Now we go over to the “Advanced Chipset Features” page. There are a lot of various options here. The first half of them refers to the memory subsystem. We can adjust CAS Latency, RAS-to-CAS Delay, Row-precharge Delay and Row-active Delay in wide ranges. At the same time, there is no such parameter as DRAM Command Rate. Seems like it’s the chipset, not the mainboard, that doesn’t allow adjusting it.

Among interesting and non-standard parameters we see “AGP (and also FSB) Spread Spectrum” and “CPU Thermal Throttling”. To cut it short, the first option allows reducing the level of electromagnetic noise from the mainboard, but also reduces system stability during overclocking. This option is quite useful: some people say that turning on the “Spread Spectrum” parameter results in better image quality output onto the TV-set through the graphics card’s TV-out. By the way, most mainboards have just simple “Spread Spectrum” parameter, without the division into FSB and AGP, and two positions (Disabled and Enabled), while NF7-S has an in-between value – a compromise between system stability and noise level.

As for “CPU Thermal Throttling”, we can’t say anything definite about this option (note: the overheat protection system in Pentium 4 has a similar name). We will tell you about it as soon as we know what it actually serves for.

The “PC Health” page contains all the info from the hardware monitoring system of the mainboard: the voltages and rotation speeds of three fans. There is the “CPU Shutdown Temperature” – the parameter that duplicates the parameter from the “SoftMenu III” page. The “CPU Warning Temperature” sets a certain max temperature, and when the system gets over it an emergency signal about CPU overheating is sent. The “Shutdown when CPUFAN Fail” option says whether the system shuts down when the CPU fan goes down. You should be careful with this option if you use some low-speed fans. The hardware monitoring just may not “see” such a fan and the system wouldn’t power up.

Let’s now turn from the descriptions to our everyday overclocking practice. We checked the ability of the mainboard to increase the FSB frequency without increasing the memory frequency and the ability of the mainboard (and memory) to speed up when the FSB and memory are in the sync mode. The second thing is more interesting from the practical point of view as NVIDIA nForce2 works best when it is synchronously clocked with the memory. Anyway, the first part will help us to check whether ABIT keeps its word about the maximum FSB frequency.

So, we raised the chipset voltage to the maximum 1.7V and set the promised 237MHz FSB. We were surprised to find that the mainboard did quite well at such FSB frequency. At least, we couldn’t make the system hang up either by running all the demos in Unreal Tournament 2003, or running 3DMark 2001 SE three times in a row, or with the help of Sandra Burn Wizard. So, ABIT’s promise about the max FSB frequency appeared hundred percent true!

Now, to the sync mode. The chipset voltage was set to 1.7V, the memory voltage – to 2.7V, memory timings were 2-3-3-7 (CAS Latency, RAS-to-CAS Delay, Row-precharge Delay, Row-active delay). We used Corsair XMS3200C2 memory modules. As this review is not about memory we didn’t play with different memory timings to find their effect on overclocking and performance.

The highest FSB frequency the system worked stable at was 212MHz. At the next value, 217MHz, the system was unstable: tumbled out of Unreal Tournament 2003 into the Desktop. We guess 212MHz is quite well both for the mainboard and memory. At the same time, we suppose the result might have been higher if we could raise the memory voltage a bit more. But we can’t yet prove it.