Archive 1997–2024

ASUS Maximus Formula SE: X38 and DDR2 Unite!

Synthetic Benchmark Performance

Everest Cache and Memory Benchmarks

9X266

9X333

9X400

8X465

8X465 DFI LP UT P35 T2RG

Our 8 hour Prime95 torture test run ended at 450FSB; 460-465FSB provided around 1 hour of torture test stability and allowed us to complete all benchmarks but Prime95. The same CPU and components managed 465FSB on the DFI P35 T2R, but with a far longer 11 hour Prime95 torture test pass - without any failures. Unfortunately, the ASUS 0505 BIOS does not provide sufficient GTL tweaking ability to extend CPU core stability. To get the board close to P35 DDR2 performance we are limited to the 400 NB strap and 3:4 memory divider ratio. At higher FSB points, the 400 NB setting is the only strap that is stable enough for 24/7 use. This makes it quite apparent that Intel's engineering favors 1600FSB support and DDR3 for the X38 and future processors. DFI P35 results have also been added at 465x8. The only way to run the Maximus Formula in a fully competitive manner is to run the memory in excess of 1200MHz, which limits users to top-end DDR2 modules.

The DFI board runs DDR2-1120 very happily at low VDimm with the right modules; the Maximus is limited to either the 1:1 or 3:4 using the 400 NB strap. Using 1:1 would have limited memory speed to DDR2-930 on the Maximus, which is quite low unless you run CAS 4-3-3 timings. Such low latency timings required a very high VDimm, and 1N did not seem to work at this speed either, so we chose to stick with DDR2-1240 CAS 5-5-5-15, which was possible at a reasonable 2.2V.

Originally published at https://www.anandtech.com/show/2374 (Wayback Machine).