Two DIMMs missing info in SPD tab in CPU-Z

[badreg]

I have 4x8GB of RAM install in my system, but two of the DIMMs are missing information in CPU-Z. They came in 2x8GB kits, and I purchased them separately, but the packaging and part number on the DIMMs are identical. The max bandwidth is also different, although the XMP profiles are identical. In BIOS, two DIMMs say "PNY DDR4 8192MB 3200Mhz", while the other two just say "DDR4 8192MB 3200Mhz".

 

Did two of the DIMMs just not get the correct JEDEC information encoded, or were these kits sourced from different vendors and are different from each other?

[ItsTheDuckAgain]

well they obviously are different :-)

since you bought them separately, even when the box got the same branding, chances are pretty high that different silicon and controllers have been used. Basically RAM manufacturers make those sticks with everything available they can get there hands on. there is a reason why 4x8GB kits exist.

wouldn't worry too much, I guess you will never feel the difference.

[NelizMastr]

They have different timings and they have a different layout (single vs. dual rank) so no way these are the same kits. You'll get in trouble with memory overclocking, but you should otherwise be fine.

[badreg]

Yeah, I just noticed that two sticks are 16-18-18-36 and the other two are 16-18-18-38. The stickers on all four sticks say 16-18-18-38, IIRC. I don't plan to overclock beyond the XMP settings, and I haven't had any problems with stability so far.

 

If each stick has different timings, what does the BIOS do when XMP is enabled? My BIOS only shows one XMP profile.

 

In case anyone is curious, these are PNY Anarchy-X 3200Mhz, which are only sold in 2x8GB kits. The packaging and appearance of the heat spreaders are identical for both kits. The stickers on each stick show the same part number and timings, but a different PO number. I guess that these were sourced from different places.

[badreg]

An update on this:

 

After running stable for a few weeks, I have been getting some random lock ups in the past few days. The mouse cursor will freeze and my system will be completely unresponsive. This happens when the system has just started up, or has been running for a few hours, and even when there is no load at all.

 

I couldn't replicate the problem when stress testing with either AIDA64 or Prime95 so I was fairly confident that my CPU overclock is stable.

 

My next step was running memtest86 to test RAM stability, and this is where I noticed problems. With XMP enabled, running at 3200Mhz 16-18-18-38, memtest86 either locks up or splits out thousands of errors on one specific test (Test 6 Block Move). Every other test passes without an error. With XMP turned off, it passes Test 6. I tried XMP on, but dialing back the speed to 3000Mhz, and this also passed.

 

There are two differences in the XMP profiles between the two different kits of RAM, according to CPU-Z. One kit has a tRAS of 38 and a tRC of 56, while the other has a tRAS of 36 and a tRC of 63. My BIOS has no option to change the tRC, but I tried changing the tRAS to 39 and 40. This seemed to reduce the amount of lock ups and errors in memtest86, but did not completely make them go away.

 

Next, I tried each kit individually with their own XMP profiles, and both kits passed without an error.

 

Finally, I rotated the DIMMs, so that I can apply the XMP profile from the second kit (3200Mhz 16-18-18-36). Predictably, this failed, but once I changed the tRAS to 38, the errors went away completely. My guess is that the tRC is still being applied, even though my BIOS is not showing it to me. Both kits are capable of 38 tRAS, but only one kit is able to do 56 tRC.

 

Hopefully, this will be the end of the random lock ups.