Memory timings vary between identical kits

 

zir_blazer

Golden Member
Jun 6, 2013
1,239
536
136
There is a year and half difference between those dates. Manufacturers sometimes reuse the same model for slighty different products (And at worse times, totally different products). In this case, they're rather close, but not identical. This is because GSkill doesn't want to openly say what DRAM chips they use on specific models, and they also can switch them if they feel like it. That can also make the specs slighty change, since they use similar but not identical chips. Check this Thread.
 

Edgemeal

Senior member
Dec 8, 2007
211
57
101
I didn't realize they used such a wide variety of suppliers! Next time I clean out my PC I'll have to pop out a couple sticks and look em up. Thanks for the link!
 

Insert_Nickname

Diamond Member
May 6, 2012
4,971
1,695
136
Is it common for memory kits to have slightly different timings even though they are the exact same kit/model #?

They're different production batches obviously. And yes, this is common.

Manufactures only guarantee DIMM from the exact same kit work together, not DIMMs from different kits. Even ones with the same model number.

That's why DIMMs are sold in kits BTW. If you buy DIMMs individually you also don't have a guarantee that they'll work together. Though if we're talking standard speed/timings (f.x. DDR3-1600@11-11-11-30) the odds are pretty high that it'll work just fine. It only when you get into enthusiast territory there can be problems. Especially with sub-timings.

If you don't mind setting your own timings manually, its easy to fix. You just set everything to what the slowest DIMM can handle.
 

Edgemeal

Senior member
Dec 8, 2007
211
57
101
They're different production batches obviously. And yes, this is common.

Manufactures only guarantee DIMM from the exact same kit work together, not DIMMs from different kits. Even ones with the same model number.

That's why DIMMs are sold in kits BTW. If you buy DIMMs individually you also don't have a guarantee that they'll work together. Though if we're talking standard speed/timings (f.x. DDR3-1600@11-11-11-30) the odds are pretty high that it'll work just fine. It only when you get into enthusiast territory there can be problems. Especially with sub-timings.

If you don't mind setting your own timings manually, its easy to fix. You just set everything to what the slowest DIMM can handle.

Ya that's what I was going to do, for now letting the ASUS P8Z77-V-LK bios do it, and it appears its running each kit with just a few minor changes between them, I didn't even know that was possible.
 
Last edited:

Soulkeeper

Diamond Member
Nov 23, 2001
6,732
155
106
What's even stranger, if you look closely at the modules sometimes the SMD components are different values. I've noticed this with mushkin matched kits before.
 

Insert_Nickname

Diamond Member
May 6, 2012
4,971
1,695
136
Ya that's what I was going to do, for now letting the ASUS P8Z77-V-LK bios do it, and it appears its running each kit with just a few minor changes between them, I didn't even know that was possible.

Neither did I. Interesting. I've got a P8Z77-V Deluxe, if I can I'm going to have to try borrowing a couple of similar-but-not-identical kits to try this with.

Perhaps this has something to do with Intel FlexMemory?
 

Edgemeal

Senior member
Dec 8, 2007
211
57
101
Neither did I. Interesting. I've got a P8Z77-V Deluxe, if I can I'm going to have to try borrowing a couple of similar-but-not-identical kits to try this with.

Perhaps this has something to do with Intel FlexMemory?

FWIW, I took out the ram and tested each kit (2x4GB) separately in the blue-slots. It is sort of interesting how the BIOS decides which settings to use from both kits, I would of just set them to the highest values from each kit and maybe lost a little performance.

UnNNKuk.png
 
Last edited:

BonzaiDuck

Lifer
Jun 30, 2004
16,333
1,890
126
FWIW, I took out the ram and tested each kit (2x4GB) separately in the blue-slots. It is sort of interesting how the BIOS decides which settings to use from both kits, I would of just set them to the highest values from each kit and maybe lost a little performance.

UnNNKuk.png

Not much or any (performance difference) between those two kits.

I have high confidence in G.SKILL. I'd be willing to bet that you could examine the specs for the most recent kit (and more aggressive secondary timings), manually set those timings under the XMP profile setting, leave the VDIMM and VCCIO/CPU_VTT voltages alone, and pass an HCI-Memtest-64 with 1,000% coverage.

But why bother with that? One thing I've learned through e-mail exchanges with G.SKILL's tech-support: they are "encouraging" on the prospect of using a Command-Rate of 1 with all the stock settings. I've found that getting my RipJaws Z and X kits to run with CR=1 requires no voltage adjustment, or at most an increase in VCCIO/VTT (IMC voltage) of at most 60 to 80mV.

And frankly, you could probably get good results with both of the tweaks I mentioned.
 

Dasa2

Senior member
Nov 22, 2014
245
29
91
the ref cycle time 128 vs 208 is probably the biggest performance difference with those sub timings in my experience it can have a fair effect