Ballistix Sport LT Overclocking

Prob old news to some but I noticed my board overvolts Dram by about 0.024-0.036 idle in bios, I had a search and found this was common and that under load that can increase from 0.044 to 0.056, so if you using 1.45 in bios you maybe pushing 1.56v. Ryzen master only reports your settings, not your actual voltage. I dialed mine back to 1.38 @3600 and its fine.
 
Last edited:
staver said:
Honestly, if I was just gaming I would probably do the same! Since this is a work and gaming rig I am a little more particular about stability so nothing important gets corrupted.
Click to expand...
Yeah, any backups won't help if there was logical level corruption in data in the first place.


staver said:
Hi folks, first off I wanted to say a big thank you to all the contributors in this thread. I have been fighting with some G.skill Trident Z Neo with my new PC for almost 2 months and was getting depressed with not even getting it stable at DOCP settings.
Click to expand...
G.Skill has also Hynix chips and other various chips in Trident serie.
So might have been some of the worser DRAM chips which don't play well with Ryzen or would need some special settings.
 
MegaPinches said:
Prob old news to some but I noticed my board overvolts Dram by about 0.024-0.036 idle in bios, I had a search and found this was common and that under load that can increase from 0.044 to 0.056, so if you using 1.45 in bios you maybe pushing 1.56v. Ryzen master only reports your settings, not your actual voltage. I dialed mine back to 1.38 @3600 and its fine.
Click to expand...

Gigabyte board?
 
michty_me said:
Gigabyte board?
Click to expand...

Yep

......so its gigbyte boards, makes sense now....


must do this with vcore too then, I knew this board was running ever so slightly hotter at load than my previous board (even though i bought much better paste and my idle temps are way lower) and producing a slightly lower R20 score due to throttling.
 
Last edited:
MegaPinches said:
Yep

......so its gigbyte boards, makes sense now....


must do this with vcore too then, I knew this board was running ever so slightly hotter at load than my previous board (even though i bought much better paste and my idle temps are way lower) and producing a slightly lower R20 score due to throttling.
Click to expand...

Yeah, was away to say. The only boards, I had read other posts of it happening were from other gigabyte owners.
Guess you have to be extra careful when tweaking voltages then!!
 
Well, I think my OC is stable now, made it through 50,000% coverage (24+ hours) in Karhus. Are there any other stability tests I should consider just to make sure everything is good?
EsaT said:
Yeah, any backups won't help if there was logical level corruption in data in the first place.
G.Skill has also Hynix chips and other various chips in Trident serie.
So might have been some of the worser DRAM chips which don't play well with Ryzen or would need some special settings.
Click to expand...
My kit had Hynix DJR according to Thaiphoon Burner and I could just never get it working right. OCing the Ballistix worked a lot better. Part of that is probably because I have an x470 board and my G.skill kit was a really late add to the QVL.
MegaPinches said:
Prob old news to some but I noticed my board overvolts Dram by about 0.024-0.036 idle in bios, I had a search and found this was common and that under load that can increase from 0.044 to 0.056, so if you using 1.45 in bios you maybe pushing 1.56v. Ryzen master only reports your settings, not your actual voltage. I dialed mine back to 1.38 @3600 and its fine.
Click to expand...
Did this show up in HWInfo voltages or just in your bios? I am curious if I might have a similar issue since I am running some voltages on the higher end of "safe".
 
staver said:
Well, I think my OC is stable now, made it through 50,000% coverage (24+ hours) in Karhus.

My kit had Hynix DJR according to Thaiphoon Burner and I could just never get it working right. Part of that is probably because I have an x470 board and my G.skill kit was a really late add to the QVL.
Click to expand...
Well, trying this test certainly won't hurt:
https://www.techpowerup.com/memtest64/

Myself very interested of 2x16GB kits so still running those settings you mentioned previously?
(tFAW 16 and tRFC 600)
Ryzen DRAM calculator keeps tRFC relatively close to 600 even at fast settings, so E-die likely needs that high unlike Samsung B-dies.


Ryzen DRAM Calculator doesn't even know those DJRs, so likely some of the worst possible chips and no wonder if they barely worked.
They aren't even mentioned here:
https://www.hardwareluxx.de/communi...lle-hersteller-12-11-19-a-1161530.html#lstH.B
(You remember the specific product code?)
 
EsaT said:
Well, trying this test certainly won't hurt:
https://www.techpowerup.com/memtest64/

Myself very interested of 2x16GB kits so still running those settings you mentioned previously?
(tFAW 16 and tRFC 600)
Ryzen DRAM calculator keeps tRFC relatively close to 600 even at fast settings, so E-die likely needs that high unlike Samsung B-dies.


Ryzen DRAM Calculator doesn't even know those DJRs, so likely some of the worst possible chips and no wonder if they barely worked.
They aren't even mentioned here:
https://www.hardwareluxx.de/communi...lle-hersteller-12-11-19-a-1161530.html#lstH.B
(You remember the specific product code?)
Click to expand...
Yup, tFAW at 16 and tRFC at 600 made it over 24 hours in Karhu RAM Test with no errors. Really happy with how well this kit turned out and thankful for the settings from @AndreiD

For my old kit, the product code was F4-3600C16D-32GTZNC which I found almost no one else using. I did find a spreadsheet that someone referenced it as CJR but Thaiphoon burner said DJR when I used it for DRAM calc. It was never close to stable at DOCP settings or even the safe settings of DRAM calc when I set it to CJR.
 
Happy you got it working, I've been very impressed with Micron E-die too, it's just excellent for the price these kits go for.
I ended up dropping mine to 3600Mhz because 3733Mhz needed me to push VSOC/VDDG/VDDP up quite a bit. The RAM can do it, but I just got unlucky with my R5 3600.
 
My findings for noobs like me, based on my sticks (2x8Gb) used on 2 boards after faffing about with these last few weeks:

Dont bother with anything other than primaries and tRC + tRFC, you will only get micro gains, if any at all and sometimes worse and usually end up spending hours trying to nail down which one is giving you errors. Yea memtest will run a bit faster but so what if you getting errors.

Let your motherboard sort out all your CAD BUS, Proc ODT and Rtt settings (unless you can't post at all), yea the Dram calculator settings will get you booting cl14 mega speeds etc if you change all these but you probably going to have errors unless you got amazing sticks anyway.

Any tRFC under 600 will eventually give errors at any high speed 3600+ (not tried lower speeds), tRRDS set to 1 will give errors always, use 3.

Power Down Mode OFF

If you get a hang, no post then bios reset after changing a lot of settings, don't give up, reload old settings then change again and see if it works. This happens sometimes even when changing to known good settings when certain settings are changed.

When overclocking infinity fabric for 3733+ use 1.05v VDDG, it will pass memtest fine at lower voltage but when you try to play a game audio will be screwed up, this is common. 1.05v is actually the default but I use less for 3600 and Dram calculator recommends even less than that. This probably down to the CPU I guess?, I have a R5 3600, some better CPU's might have better IF that runs at lower volts.

Check actual voltage using HWinfo as some baords over volt ram, in my case by 0.024-0.036 so I use 1.38v for 1.4v. Ryzen master will only show you what you input not actual.

Settings that absolutely work for me, 10+ memtest plus hours gaming.

3600 16-19(19 tRCDRD + 16 tRCDWR)-16-36-62-620, Soc 1.1, VDDP 0.9v, VDDG 1.0v, 1.40v Dram, Spread spectrum off.

3733 16-19(19 tRCDRD + 16 tRCDWR)-16-36-62-620, Soc 1.1, VDDP 0.9v, VDDG 1.05v, 1.42v Dram, Spread spectrum off.

I imagine these Dram Voltages can be tweaked down a bit too.

Ive benched the 3600 profile against the Safe 3600 Dram calculator profile (that has much tighter secondaries downwards), there's no difference in bandwidth or latency.

My 3733 vs the 3733 Fast Dram calculator settings is fractionally lower bandwidth and same latency but the Dram Calculator settings will give errors for me.

I'm just a noob with Ryzen builds but hope this helps some save a lot of hassle chasing holy grail settings that are not worth it.
 
Last edited:
I Cant get mine to post over 3500 no matter the timings

3500 is Memtest 10 Pass Stable 16-18-18-72-560 CR1

Setting Timings back to auto = 18-20-20-595 CR2 3600 wont post
VCCIO auto = super high wont post
System Agent auto = super high wont post
Dynamic Vcore (DVID) +0.000 = wont post

1.30 Volts 3500 16-18-18-560 CR1 = Memtest86 10 Pass Stable

No Matter what i do i cant seem to get it to post at 3600
well i have twice but it was super flakey and took around an hour

It doesnt seem to be a ram limit
Looks like a Motherboard/IMC problem

anyone have any idea's
also if anyone has any easy sub timings that i missed at the current 3500
(I just set them back to the 3200 timings as my bios auto loosened them off when i set 3500)


Bios setup and pc info here
https://postimg.cc/gallery/1hx2n7jvo/
https://valid.x86.fr/1sxa5b
https://www.userbenchmark.com/UserRun/21655032
 
Last edited:
Crazy how it varies wildly from ram to ram even if exact same kit. The ram lottery is even worse than cpu lottery these days it seems.
 
BLS2K16G4D32AESB, 32Gb 2*16Gb, 3200MHz RAM Speed, CAS 16-18-18-36 Timings, 1.30-1.35v VDIMM

Can these reliably do the above timings at 3600MHz with around 1.38v on Ryzen 3000 with X570 or is it hit and miss? I'm unsure if to future proof a new system now with this kind of kit, or if to just by 2*8gb bdie for ease and change out in a year or two.
 
Drollic said:
Can these reliably do the above timings at 3600MHz with around 1.38v on Ryzen 3000 with X570 or is it hit and miss?
Click to expand...
overclocking is never guaranteed lol
but in general most will be able to. 3600c16 should be around 1.4v rather than 1.38v - not that the 0.02v makes a large difference in heat output
 
You must log in or register to reply here.
Back
Top Bottom