RAM Tuning - Help with subtimings.

[Kodehawa]

Hello there!

I have a pair of T-Force DDR4 3200MHz RAM I managed to push to 3600MHz (and to verify it's stable, which was a pain but )
Was wondering whether any of the subtimings are suboptimal, since the automatic settings on the motherboard seem to be fairly lenient on them, but I have no framework to work with.

About main timings: CL 18-21-21-21-40 at 3600 is the least that will not produce errors over a 12h long memory test. Doubt it's possible to push that further (even doing 18-20-20-20-40 fails) at 1.4V (1.45V too, but 1.4V works just as well on that).
Taiphoon Burner does not tell me the RAM chip type of this specific RAM. Screenshot attached. If anything, I wonder if anyone knows what specific memory chips this uses.

Thanks you all!

 

[droid-I]

^ Interesting. Yesterday ordered one pair of Team Group 16GB (2 x 8GB) T-FORCE VULCAN Z, DDR4 3200MHz, CL16 kit. Never had these before, but it'll take a while before I have chance to test them.

​

​

 

[Deleted member 185158]

Try a 2T command rate instead.

 

[Kodehawa]

Try a 2T command rate instead.

Will do. I guess I could tighten the other timings with that, with what I've read?

 

[3x0]

Since we're not sure on the chips (Micron rev E?), these are just approx guides. Start with changing one line at a time as written below.

tRC try 61 (tRC=tRAS+tRP), increase slightly if unstable. If one of those values don't follow the formula, they will be autocorrected to the higher value that satisfies the formula.
tRRDS/tRRDL/tFAW 6/8/24, if that works, try 4/6/16 as well (probably won't).
tWTRS/tWTRL 4/12
tWR/tRTP 16/8 (tWR=2*tRTP, but not absolutely necessary)
tRFC will depend on what chip it is, try slightly below 600 (550-590 for ex.).
*SCL to 4 (both)
tRDWR 10, increase if unstable
tWRRD 3
Tertiary:


GDM disabled 2T will allow you to run odd tCL, up to you if you want to try it.

Also, check tPHYRDL values if they are the same 26/26 on both sticks (drop down menu in ZenTimings). They should be the same for your config, but just make sure.

 

[Kodehawa]

Since we're not sure on the chips (Micron rev E?), these are just approx guides. Start with changing one line at a time as written below.

tRC try 61 (tRC=tRAS+tRP), increase slightly if unstable. If one of those values don't follow the formula, they will be autocorrected to the higher value that satisfies the formula.
tRRDS/tRRDL/tFAW 6/8/24, if that works, try 4/6/16 as well (probably won't).
tWTRS/tWTRL 4/12
tWR/tRTP 16/8 (tWR=2*tRTP, but not absolutely necessary)
tRFC will depend on what chip it is, try slightly below 600 (550-590 for ex.).
*SCL to 4 (both)
tRDWR 10, increase if unstable
tWRRD 3
Tertiary:
View attachment 286059

GDM disabled 2T will allow you to run odd tCL, up to you if you want to try it.

Also, check tPHYRDL values if they are the same 26/26 on both sticks (drop down menu in ZenTimings). They should be the same for your config, but just make sure.

tRC 61 caused a POST failure, not even instability, oops

What other value should I try there to make sure it doesn't boom on me? 70-something?

Also, check tPHYRDL values if they are the same 26/26 on both sticks (drop down menu in ZenTimings). They should be the same for your config, but just make sure.

One is 26, the other is 28. All the other values are equal.

 

[3x0]

tRC 61 caused a POST failure, not even instability, oops

What other value should I try there to make sure it doesn't boom on me? 70-something?


One is 26, the other is 28. All the other values are equal.

Yes, try 65-70, if that fails, then 70+. But it should work slightly below 70.

As for tPHYRDL, in my experience, lowering ProcODT helps. Ignore the mismatched values for now, they don't contribute to system instability, just slightly lower perf.

 

[Kodehawa]

Yes, try 65-70, if that fails, then 70+. But it should work slightly below 70.

As for tPHYRDL, in my experience, lowering ProcODT helps. Ignore the mismatched values for now, they don't contribute to system instability, just slightly lower perf.

68 booted fine. Gonna run a memory test now, so I'll report back in an hour or two.

 

[3x0]

Which program are you testing with?

 

[Kodehawa]

Which program are you testing with?

TestMem5 with the Extreme@anta777 profile now.

 

[3x0]

Might want to switch to the 1usmus profile since there are explanations/possible reasons for specific errors.

 

[Kodehawa]

Passed two cycles. I'll mark it as good and do a longer test later. Will try tRRDS/tRRDL/tFAW 6/8/24 in a few hours and see how that fares

Might want to switch to the 1usmus profile since there are explanations/possible reasons for specific errors.

Will do on next round!

EDIT 1: Switched to 1usmus now, and testing tRRDS/tRRDL/tFAW change. Will report back in a few hours.
EDIT 2: 1usmus passed, but it felt rather short. Gonna run anta777 for a few hours to make sure.

 

[3x0]

You can edit the number of cycles or the test duration manually in the .cfg

 

[Kodehawa]

You can edit the number of cycles or the test duration manually in the .cfg

Oh, yeah, thanks. Well I made it 18 cycles. Also went for tWTRS/tWTRL now. Slowly going down the list. This memory tends to fail inmediatly if anything is off lmao, if I do CL18-20(...) it will throw thousands of errors and bluescreen in an instant, so at least it lets me know quickly...

I'll let this run for about two hours now and go down the other line after this.
Thanks you btw I didn't understand a thing about tweaking subtimings, and definitely not where to start.

 

[3x0]

We all have to start somewhere.

The "fun" part is when you get 1 error in 12h of testing. Took me a while to figure out the bugger on my kit.

 

[Kodehawa]

We all have to start somewhere.

The "fun" part is when you get 1 error in 12h of testing. Took me a while to figure out the bugger on my kit.

Guess I'll leave it overnight before going down the list further honestly lmao

 

[3x0]

Just to make sure, open HWInfo sensors in the background while testing to see if any WHEA 19 errors pop up (last entry in the list). WHEA 19 are usually related to the Infinity Fabric, but 1800MHz shouldn't be a problem for it.

 

[Kodehawa]

Just to make sure, open HWInfo sensors in the background while testing to see if any WHEA 19 errors pop up (last entry in the list). WHEA 19 are usually related to the Infinity Fabric, but 1800MHz shouldn't be a problem for it.

Left 1usmus overnight, but it stopped after 7 cycles. No error, it even kept counting the hours and all, but it wasn't using memory, nor using CPU. I imagine the memory manager got killed by windows, but ugh. First time it happens. Well, with the program, I hadn't tried 1usmus profile before.

 

[3x0]

Yeah, happened to me too. Usually it's related to the windows memory manager as you said.

 

[80251]

Is it worth it to have such loose primary timings @ 3600Mhz. if you can get a 1T command rate?

 

[Kodehawa]

Is it worth it to have such loose primary timings @ 3600Mhz. if you can get a 1T command rate?

Stock ones are also loose, 16-20-20-20-40. Maybe on 2T you could get close to stock, but I doubt you could go much further? I tried 2T CL17, failed

 

[Kodehawa]

Yeah, happened to me too. Usually it's related to the windows memory manager as you said.

Hello, sorry for the delay, 'twas a tiring weekend and I couldn't get testing done.

Those timings seem stable at the moment. Time to try the others you've mentioned.

tWR/tRTP 16/8 (tWR=2*tRTP, but not absolutely necessary)
tRFC will depend on what chip it is, try slightly below 600 (550-590 for ex.).
*SCL to 4 (both)
tRDWR 10, increase if unstable
tWRRD 3

Are the ones I haven't done. I'll change tWR/tRTP and test them later today.

--------

Tested tWR/tRTP, *SCL and tRFC in quick tests. I made tRFC 580. Maybe less is possible, but I wanna see tRDWR and tWRRD before I change anything else.

I wonder if any of the primary timings besides tCL could be tightened. I so far tried to do tRCDWR, tRCDDR and tRP the same, but I've seen people doing different values.

Current, missing tRDWR and tWRRD.
EDIT: I have no clue why one image said it was running at 4000MHz. No, it's not LMAO, edited.

 

[3x0]

Regarding tRFC, a "safe" value would be tRC*8, in your case 68*8=544. But greatly depends on RAM temps/voltage and specific chip.

 

[Kodehawa]

Regarding tRFC, a "safe" value would be tRC*8, in your case 68*8=544. But greatly depends on RAM temps/voltage and specific chip.

Hello again. After reading a bit I've come to the conclusion this is Micron Rev. E indeed.
This obviously came with a realization, I can run lower tCL if I don't touch tRCDRD.

I've ran memory tests on 3600 16-21-18-18-40 (21 being tRCDRD) and... it works, with all the other sub-timings tweaked as stated before.

3600 16-21-16-16-40 won't POST even at 1.45V though (CPU light, guess memory controller isn't so happy). Guess I'm stuck at 16-21-18-18-40, but that's a big improvement even from stock timings (3200MHz 16-20-20-20-40)

Passed a "quick" test. Gonna do the rest of the tertiaries now.
Knowing it's Micron Rev. E, got any other suggestions?

 

[3x0]

Knowing it's Micron Rev. E, got any other suggestions?

Not really, but they scale with voltage pretty well, so if you ever decide to go above 3600 don't be afraid to pump more voltage.

When you finish tertiaries, try tRRDS 4 tRRDL 4 tFAW 16 tWTRL 8