Operational Specifics in Nominal Mode

There is only one thing you can usually say about the mainboard’s operation in nominal mode: it works. And how could it be otherwise, as it is the most natural mode. Any operational mainboard should work flawlessly in nominal mode. However, no matter how natural this may seem, this statement underwent severe pressure from ASRock P55 Deluxe. The board is operational, but we did encounter some issues particularly in the nominal mode. And unfortunately, these were not just one or two flaws, but a number of things.

Let’s start with the fact that if you disable the startup logo, you will see the progress of the POST procedure on the screen. In particular, we are told that Intel Core i7-860 processor works at its nominal frequency of 2.8 GHz. In fact, this is not true and the actual frequency is higher not only because of the active Intel Turbo Boost technology that increases the CPU frequency when the workload appears.

But let’s not anticipate things. We boot the OS, launch Everest utility and see that all Intel processor power-saving technologies are up and running. In idle mode the CPU lowers its voltage and frequency. Only the reported voltage is somewhat lower, it is usually about 0.1 V higher. Is it a monitoring mistake? No it is not. The thing is that we enabled ASRock’s brand name power-saving technologies in the BIOS and in this case not only the number of active phases in the processor voltage regulator circuitry changes, but also the processor core voltage may be lowered if necessary.

In fact, it is a pretty popular approach used to save some extra power. Power-saving utilities of Asus or Gigabyte mainboards allow selecting a mode where the processor core voltage will be below its nominal value. As a rule, this doesn’t affect the processor stability in any way, but allows lowering its power consumption and temperature. However, we make this choice knowingly and deliberately, by default Asus and Gigabyte mainboards, unlike ASRock, send the nominal voltage to the CPU. However, we don’t have this choice here. “Intelligent Energy Saver” parameter only allows us to enable or disable power-saving, but doesn’t allow managing the modes in any way.  Either power-saving is off, or the processor Vcore is being lowered – these aren’t the alternatives I would want to choose from. However, in the end we have to agree to the lowered CPU Vcore.

As we proceeded with testing of ASRock P55 Deluxe mainboard, we decided to load the CPU to the utmost extent. Of course, Intel Turbo Boost technology increases the processor clock multiplier to 22x, but for some reason the base frequency also increases in this case. Instead of the processor’s nominal 133 MHz it jumped to 141 MHz!

However, we were not surprised with this unexpected development. We wanted our system to work in the most optimal way in the nominal mode that is why we selected “Load Performance Setup Default” in the “Exit” BIOS section. In this case “Overclock Mode” parameter is set to undocumented Optimized value. From our previous experience with ASRock mainboards we already know that in this mode the base frequency will increase under load. Well, this is a known trick used to provide the board with some advantage over the competitors. Just recently in our Gigabyte GA-P55-UD6 mainboard review we saw that that board also increased the base clock frequency to 136.7 MHz. However, we could manually adjust the frequency on Gigabyte mainboard and return it back to the nominal 133 MHz, while on ASRock mainboard it is still set at 141 MHz. You can only see the changes when you set “Overclock Mode” parameter to Auto. Only the frequency doesn’t go back to its nominal value, it is still increased, but this time “only” to 138 MHz.

Surprisingly, inability of ASRock P55 Deluxe mainboard to provide the CPU with normal core voltage is not the only issue. In addition it doesn’t allow setting the nominal values for processor and memory frequencies as well as for all other busses, which frequencies are directly connected to the base clock. I have come across mainboards that cannot ensure system stability during overclocking many times, but I could never expect to see a mainboard that can’t work in nominal mode.

Sometimes manufacturers resort to all possible and impossible tricks to ensure that their mainboard will have an advantage in allegedly identical testing conditions. ASRock seems to have turned on protection against extremely honest reviewers. No matter how hard you are trying to make sure that all mainboards work in the same fair conditions, ASRock solution will still run at a higher frequency. We do not level out the mainboards’ chances in nominal mode. Our tests show the performance provided by the mainboards with practically no user interference. Therefore, we didn’t even comment on the advantage of ASRock P55 Deluxe, as the CPU installed in it will work at 100 MHz higher frequency than the CPU in Gigabyte GA-P55-UD6 (which also sets higher clocks). As for the Asus P7P55D Deluxe working at the honest nominal speeds, the advantage of a processor in an ASRock board will hit all 170 MHz. So, you can imagine how surprised we were when the anticipated indisputable victory of ASRock P55 Deluxe didn’t really turn out that obvious anymore. Take a look:

12 % advantage in memory read speed is definitely quite impressive. And 5 % advantage in heavy duty Cinebench and Fritz tests also looks very convincing. But there are no other numbers like that. Moreover, sometimes ASRock P55 Deluxe is even falling behind, although we compared it against Asus board, and not Gigabyte that also uses higher frequencies (though not as high as ASRock anyway). How did it happen that the system working at higher CPU and memory frequencies takes longer to complete the calculation of the Pi? Strange things continued when we got to the power consumption tests and discovered that ASRock was losing again, even though it lowered the processor Vcore!

As a result, we decided to undertake a paradoxical move: disable the Optimized mode and rerun the tests with “Overclock Mode” parameter set at Auto, even though in this case the frequencies wouldn’t be increased that much. The results totally surpassed our expectations. Yes, the frequencies are lower in Auto mode that is why there is nothing strange about a 2 % lag in Cinebench and Fritz tests. However, in all other cases the performance in Auto mode is about the same or even higher than in Optimized mode!

Moreover, power consumption tests, which we are going to discuss a little later today, have also become normal. So, it is certainly logical that the final comparison was performed with the Auto mode set on ASRock mainboard instead of Optimized.

I believe that the results of our investigation of ASRock P55 Deluxe performance in the nominal mode won’t really surprise you. The board gets an F for a very weird Optimized mode implementation and actual inability to ensure stable CPU performance at nominal frequency and with nominal core voltage settings.