Memory Information

[zenmota]

I have two G Skill memory sticks. Model G.SKILL Ripjaws S5 Series 64GB (2 x 32GB) 288-Pin PC RAM DDR5 6800 (PC5 54400) Desktop Memory Model F5-6800J3445G32GX2-RS5K. I notice in the System Information (are any place that shows memory stats) for both BurnIn Test and Performance Test it lists the memory as PC5-38400. Shouldn't it be PC5-54400? Where is it getting PC5-38400? In HWiNFO 64 it lists memory as DDR5-6800 / PC5-54400 DDR5 SDRAM UDIMM. Have attached screenshot of both. Thanks

[David (PassMark)]

Data is collected from the RAM's SPD flash storage chip via the SMBus. If that isn't available the SMBIOS data is used.
But RAM typically has multiple profiles. Maybe the XMP profile isn't being picked up here and it is displaying the base JEDEC profile. Or maybe SMBIOS data is bad (which would be a firmware bug).

Can you run BurnInTest in debug mode and post the log files.
https://www.passmark.com/support/bitdebug.php

[zenmota]

Here you go. I zipped up all the files it made. I didn't run any test. Just opened the System Info Page. I understand what you said about the SMBIOS data maybe being bad, but HWiNFO 64 reads it correctly, so I wasn't thinking firmware bug, but you never know. Thanks

[zenmota]

When using MemTest86 Pro 11.6, when I read the SPD it shows PC5 54400 on one screen, then PC5-38400 on the next screen. Tried to get screenshot, but it didn't work. All these memory stats are a little confusing. Thanks

[zenmota]

I think maybe where it is getting the PC5-38400 is because I believe that is DDR 4800MHz which is default memory speed for ddr5 and the Z790 chipset. And it shows the memory clock at 2400MHz which with ddr that would be 4800MHz. Anyway, just a thought. Thanks....

[keith]

Thanks for the logs. It looks like the standard JEDEC timings were reported (PC5-38400), as opposed to the XMP timings (PC5-54400).
This has been fixed for the next public release, which should be available shortly.

[zenmota]

OK thanks. Keep in mind this was happening in both BurnIn Test and Performance Test.

[David (PassMark)]

We use a common code library for for hardware detection for both BurnInTest and PerformanceTest. So not a surprise that the behavior is the same.

[zenmota]

Thanks for the quick fix on both BurnIn Test and Performance Test. Especially since it wasn't a major problem.