Kernel Message re: ACPI-Safe quality

[Constantin]

Today I got a puzzling kernel log message from my FreeNAS after recently upgrading its RAM to 64GB using UDIMMs on the approved hardware list. As for activity, the pool was being backed up (so far ~1,5TB) and the load was around 1.

First, the kernel log noted two different memories (I guess expected) and it then noted that the ACPI-Safe frequency is 3579545, with a quality of 850 instead of the expected 1000 (like the TSC-low).

Anyone have an idea what the issue is?

 

[m0nkey_]

Can you paste the output from the log using the [ code ] tags?

 

[Chris Moore]

Anyone have an idea what the issue is?

You said that you upgraded the memory. Is the new memory exactly the same as the old memory? It could be that the timing of the new modules and the timing of the old modules is not exactly the same.

 

[Constantin]

This is what I was sent last night:

Code:

local kernel log messages:
CPU: Intel(R) Atom(TM) CPU  C2750  @ 2.41GHz (2416.72-MHz K8-class CPU)
real memory  = 68451041280 (65280 MB)
avail memory = 64296955904 (61318 MB)
Timecounter "ACPI-safe" frequency 3579545 Hz quality 850
ugen0.5: <vendor 0x046b product 0xff01> at usbus0
uhub2: <vendor 0x046b product 0xff01, class 9/0, rev 2.00/1.00, addr 5> on usbus0
SMP: AP CPU #2 Launched!
SMP: AP CPU #4 Launched!
SMP: AP CPU #6 Launched!
SMP: AP CPU #7 Launched!
Timecounter "TSC-low" frequency 1208361284 Hz quality 1000
-- End of security output --

However, I suspect you want the whole kernel log? I will be able to send that later tonight when I return home. I hope it's easy to find!

Cheers and thanks! Constantin

 

[Constantin]

You said that you upgraded the memory. Is the new memory exactly the same as the old memory? It could be that the timing of the new modules and the timing of the old modules is not exactly the same.

That may very well be the issue. The new memory modules are Model # IMM2G72D3DUD8AG-B15E, which are on the approved hardware list. Do I have to change something in the BIOS to accept it? The old memory was the OEM-supplied Samsung sticks.

 

[Chris Moore]

That may very well be the issue. The new memory modules are Model # IMM2G72D3DUD8AG-B15E, which are on the approved hardware list. Do I have to change something in the BIOS to accept it? The old memory was the OEM-supplied Samsung sticks.

I am not familiar with the settings in that system BIOS but I have seen systems where there was a setting to accommodate memory that wasn't perfectly matched. It may just be an informational message more so than a fault. The 'safe' speed is meant to be more compatible to ensure system stability where the 'fast' speed gives you a little more performance.. You shouldn't see a huge difference.
If you were able to reboot the system a few times, you could try the new memory alone to see if the message is generated. Then try the OEM memory alone to see if the message happens and then try both sets of memory again to see if the message still persists.
It is a good possibility that the new memory has a faster timing than the original memory due to improvements in technology over time.

 

[Constantin]

Thanks for the help! Currently, the only memory in the system is the new stuff, hence 64GB, the maximum that the ASRock C2759DI board supports. The OEM memory never triggered this message to the best of my recollection.

I'll have a look tonight inside the BIOS to see if the Northbridge settings include the DRAM frequency being set to [AUTO] (see page 43 of the C2750 user manual) as well as the DRAM voltage being set to [AUTO] as well (see page 37).

I'd expect that to be the best place to start, no?

 

[Constantin]

Both DRAM frequency and voltage were set to AUTO. The BIOS didn't give me the necessary insights into what choices the motherboard made. FWIW, I chose 1.35V as the DRAM voltage and that seems to be working so far. What specific kernel file is the one to post here? Is it var/log/messages with a grep to isolate the entries of interest?