facebook rss twitter

Review: AMD Athlon 64 FX-53 CPU

by Tarinder Sandhu on 18 March 2004, 00:00

Tags: AMD (NYSE:AMD)

Quick Link: HEXUS.net/qaw7

Add to My Vault: x

3DMark2001SE, Comanche 4, X²

Both AMD and Intel often position their ultra-expensive CPUs as the first choice for die-hard gamers. When every frame counts, as the saying goes, accept on the very best that money can buy.



A couple of interesting points here. We can call 3DMark2001SE a tie between the FX-53 and 3.4GHz Extreme Edition. ~21,500 marks is impressive going with a stock Radeon 9800 XT 256MB. What's more interesting is how the FX-51 loses over 300 marks to the Model 3400+ S754 Clawhammer. Non-3D tests have shown a propensity towards bandwidth over latency. 3DMark2001SE goes the other way. Of course, the performance discrepancy isn't purely a latency case. Different chipsets are being used for either processor, and VIA K8T800 (Model 3400+) may well have a better AGP implementation than NVIDIA's S940 nForce3 150. Still, it's strange to see a supposedly inferior CPU turn the tables. FX-53's result can be found here and the FX-51's here



Order is restored, for now. A 200MHz speed bump to the FX architecture sees it take first place back from Intel's ever-so-expensive Extreme Edition CPU. Comanche 4's a fan of large on-chip cache counts, as seen by the performance improvement of the 3.4GHz Northwood Extreme Edition over the regular Northwood. That may explain away the FX-51's lead over the Model 3400+.



Comparative X²- The Threat performance is not as bizarre as it first seems. When we looked back at a couple of S754 nForce 3 150 and VIA K8T800 boards we found that the latter, running the same CPU, enjoyed a significant performance lead over the former. That seems to have carried through to S940 ranks, too. AMD's Athlon 64 Model 3400+ isn't intrinsically 11FPS faster than the FX-51 - it's just that the chipset choice heavily colours results. We're not really complaining at the FX-53's 137FPS average, however.