Intel BCLK Detection & RAM Speed Issue

[chinobino]

I have an i7 3770K in an Asrock Z77 Fatal1ty Pro motherboard, running 8 GB of DDR3-2400 RAM at DDR3-2200 per the UEFI.

HWiNFO64 v3.96-1660 shows my CPU Base Clock on the sensors page as 100 MHz and my RAM at 1100.0 (2200.0 MHz DDR3) and never deviates.

HWiNFO64 v4.16-1900 & v4.17-1912 shows my CPU Base Clock ranging from 98.4 MHz to 99.2 MHz and my RAM ranging from 1073.6 MHz (2147.2 MHz DDR3) to 1098.3 MHz (2196.6 MHz DDR3).

Did something change in the way base clock is detected?

I have observed this with CPU-Z also.

CPU-Z v1.61.5 and earlier show my base clock as 100 MHz and my RAM as 1100 MHz (2200 MHZ DDR3) without deviation, every version after this shows the base clock fluctuating in a range from 97.96 to 98.26 MHz, and my RAM from 1077.6 MHz (2155.2 MHz DDR3) to 1081.4 MHz (2162.8 MHz DDR3).

Being that the newer versions of HWiNFO and CPU-Z do not show the same value it has become extremely difficult to know what frequency my RAM is actually running at (as small changes in base clock effect the RAM speed significantly).

Should I just assume that the latest version of HWiNFO is correct and that my RAM frequency is constantly changing in frequency, anywhere from 2147.2 MHz DDR3 to 2196.6 MHz DDR3?

CPU-Z shows a smaller range, from 2155.2 MHz DDR3 to 2162.8 MHz DDR3, which is much lower than the 2200 MHz set in the UEFI... Please help!

 

[Martin]

Yes, newer versions of HWiNFO use a different method to measure and report CPU clock. Since RAM clock is based on the same source (BCLK) it affects this too.
The way how HWiNFO (and probably CPU-Z too) work is to make performance measurement per every cycle. Unfortunately there's no better method how to measure the exact and precise BCLK speed, so that's why you see small fluctuations. In fact BCLK and RAM run at a fixed clock unless you change that clock manually. So the real clock is a kind of average value of the values shown. The values of BCLK shown (98.4 - 99.2 MHz) are indeed a bit lower than expected, are you maybe running some CPU intensive tasks in background that could have influence on measurements made by HWiNFO or CPU-Z ?

 

[chinobino]

Hi Martin, thanks for the response.

My computer was idling (no running tasks) when I took those measurements, it just seems that my motherboard clocks quite a bit lower than it should.

I may be able to increase the base clock in the UEFI to compensate, if I add 1.2 MHz to the base clock it should bring the base clock range up to 99.6 - 100.4 MHz, (the average and median would then be 100 MHz).

I wonder what Intel has specified as acceptable base clock tolerance for motherboard manufacturers.

[Edit] I just set the base clock to 101.2 and it read as 101.2 in the latest versions of both HWiNFO64 and CPU-Z.

Then I set the base clock back to 100.0 and now it is reading as 100.0 MHz with no fluctuation in either HWiNFO64 or CPU-Z - changing the base clock and saving to CMOS seems to have fixed the issue.

At this point I have to assume something was corrupt in CMOS or the previous settings did not save correctly?

 

[Martin]

I'm sorry, I don't know why it was so off.. You might also verify the measured BCLK with ICC values displayed in HWiNFO if it matches.

 

[chinobino]

I'm sorry, I don't know why it was so off.. You might also verify the measured BCLK with ICC values displayed in HWiNFO if it matches.

I noticed those values, and originally they weren't matching up but now they are.

Thanks for your input and time.