Intel Pentium D 820 (operating at 2.8GHz - 14x200MHz); Shuttle SD31P (Intel 945G, ICH7R); 2x 512MB Corsair XMS2-5400UL (operating at 533MHz, 3.0-2-2-12*); XFX GeForce 6800 GT (operating at 350/1000MHz); Western Digital WD2500 Caviar 7,200RPM 250GB SATA150 hard disk drive; Windows XP Professional Service Pack 2; DirectX 9.0c; NVIDIA Forceware 71.89 WHQL.
Intel Pentium D 820 (operating at 2.8GHz - 14x200MHz); Gigabyte GA-8N-SLI Royal (NForce 4 SLI Intel Edition); 2x 512MB Corsair XMS2-5400UL (operating at 533MHz, 3.0-2-2-8); XFX GeForce 6800 GT (operating at 350/1000MHz); Western Digital WD2500 Caviar 7,200RPM 250GB SATA150 hard disk drive; Windows XP Professional Service Pack 2; DirectX 9.0c; NVIDIA Forceware 71.89 WHQL.
* We mentioned earlier that we had problems setting TRAS - we actually selected TRAS 8 in BIOS, but according to CPU-Z the board was running at TRAS 12. We opted to set the GA-8N-SLI Royal to TRAS 8, seeing as we have found that to be the optimal timing for our Corsair 5400UL.
General Performance
Summary:
We cannot quite nail down the reason why the Shuttle SD31P is slower than the Gigabyte GA-8N-SLI. It could be down to the TRAS adjustment bug that we mentioned on the previous page, or it could be that the board lacks a 1T command rate. We find the latter may not be the case though, because the performance differential doesn't seem to relate to a typical performance difference between the 1T and 2T command rate. We are left to conclude that NForce4 for Intel is faster than 945X.