Ballistix Sport LT Overclocking

Associate
Joined
12 Mar 2017
Posts
1,115
Location
Ireland
Should be fine as long as you use >80% of your total RAM, you'll find errors sooner or later if they're there if you leave it running long enough. Usually a couple of hours or leaving it running overnight should be good.
 
Man of Honour
Joined
26 May 2012
Posts
16,450
I'm using the windows one bundled with Dram calculator, I open up 6 instances using 2048 to 3072mb on each and leave about 500mb spare, is this the way to do it or should you completely max out the capacity?
Should be fine as long as you use >80% of your total RAM, you'll find errors sooner or later if they're there if you leave it running long enough. Usually a couple of hours or leaving it running overnight should be good.

guys, i'll give you a massive tip...

if you use ryzen dram calculator...
under the tab "membench"
if you select the membench mode as "memtest"
it will fully automate the opening of memtest and ram testing - saves the hassle of manually opening N amount of instances

:)
 
Associate
Joined
12 Mar 2017
Posts
1,115
Location
Ireland
I just use Karhu RamTest, it's paid software (10 euro), but it's the fastest memtest that I've found and it's been very reliable when it comes to finding errors. With Memtest I recall getting 1000% coverage per thread should give you a pretty good confidence level, but still recommended to leave overnight in case you have some rare errors.
 
Soldato
Joined
13 Jun 2009
Posts
6,847
So it turns out if I leave my RAM at 3600 MT/s but bump the IF to 3733 MT/s, with default VDDP, CLDO_VDDP, and CLDO_VDDG, it doesn't immediately fail an OCCT AVX2 CPU test. Lasted 6 minutes so far. I can't say for sure that it's stable but when both the RAM and IF are at 3666 MT/s or 3733 MT/s, it falls over within 10 seconds. So I think my issue is the RAM rather than the IF or VDDG: it just doesn't like being above 3600 MT/s at 1.4 V. Not sure I want to go higher than that with Micron-E die to be honest, what is everyone else using?
 
Associate
Joined
12 Mar 2017
Posts
1,115
Location
Ireland
3800/1900 not possible?
Haven't tried with 1.125v SOC, but with 1.1v 3800/1900 didn't boot for me. Might give it a shot later but I doubt my chip can do 1900 FCLK.

@michty_me 1.45v should be more than fine for 24/7 on Micron E-die. Buildzoid had a hairdryer pointed at some Crucial Ballistix and they didn't start popping errors even at 73C, Micron E-die is a lot less temperature sensitive than Samsung B-die it seems.
 
Associate
Joined
26 Oct 2019
Posts
2
Hello,

I also bought some CB 2*8GB Ram for a Ryzen 3600 computer (X570 Tuf Gaming Wifi with a Scythe Ninja 5 cpu cooler). I am a bit confused, therefore I have 3 questions:
  1. I used dram calculator and it states 3200/tCL14 as the optimal timings, but when I watch youtube people first raise mhz to try to reach atleast 3600mhz. If I alter the memoryspeed in DRAMcalculator, I get nothing because it says it is not safe. So what steps should I use to get the most optimal overclock possible?
  2. When I run the membench from DRAMcalculator it gives a lot of errors on 3600/tCL16. Should tests be errorfree in this software? Like, how many errors is acceptable rule of thumb?
  3. My second question is about voltage. When do you increase voltage and when do you stop?
Thx!

Fh6f4gn.png
 
Last edited:
Man of Honour
Joined
26 May 2012
Posts
16,450
@govie
dram calculator (in my experience) isn't that accurate for the crucial e-die sticks
if you go through this thread, multiple forumers have posted their settings...i suggest you have a look and a play with these settings (loosen them a bit and then tighten them depending on silicon lottery of course)
 
Soldato
Joined
13 Jun 2009
Posts
6,847
You need to use Thaiphoon to get stats for your sticks, then import that into DRAM Calculator, otherwise it gets a few timings wildly wrong (e.g. for me tWR and tRC).
 
Associate
Joined
12 Mar 2017
Posts
1,115
Location
Ireland
@govie try these timings & voltages for 3600Mhz C16:
tCL 16, tRCD(RD) 20, tRCD(WR) 20, tRP 16, tRAS 38, tRC 58
tRRDS 4, tRRDL, 6, tFAW 16, tWTRS 4, tWTRL 12, tWR 14, tRDRD SCL 4, tWRWR SCL 4, tRFC 540
tCWL 16, tRTP 12, tRDWR 8, tWRRD 2, tWRWR SC 1, tWRWR SD 6, tWRWR DD 6, tRDRD SC 1, tRDRD SD 4, tRDRD DD 4, tCKE 1
(tPage Auto, tRFC2 Auto, tRFC4 Auto)
ProcODT 40 Ohm, CMD2T 1T, GearDownMode Enabled, PowerDownMode Disabled
RTT, CAD Bus settings all on Auto
Voltages 1.4v VDIMM, 1.1v vSOC, 1.05v CLDO_VDDG, 0.9v/Auto for CLDO_VDDP
 
Associate
Joined
12 Mar 2017
Posts
1,115
Location
Ireland
Managed to break the 70ns barrier with 4 DIMMs, but wasn't stable with 1.45v in RamTest, probably needed 1.46~1.47v to get it stable:
wVZ1U38.png
With 2 DIMMs I reckon that would be ~66ns.
 
Associate
Joined
8 Oct 2019
Posts
6
Are the 3000Mhz (CL15) and 3200Mhz (CL16) essentially the same memory and it doesn't matter which one you buy?
I mean, the 3000MHz kit could just run at the 3200MHz with the same slightly looser timings.
 
Back
Top Bottom