DDR4 ram 3600Mhz cl16 team-group

[th1sm4dn3ss]

Hello,i have an asus rog strix x570-e gaming mobo and a ryzen 9 5950x cpu
gpu rtx 3090fe.
Now i bought a ram set from overklockers uk 2x16GB 3600Mhz cl16-16-16-38 @ 1.35volt
First i ran it in auto,now in docp and ran cpu-z,but the bank cycle time doesnt match???
I also ran thaiphoon burner and ryzen dram calculator but i dont know what to do,
i dial in the numbers in the overklock section of my mobo,do i set the memory on docp or manual
and than dial in the numbers?
Than u have memory settings in the bios also,i upped the voltage but did nothing so please
where do i start??? kind regards

 

[EarthDog]

Welcome!

You can adjust it manually. DOCP, save, reboot, then go in and adjust that number. I wasn't sure if the profiles adjusted TRC or if that is a secondary timing that's on auto. That said, you won't notice a difference performance-wise, so it's good as is.

Also, feel free to crop the abundance of whitespace out of future images, lol.

EDIT: Pretty sure that's an auto setting........just checked my PC (Intel) and my TRC doesn't match either using XMP profile. I didn't bother changing it.

 

[mackerel]

On the screenshots, looks like that was pasted into paint? Tip: before you paste, reduce the canvas size to something like 100x100 pixels. It'll expand to the pasted size so there isn't all that wasted space.

As for the ram, out of interest does the bios give you multiple options for DOCP? On my Intel and newer AMD systems, Asus give e.g. EXPO I and II. Profile I is what Asus thinks is best for subtimings. II is what the ram profile actually says. I you have the option try II.

 

[infinitevalence]

In windows you can use the WIN+S key to open the snipping tool and then copy paste or mark up a section of a screen shot.

As for the memory, I agree with Mackerel check in the BIOS for either a drop down or extended menu in the XPM section to see if there are more timing options. I have had sticks with 2 or 3 profiles before.

 

[Woomack]

Here is a thread/review of something similar:

FEATURED - Team Group Dark Pro 8Pack Edition 2x16GB DDR4-3600 CL16 - TDPPD416G3600HC16GBK

Team Group Dark Pro 8Pack Edition 2x16GB DDR4-3600 CL16 - TDPPD432G3600HC16GBK Today I have a bit unique memory kit as it's marked as the 8Pack Edition while the famous overclocker 8Pack is out of the scene for some longer. Either way, there is no Dark Pro memory kit listed on the Team Group...

www.overclockers.com

Maybe you will find some helpful settings.
In general, it's a Samsung B dual-rank, so there are plenty of overclocking guides for this IC around the web. Some guides/tips from 2x8GB Samsung B kits work too, Just don't expect much more than 4000. It's great RAM for 3600-3800 CL14-14-14 or 14-15-15 (depending on the voltages) 1:1 ratio on AMD.

I recommend enabling XMP/DOCP and manually adjusting whatever you need/want to change. This way, it will use safe (but faster than standard) settings from the programmed profile for every timing you have no idea how to set, and you can adjust everything else step by step.
Remember that you can save BIOS profiles (one of the last tabs in ASUS BIOS), so do that slowly ... change 1-3 settings max, test stability and move forward or back to previously saved settings.

 

[th1sm4dn3ss]

Hello,my first overklock is finally working.But when i use thaiphoon burner i got memory chips with a quality of 92%
and overklockable to 4050Mhz cl16?
In gpu-z memory controller load is alternating from 7 to 26 to 16 to 22 to 14 etc...
I want to run it @ 4000Mhz fclk is 2000 uclk,correct me if i am wrong @ 14cl but i dont know how?
No i dont have more options in the docp,only auto and manual but for intel cpu s there is xmp 2.0
I assume the die type is samsung B-die,because thaiphoon burner doesnt shows it? 8GB 1die???

But why is there in the bios an extra tab amd overklocking section,if u click that menu than u must accept terms???

 

[Woomack]

I wouldn't trust the DRAM calculator. Some guy made it as a side project and various users get various results. For a long time after releasing this app, he had only a single-rank Samsung B kit for tests and later added a Hynix kit (when on the market were already 5-6 different ICs). Results were all over the place, but clueless forum members (other forums, not OCF) trusted it as they had no idea how to set everything manually.

As I said in my last post, this memory kit is Samsung B dual-rank. I gave you a link to pretty much the same memory kit that I was testing some years ago. Team Group made these kits only with Samsung B IC.
You have barely any chance of making it run at 4000 1:1. That's because maybe 1% of CPUs/memory controllers can handle it. That's why I said that the best is to focus on 3600-3800 1:1, as it will give you not much worse results while being much easier to stabilize.
Of course, you can try 4000 1:1 ... simply set it at more relaxed timings like 18-18-18 1.45V and check if it works. Once you stabilize it, set tighter timings like 16-16-16 or 14-15-15 (CL14 will need ~1.5V.

No one can tell you how high this kit will go. You have to check it yourself. It's very random. The best bins could make 4400 (G.Skill was selling it at 4400 CL17-18-18). I had kits that could make 4100 max, and one that could go up to 4500.
You can set 4000+ 1:1 on APUs as they have better memory controllers. Only on APU, I could go up to 4400+ 1:1. On Ryzen 3000/5000, the max was ~3866 on my CPUs.
If I remember correctly, on OCF, we have literally one member who could stabilize RAM at 4000 1:1 on Ryzen 5000.

 

[th1sm4dn3ss]

So far...So good...
I didnt do a memory test so should i do it and what is a good one?

 

[Woomack]

Everything that uses a lot of RAM and performs mixed calculations is good. You can use AIDA64 stability test (memory+cache), memtest86+ (bootable from a flash drive), HCI memtest (or any option of that - since the free version uses 2048MB RAM max then you have to run multiple instances or use a 3rd party software which set that for you). Prime 95 blend test works fine too. Some need 1h, and some need 2-3h. Typically if HCI passes 100-300%, then it's already good. I run AIDA64 for reviews for about 2-3h, as it's typically enough. For quick tests, 30 minutes is more than enough as most errors are in the first 3-10 minutes. Once you tune up sub-timings, then run something longer.
You will see that not always a long RAM stability test is enough. Sometimes even if it passes, then some games may crash as calculatons can be totally different. This is why there is no one, perfect stability test.
I recommend trying how high it will boot. It's usually a 0.5-1 memory ratio above the safe/stable frequency. So if you can make it boot at 3866, then 3800 gives a high chance to be stable. In the worst case 3733.
You can browse various Samsung B-die guides around the web, but most of them are only about 8GB/single rank modules. Most sub-timings are similar, but if you see settings for 3800, then on a dual-rank kit, they will probably work at 1-2 memory ratios lower.

 

[th1sm4dn3ss]

Hello,this forum is the greatest i have ever been on.
i dont say it to mudging,but it really is because people are so helpfull
instead of the Belgian/Dutch fora where they be mean @ people instead of being helpfull...
So here is my latest oc on my ram,what should u guys do?
Leave it like this or???
The ddr4 ram is now on 1.51v @3800Mhz fclk 1800uclk correct me if i am wrong.

 

[Woomack]

It looks fine, except for tRP, which should be 15. Uclk = mem clock, so it's fine.

 

[MaddMutt]

IIRC except for benching you will not notice a difference from running 3600 over 3800. This is a setup for stability not speed correct? If you want to see your setups max capability, WE have a large area dedicated to OCing. Whichever way you go, we will try and assist you

 

[infinitevalence]

IIRC except for benching you will not notice a difference from running 3600 over 3800. This is a setup for stability not speed correct? If you want to see your setups max capability, WE have a large area dedicated to OCing. Whichever way you go, we will try and assist you

depends, if they are doing lots of video encoding, bandwidth can make a difference, or if they are using a software storage solution like ZFS RAM bandwidth again can make a performance difference that is noticeable.

I would say its more important to match the CPU bus speed with the lowest stable timings possible in most cases with AMD.

 

[th1sm4dn3ss]

Hello,thanks but i want it 2 oc it all the way so i now ocd my gpu and

put my score on port royal.

 

[th1sm4dn3ss]

Are there people on that site???

 

[EarthDog]

Are there people on that site???

What site???

 

[Woomack]

What people???

 

[Janus67]

 

[th1sm4dn3ss]

Unigine benchmarks?

 

[EarthDog]

That website? Those people?

I have no idea, lol.