The mainboard allows adjusting:
- the CPU multiplier;
- the base (clock generator) frequency from 200 to 250MHz;
- the AGP frequency from 66 to 100MHz;
- Vcore from 0.8v to 1.55v stepping 0.025v;
- Vmem from 2.6v to 2.8v stepping 0.1v;
- Vagp from 1.5v to 1.8v stepping 0.1v;
- Vchipset from 1.6v to 1.9v stepping 0.1v.
The Cool’n’Quiet technology works fine, but you can disable it in the BIOS, if you wish. It is automatically turned off the moment you decrease the CPU frequency multiplier, while the option of increasing the multiplier is locked in Socket 754 processors.
You cannot change the voltage on the HyperTransport bus, but you can adjust its frequency up to 1000MHz. Memory timings can also be adjusted – they are not many as in modern mainboards, but all the main timings are present and you cannot get confused:
- 1T/2T Memory Timing: Auto, 1T, 2T;
- CAS# Latency: Auto, 2.0, 2.5, 3.0;
- RAS# to CAS delay: Auto, 2 – 7;
- Min RAS# active time: Auto, 5 – 15;
- ROW precharge time: Auto, 2 – 6.
The SmartDoc Anti-Burn Shield section allows controlling all the basic system health parameters: temperature, voltages, fan speeds. You can also specify the threshold after which the fan speeds will start to grow up. The possible threshold range is 45-65°C for the CPU and 30-50°C for the system, stepping 5°C.
Alas, I couldn’t set up the Motherboard Monitor program to control the system from Windows, so I used Soltek’s exclusive Hardware Monitor:
When the system was idle, the fan speeds didn’t exceed 1500rpm, and the temperature was around 40°C. The computer was practically noiseless. Under a higher load, the fan speeds increased to the maximum of 2900rpm. The noise was audible then. The temperature reached its highest peak of 72°C when I had been frying the CPU up with the Prime95 utility. During game tests, the CPU temperature never exceeded 70°C, but was above 60°C.
The first unpleasant surprise was found soon after the start of the tests: the mainboard didn’t respond to any adjustments of the memory timings. According to the CPU-Z utility, the timings were always set to 2-4-4-8, irrespective of the memory frequency or the user-defined values. My checking with SiSoftware Sandra 2004 SP2 version 9133 confirmed that the memory speed was always the same, whatever memory timings you select in the BIOS.
The timings are most important for the memory controller integrated into the CPU die, but that’s not the only piece of bad news. CPU overclocking proved to be impossible with this mainboard. The system booted up normally at 250MHz base frequency and x9 multiplier, but it was not stable in this mode as well as at 240MHz. I didn’t try to reduce the clock generator frequency further since the resulting CPU frequency bonus would have been too small.
With its standard x10 multiplier the processor I used in the tests is known to be stable at 220MHz base frequency even with the nominal voltage, but now it could only start up and I had to drop the frequency to 210MHz to boot up Windows. Overclocking to 2200MHz is not very much and finding a miserable speed bonus from extra 100MHz makes no sense.
Soltek’s website offers an experimental version of the BIOS that allows increasing the clock generator frequency to 300MHz, but I had no success with it either and returned to the BIOS version 1.5. This overclocking failure is very strange since the system turned to be operable at 250MHz. Reducing the multiplier to x8, I made it pass all the tests, including the burning with Prime95. I used this to compare the system’s performance in this “overclocked” mode and at its regular frequencies.
Systems with K8 family processors have no front-side bus proper and cannot be said to work with memory synchronously or asynchronously. In both cases, the CPU worked at its rated frequency of 2000MHz (200*10 and 250*8). In both cases, the memory actually worked as DDR400 SDRAM at 200MHz and the memory devisor was the same (10), since in the second case I identified the memory as DDR333, but the CPU and the memory were simultaneously overclocked.
The only difference between the two system modes was the frequency of the HyperTransport bus: 1GHz at “overclocking” against 800MHz in the regular mode. I had thought that the speeds should be the same in both cases, but this was not so.