ASUS Striker II Formula under the Microscope

Specifications - ASUS Striker II Formula

Over a year ago, the NVIDIA 680i reference board gave the Striker Extreme 680i a run for its money. This time, ASUS has refined the power delivery circuitry to key areas of the Striker II Formula motherboard (memory and chipset power), giving it an apparent technical edge.

At present, online pricing of the Striker II places the board around the $300 mark, while the EVGA 780i A1 boards are priced at $259 direct from the EVGA site or around $230 at other online vendors. We have tested both boards in the labs and initial comparisons do show a marked difference in component and engineering choices between boards. The ASUS motherboard uses better components all around (something you would expect at the price). The question is whether these changes will justify the additional expenditure to the consumer.

Ultimately, we feel that the choices are going to come down to a couple of things. While the NVIDIA reference boards generally have the upper hand for BIOS base-code fixes (these eventually filter down into the non-reference boards), the rest will be determined by how hard you intend to push each board for overclocking and if you get a return for the extra ~$70 that the Striker II costs over its EVGA counterpart.

Making a decisive judgment call on overall overclocking potential is not easy though. The NVIDIA chipsets have shown remarkable variance in overclocking headroom from board to board in the past. A disparity as large as 25MHz FSB between boards of the same model is not unheard of between brands. In order to keep this variance to a minimum, ASUS screens the SPP units for their R.O.G. boards to ensure the best chipsets are utilized.

At first glance, we would say that comparing dual-core overclocking with up to 4GB of memory is like splitting hairs, as both boards would appear to have sufficient engineering to cope with these kinds of loads. The big question will be over the much-publicized Penryn overclocking ability that hounded the NVIDIA 680i chipset motherboards. The change that threw 680i-based boards off guard was centered on lower processor reference voltage levels (known as GTL) for the 45nm Intel CPUs. The 65nm CPUs require a reference voltage in the range of 67% of VTT, while the 45nm process requires around 63-64% of VTT (termination voltage) to achieve stability. Changes like these require a component level rework on boards that lack EPROMs and/or the required divider networks onboard that can be switched via BIOS coding to offer fine voltage changes to GTL reference voltage levels.

A larger cache size for the 45nm processors (especially the quad-core processors) also places more of a capacitive load upon the chipset, requiring additional signaling drive, again something that just can't be added via a BIOS workaround if the hardware is not adequate. This is why the early 680i boards are not stable with the 45nm processors or even the 1333MHz FSB 65nm units. Essentially, we are still working with the same Northbridge (680i) with improved GTL reference value adjustments to give higher stable FSB speeds with Penryn processors.

We would like to think that 400FSB with a 45nm quad-core processor will keep NVIDIA's solution competitive with the overclocking ability of the Intel chipset based boards. Our primary aim to test this today is to strap a 4GB and 8GB memory load to a QX9650 processor and see how well the board will work for stressful loops of PCMark Vantage and games like Crysis.

These tests will be repeated on the EVGA 780i board over the coming week, with a full report and cross comparison of both boards. We had hoped to have the EVGA results today, but our board failed before we could complete our torturous overclocking routines. It appears we had a weak SPP as the board gradually failed to overclock our quad-core processor, and at the bitter end it refused to even run at stock speeds. We sent the board off for a quick burial, but not before a 21 water gun salute.