I want MORE !! MOAARRR :-D

[MuHcOw]

It's fine. Higher frequency = higher bandwidth, but you can see 7ns lower latency. If you check my screenshots then you see that you have about the same bandwidth, but additional timings can give you 2-3ns lower latency. It can be also a matter of the software in the background or something else.
You can play with additional timings, but if it's stable then I would leave it like that as you won't make it much faster. As I said, maybe 2-3ns latency or maybe 1GB/s higher bandwidth.

Do I still test with Prime95 (Large FFTs) and if so, for how long?

I'm pretty sure the following is going to go way over my head, but...
Let's say I wanted to try and push things a bit further...

If I upped the voltage level and had better RAM cooling (no, I'm not planning on water cooling my DIMMs) - Would that enable me to do some extra tweaking??

BTW - Testing this one now:


[EDIT] - Didn't really improve anything that I can see... I probably missed some important detail, again...

 

[Woomack]

The performance won't be so much different at both settings. If you rerun the test, then it will give you a bit better results. Usually it's +/- 1GB/s and +/- 2ns. As I mentioned before, additional software in the background or some Windows services may slow it down a bit, so maybe it's faster, but you can't see it.
This RAM is scaling up to 1.6V, but at 3600-4000 you can't make more as it won't run at lower CL. The highest kits had 1.50-1.55V XMP profiles and over 5000+ clock, and were not sold with additional fans. Crucial would add it if it was required. I had no problems at 1.50V for 24/7 use, and I guess you won't really need a fan if you stay at 1.5V. Micron IC runs fine at higher temps than Samsung or Hynix, but of course, there are some limits. I guess you have to check it yourself, but as I said, I don't think you will need higher voltages.

 

[MuHcOw]

Dang - Current settings not stable after all

Had a long play session the other day where the game I played crashed a few times (not unusual for Star Citizen) - But I decided to run another Prime95 test just in case - This time I just let it run for longer...

I am now back where I started

Pretty sure I'll ditch two DIMMs after I'm done with my video project in a few months.

Thanks for ALL the time you guys put into this!!!!!

I WILL be back !!

 

[Zerileous]

eh just try going back to tRP 17. You're chasing fractions of a fraction of a percent at this point.

 

[MuHcOw]

Guess who's back in town

I just wanted to share the following screenshots:


Std. XMP - Everything set to AUTO (1:2 IF (or is it 2:1?)).
I forgot to run this test the very first time around.
4x 16GB sticks




After fiddling with some Voltages (1.4V) and a few timings etc. I ended up here:




Well - Latency went down a bit but so did a lot of other stuff
I may have been able to go ONE step further up the ladder from 1833MHz (I tried TWO steps and that crashed my system at 1.4V).

Once again, I'm probably missing something.
But it appears that Read and Write speeds really took a plunge here.


PS. Yes - I have now bought my very own legally licensed version of Aida64 (Christmas sale).