4800+, can be found for slightly under $900. SimHQ will compare the performance of these parts from AMD and include scores from Intel’s Pentium 4 3.73GHz Extreme Edition since its price is very close to that of the FX-57.

Test Systems Setup

  • Athlon 64 FX-57
  • Athlon 64 FX-55
  • Athlon 64 X2 4800+
  • ASUS A8N SLI Deluxe (nForce4) motherboard
  • 1 GB (2x 512 MB) Corsair DDR400 memory (2-2-2-5)
  • ATI Radeon X800 XT PCI Express graphics board (Catalyst 5.8)
  • Windows XP Professional (SP2)
  • DirectX 9.0c
  • Pentium 4 3.73GHz Extreme Edition
  • Intel 955XBK (955X chipset) motherboard
  • 1 GB (2x 512 MB) Micron DDR2 533MHz memory (4-4-4-12)
  • ATI Radeon X800 XT PCI Express graphics board (Catalyst 5.8)
  • Windows XP Professional (SP2)
  • DirectX 9.0c

The benchmark suite used for testing is listed here; all software titles used for testing were installed with the latest, most up-to-date patches. As standard practice, 32-bit color and trilinear texture filtering are the default baseline during testing, and anti-aliasing and anisotropic texture filtering were disabled throughout all tests. Also, Windows XP Professional was configured to have Automatic Updates, System Restore, and all unnecessary startup services disabled. Last, unless noted otherwise the licensed version of Fraps 2.6.4 was used to record performance scores.

Go To Page 2