Overclock.net banner
1 - 2 of 2 Posts

·
Registered
Joined
·
1 Posts
Discussion Starter · #1 · (Edited)
For starters, here is an image with unaltered XMP AIDA64 results for my G.SKILL F3-2400C10D-16GTX kit on the left, and unaltered 1T command rate XMP AIDA64 results for the Avexir 2666 kit on the right:

2519305


Quite obviously, the read / write / copy bandwidth for the Avexir kit is unacceptably / abnormally low there relative to the frequency it's running at.

Next, here's an image of the Avexir kit running the XMP profile at a 2T command rate with no manual adjustments other than setting tRDRD and tWRWR to 4 (down from the much higher defaults my motherboard puts them at).

2519307


As you can see, those two timings have an absolutely enormous impact on bandwidth, and clearly need to be at no higher than 4 for the kit to actually be worth using at all.

Only problem is, the configuration shown in the second image consistently fails just about every memory test I run it through. So my question is, what would be the best steps to take in order to get everything stable while maintaining the proper (or perhaps even slightly better) bandwidth levels?

Thanks a lot for any advice anyone has.
 

·
Collector
Joined
·
149 Posts
Stabilizing those two settings and CR 1T can require adjustments to other timings, voltages, slopes etc. It might also not always be possible, depending upon the capabilities of your hardware e.g. motherboard, processor/memory controller and lastly the memory itself.

Font Material property Screenshot Electronic device Technology


Two OEM Samsung modules with full manual settings. The performance difference between 1T and 2T is not that big, so focus on the other timings first.
 
1 - 2 of 2 Posts
Top