Navigate: Advice Beginners BIOS Guide CPUs Links Mainboards Memory Network Storage Video/Sound Cards Contact Forum SiteMap Sponsors WebNews Home	.	.
Prices: Mainboards ABIT ASUS Chaintech Shuttle Soyo Tyan CPU Intel P4 2.4C-800 P4 2.6C-800 P4 2.8C-800 P4 3.0-800 P4 3.2-800 AMD AthlonXP XP 1700+ XP 2000+ XP 2400+ XP 2500+ XP 2700+ XP 3000+ XP 3200+ Athlon64 Athlon64 3200+ Athlon64 FX-51 Opteron Opteron 240 Opteron 242 Opteron 244 Opteron 246 Memory Corsair Crucial Kingston Mushkin OCZ Search Prices:

LOSTCIRCUITS	SHORTCUTS: Top Page Test Configurations Test Configurations Overclocking Winstone Sysmark2004 3ds max Amorphium, Caligari Cinebench3 Gaming Benchmarks 64bit Performance, Conclusions Athlon64 3400+ Pricing Give Us Some Feedback to Help Us Improve our Reviews
AMD Athlon64 FX53 Back to the top
(Review by MS, March 18, 2004)

Overclocking per se has become somewhat obsolete, there is not much to be gained in useable performance - in other words, who cares whether Quake can run at 800 or 1000 frames per second. On the other hand, overclocking is still an indicator of the headroom within any component. In the case of the Athlon64 (including -FX) series, some of the limitations are in the hypertransport and the fact that all clocks are depending on one single master clock. That is in contrast to most other chipsets, there are no locked AGP / PCI buses currently available, meaning that timing limitations of these two buses are capping the clocking attempts to approximately 225-230 MHz bus speed on average. There are certain exceptions, especially with the use of third party SATA controllers that are using their own clock, it has been possible to push the system clock a bit higher, however, the general limitations persist - something to keep in mind when evaluating the general OC-capabilities.

The way towards these levels of overclocking was a bit bumpy, at first, the system refused to run anything higher than 215 MHz bus speed even with the voltages maxed out. As it turned out, the key to reaching higher frequencies was actually to run the memory at the lowest (2.6V) voltage and bump the CPU core voltage up by a modest 150 mV to 1.7V. At those settings, the system POSTed at 2725 MHz (227 MHz bus), got into Windows at 225 MHz bus speed and ran most benchkmarks at 222 MHz (2664 MHz CPU clock).

For most benchmarks we settled at 217 MHz for the simple reason that the resulting clock speed in that case is 2604 MHz, which is approximately where the next speed bump (FX55) will be and where the system ran without a single crash. In all benchmarks, the system was running with the memory in synchronous mode at 2:3:2 (CL:tRCD-tRP) latencies.

next page:    => Sysmark - Business Winstones =>