Team Group Dark Pro "8 Pack Edition" 16GB (2x8GB) DDR4 PC4-25600C14 3200MHz Dual Channel Kit - Black

Try changing the 31 to 36 or 38. Try tRC at 60 and tRFC at 333 also.

Also worth trying 2T and geardown disabled just whilst troubleshooting.

Does seem odd though either a very weak IMC or unusually poor bin of memory.

Can also be worth testing with CPU at stock.
 
I'm also having trouble reaching 3200 with this memory - this is on a B350-F Strix gaming board and 1700x at stock (currently). If I use the DOCP profile I get boot loops until it gives up and defaults back to 2133Mhz, even with SOC 1.25v and the mem at 1.4v. If I try Ste's suggestions above it will boot about 1 in 3 times. I'm guessing 1.45v is the next thing to try? Any advice would be welcome, as 2666 seems to be the highest I can boot reliably so far.

Edit: Meant to add, running the latest 0805 BIOS, and 1.45v has not helped either
 
Last edited:
For the 8 pack 3200 I just used docp and it ran fine at 3200 c14. Changed divider to 3333 also fine at c14. Needed c16 timings for 3466 and boots to windows at 3600 but not stable under load. This is on ch6 1403 bios. I am guessing you have it but make sure you have the latest bios.

I would try saving settings and then setting motherboard back to default settings so you can just test the memory. There is a review on the internet for the ram which has timings in it as well. If not possibly it is faulty.
 
Try changing tRRD_S to 12 and tFAW to 48, it should work fine. I found that auto defaults those to very aggressive subtimings which will result in errors.
I don't think you need to lower tRC that much either, 52 should be fine on those Samsung ICs. ProcODT at 53.3Ohm (default) is also fine for Samsung B-die.

(I have the same board, X370 Prime, with a Team Dark Pro kit, had the same issue until I looked at subtimings)
 
AndreiD said:
Try changing tRRD_S to 12 and tFAW to 48, it should work fine. I found that auto defaults those to very aggressive subtimings which will result in errors.
I don't think you need to lower tRC that much either, 52 should be fine on those Samsung ICs. ProcODT at 53.3Ohm (default) is also fine for Samsung B-die.

(I have the same board, X370 Prime, with a Team Dark Pro kit, had the same issue until I looked at subtimings)
Click to expand...

Giving that a go. Thanks. :)

So what do you have soc and voltage at for the ram?

Have you got a ryzen checker screen shot?
 
Try a lower DRAM speed just to check. Some CPU cant reach 3200 apparently although I never tested one like this. But end users on here have reported it. Maybe try one stick at a time too see if one stick is mem test stable.
 
8PACK - ok after some more pocking in the bios I have my 1700 at 3800Mhz, 1.275v (might be able to get it lower with time)

Ram I now can boot into windows with 3466 14-14-14-31, Gear down mode off, BankGroupSwap off. 1.475v (first try, so may go lower?)

I just played a round in Player Unknown Battlegrounds and it was fine.

Where do I go from here? How do I test memory stablilty? I am praying I can keep these settings. Feeling a bit happier now.

I have the Asrock Taichi mainboard, I think for me all this has been a massive learning curve, I last built a PC in 2011 with the 2500k, 3 button presses in bios and I was 12 hour P95 stable, job done. This AM4 setup is very finerky to setup.

I think uping the volts have helped massively and using XMP properly and only changing those settings and not a mix of Advanced settings as I read today that bit of the BIOs is buggy.
 
If its stable no need to go lower.

To test stability for your applications just use the PC as you wold normally and tune from that. Synthetics like Realbench stability, Aida stability or HCI memtest can help. But the best way is to just use it.

So at 1.275 it can do 3.8 but no 3.9 at 1.45?? Really strange scaling.
 
I can't get stable currently, and am now trying the SPD settings. I built two machines over the weekend with 1700X, B350-F and this memory and both behave the same, I'm beginning to think there's a compatibility issue with the memory and boards. None of voltages, timings, command rates, nor gear down seem to make any difference. I always get random boot failures or black screens in windows when running mixed workloads. Stress testing tends to last longer than a mixed workload. These are both on clean installs of Win 10 with both 0805 and 0806 BIOSes. Everything else is at stock, and all other components were carried over from our previous builds.

8 Pack, is the B350-F one of the boards you've tested?

Edit: now back after another black screen whilst just browsing, that was at stock settings and running the memory at SPD 2133. Time to go back to my previous rock solid Z97 builds I think :(

Edit2: A little progress... https://forums.overclockers.co.uk/posts/31036827/
 
Last edited:
8pack

My 3200 14 14 14 31 kit is fully memtest stable at stock.

Left memtest running on 3333 14 14 14 31 at 1.475 volt. It was over an hour in on second pass with no errors when I left for work.

Last night I tested it but left doc volt auto and I had one error after like 40 mins. I am assuming that the volt boost on soc will stablize it. Is this the likely issue? This morning I set it to 1.1v and it's gone further through the memtest without an error.

If it's stable what do I do next?

Do I try to reduce the volt and test each time?
 
Last edited:
It seems the black screens were unrelated to the memory and actually caused by running CPU-Z.

I now have one machine running at 3200 via XMP with SOC at 1.15. However the other won't seem to reach 3200 stably even with 1.45v and SOC at 1.2 and a range of ProcODT values(boot loops and mem test failures). Currently validating the sticks at 2133 before trying to work upwards again, and might swap the ram kits between machines to see what happens.
 
zola25 said:
It seems the black screens were unrelated to the memory and actually caused by running CPU-Z.

I now have one machine running at 3200 via XMP with SOC at 1.15. However the other won't seem to reach 3200 stably even with 1.45v and SOC at 1.2 and a range of ProcODT values(boot loops and mem test failures). Currently validating the sticks at 2133 before trying to work upwards again, and might swap the ram kits between machines to see what happens.
Click to expand...

Having applications polling the CPU will help instigate any instability. So it's not so much that it was CPUz making the system unstable, the system likely is just unstable. Make sure to test the memory with the appropriate methods (HCI, GSAT)
 
I use this memory alongside a 1700 at 3.8 on a Crosshair 6 Hero. It booted first time for me using the XMP/DOCP profile but it doesn't run at 1.35v. Last I checked it was at 1.395v - will this cause any damage or anything like that? It's stable and temps are fine?
 
You must log in or register to reply here.
Back
Top Bottom