• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Subtiming help (Ryzen, DDR4, hynex c-die)

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

pqwoerituytruei

Member
Joined
Nov 24, 2011
I have a x470 gaming plus, a ryzen 3600, and a kit of 16gk (2x8GB) DDR4 3200 that has Hynex C die in single rank (see attached zip file w/ typhoonburner report)
I have tried using the ryzen timing calculator, but at best i can post and run memtest86+ a while, but get windows to boot :bang head (i think i made a typo so it actually posted as i have not been able to reproduce it)
using my boards try it memory i could not get 3600 stable, but i got 3466 to work, but i am using auto subtimings that are probably awful, but i have no idea what i am doing
View attachment ddr4-3466_00.bmpView attachment ddr4-3466.bmp
 

Attachments

  • G.Skill F4-3200C16-8GVKB.html.zip
    2.7 KB · Views: 24
To get 3600 Mhz?
to get the calced 3200mhz timing to work?
from what i tested it was able to do 3466 with the above timings at 1.37v, did not try 1.35, the try it feature uses 1.4v
 
Yes, I would try 3600Mhz 16-18-18-18-38 @ 1.50v and see if it's stable. If so, try and drop your dram volts to 1.45v.

Don't be surprised if you are not able to hit 3600Mhz with this kit though. Woomack reviewed a similar kit and maxed out stability at 3466Mhz.
 
Based on what i was trying i do not think i will get 3600 out of it, at 1.4v the best i got was about 3 seconds of stability testing in windows before it crashed, this was at cl 18 20 20 20, maybe even slower than that
that was why i was asking for sub timings for 3466
right now i have my new parts(spare PSU, spare GPU, hyper 212 (just till it goes in my daily build w/ a NH-D14), 470 gaming plus, ryzen 3600) in a cardboard box where i have not put them into my daily build yet, wanted to get my settings right 1st, and about 2 hours is not enough time to play with it give i have setup and teardown time to deal with, i cant just leave everything on my desk for my cats to chew on

once i get my settings right and get some benchmarks done i can compare it to my currently daily build
 
Here is the review I was referring to.

Overclocking results from his set were.

3333 16-16-16-36 1.40V

&

3466 15-18-18-36 1.50V

I would venture to guess that 3600Mhz @ 1.40v would crash at any timings.
 
I tried something similar to that the other day
with 16-16-16-36 i got stuck at 3266, not sure if i tried that at 1.4v though
i know you can put the ddr4 voltage quite high, but how high would you suggest for daily use
i'll do some more testing tomorrow morning
 
With good case air flow I'm comfortable with 1.50v.


EDIT: I feel I need to add the disclaimer that if the OC doesn't provide a "noticeable" difference than there is little to no reason to stress your system. From 3200-3600 there is an improvement, but for your particular application it may not be enough to be worth it.
 
Last edited:
I just want to get the most out of my budget ram, ryzen may not OC, but ram will; i have to OC something and my GPU has nothing left in the OC tank
I'd rather not have the voltage high enough to require using my NF-A4x10 fan on the ram, id rather have that on the VRM (airflow dead zone), my Z97 board needed that to stability test the cpu OC (with 2 hrs prime95 1344 FFTs it can't handle it)
 
It's Hynix so I wouldn't be surprised if you had to set something like CL18-20-20 or 18-22-22 to make it work at 3600. GSkill is using Hynix IC in their 3600 CL19-20-20 kits, the same as some other brands, just because this IC couldn't run stable at 3600 CL18 or lower at 1.35V in mass production. Even when I got GSkill SniperX 3600 CL19 memory for review then about a day before publishing, GSkill removed 3600 CL19 kit and replaced it with CL20.
Another thing is that X470 motherboards usually work well up to 3466 (some up to 3600 but not all). I would check something like CL16-19-19 at 3466, 1.40V and SoC 1.1V.
 
Do i need to manually define SOC to 1.1v or is auto 1.1 good enough?
CL 16-18-18-18 is better than 16-19-19-19 right?
I know i can do 3466 CL 16-18-18-18 at 1.4v w/ 1.1 soc
sadly i did not get time to mess with anything today, maybe tomorrow...
 
Do i need to manually define SOC to 1.1v or is auto 1.1 good enough? For stability I prefer manual settings
CL 16-18-18-18 is better than 16-19-19-19 right? If it's fully stable, Yes
I know i can do 3466 CL 16-18-18-18 at 1.4v w/ 1.1 soc
sadly i did not get time to mess with anything today, maybe tomorrow...
 
Yes, I would try 3600Mhz 16-18-18-18-38 @ 1.50v and see if it's stable. If so, try and drop your dram volts to 1.45v.
posted at 1.45v it pass all of 3 seconds of memtest
at 1.5v windows crashed during boot
16-20-20-20 ram memtest for almost 10 seconds
3533 16-18-18-18-38 1.45v looks like it might work (200% scope in memtest passed via ryzen ram calc), but it failed at 1.42 as for 1.43 and 1.4 they failed to get into windows
now running long test for 1.45v 3533

3533 is probably be the limit for this kit, unless we throw bclk at it and wreck everything else's stability

- - - Auto-Merged Double Post - - -

Here are the benchmark scores i got in aida64 (*scores are CPU bound, so a faster CPU will score higher with the same ram*)
  • XMP (16-18-18-18)
    • Read: 44881
    • Write: 25599
    • Latency: 76.3
  • XMP MOD (16-16-16-32)
    • Read: 45199
    • Write: 25599
    • Copy: 44144
    • Latency: 76.3
  • XMP MOD (3266)
    • Read: 45938
    • Write: 26666
    • Copy: 44351
    • Latency: 74.5
  • XMP MOD (3333)
    • Read: 46794
    • Write: 26133
    • Copy: 45075
    • Latency: 74.4
  • 3466 Try It
    • Read: 48450
    • Write: 27732
    • Copy: 45659
    • Latency: 72.8
  • 3533 16-18-18-18-38 gear down
    • Read: 49354
    • Write: 28266
    • Copy: 47832
    • Latency: 71.4
Hoping to get some sub-timing recommendations to try for 3533
*dam you screensaver, i thought it crashed*

edit:
Test Complete, it passed, i tried the tight timing suggested by the ryzen meme calc tool for 3533 at 1.5v and as i expected it would not post

Edit2:
I tried the 'safe' calced values for 3200 at 1.4v using the tRFC at 448 and tWRRD at 3 and it did post and crashed on the windows desktop
 

Attachments

  • ddr4-3533_01.png
    ddr4-3533_01.png
    51.1 KB · Views: 1,222
  • ddr4-3533_00.png
    ddr4-3533_00.png
    50.4 KB · Views: 1,219
  • ddr4-3533.png
    ddr4-3533.png
    53 KB · Views: 1,222
Last edited:
Just as an info, I'm testing new Hynix C-die kit right now. It works at 3600 15-19-19 1.35V, 3600 14-19-19 1.45V, 3800 16-19-19 1.35V, 3466 15-17-17 1.35V.
 
What was it rated for?
while i was testing 3533 sometimes it would boot sometimes it would not, but when it does boot there are no stability issues for testing, using the stability test in the ryzen calculator i tested 14/16GB to 1000% with no errors
also had some settings pass 1.5hours of bootable memtest86+ and then fail to boot windows
here are a bunch of things i have tried if you want to take a look at it

is the benchmark tool in the ryzen calculator, even that useful, i got a very diff score on 3533, not sure why the other one was so high (maybe it was windows doing something or the bios update i did today)
 

Attachments

  • Ram OC Files.7z
    450 KB · Views: 35
I'm not using ryzen calc, I never found it useful but that's me.

Memory is Patriot Blackout, 3600 CL17-19-19 1.35V, a bit weird is CL17 when it's designed for Ryzen. So far couldn't boot it at 4000+.
 
I think my biggest issue was power down mode, turned that off and now i can test 3533 16-18-18-18-38 at 1.35v

not sure if it is the ram, motherboard, or ram settings but sometimes after a reboot or i save bios settings and the thing reboots it will not post till i hit reset, sometimes it takes multiple resets or a power cycle
I feel this happens more often with questionable ram settings when i do not want to redo my settings from scratch again
edit: after rebooting from memtest86+ i had to hit reset 3 or 4 times before it would post
edit: was able to run memetest86+ at 3600 on the same timings at 1.4v, memtest86+ crashed at 1.35v

only did test for 5-10 min
 
Last edited:
This is the best i have been able to get the primary timings
more voltage did not seem to help, unless i was trying to run at 3600, even then it is only stable enough fail a stability test in the 1st minute or 2 with really slow timings
what secondary timings should i start trying lowering? (priority order)
what timings need to maintain a certain ratio?
 

Attachments

  • MSI_SnapShot_ram.png
    MSI_SnapShot_ram.png
    154.4 KB · Views: 1,099
It's Hynix so I wouldn't be surprised if you had to set something like CL18-20-20 or 18-22-22 to make it work at 3600. GSkill is using Hynix IC in their 3600 CL19-20-20 kits, the same as some other brands, just because this IC couldn't run stable at 3600 CL18 or lower at 1.35V in mass production. Even when I got GSkill SniperX 3600 CL19 memory for review then about a day before publishing, GSkill removed 3600 CL19 kit and replaced it with CL20.
Another thing is that X470 motherboards usually work well up to 3466 (some up to 3600 but not all). I would check something like CL16-19-19 at 3466, 1.40V and SoC 1.1V.

I would try the timings Woomack suggested. You're trying to run primary timings tighter than stock and take a significant frequency boost. I just don't think it's going to happen. Don't worry about the secondary timings. Auto will be stable if the speed and primary allow it.
 
I will drop a thread with Patriot/Hynix C memory so you can see how it works. Sub-timings can be at auto as it won't matter much and the motherboard should set them right (as long as BIOS is well-tuned).
 
Back