help me overclock 64GB (2x32GB) G.Skill Trident Z5 DDR5 6400 CL32 (HYNIX)

[rcarlos243]

I have a 13600K with Z690 Unify X and 64GB (2x32GB) G.Skill Trident Z5 DDR5 6400 CL32 (F5-6400J3239G32GX2-TZ5RK).

 

I do not plan on adding active cooling or raise the voltage. It is currently running at 1.4v through XMP and hits up to 80c when stress tested through TM5 anta777 extreme

 

It is a Hynix according to Thaiphoon Burner

 

I was able to increase tREFI from 8k into 32767 (tried 65k but threw errors when stress tested).

I was able to to decrease tRFC2 from 510 into 383 (tried lower but it won't post).

 

 

which timing would give me better performance in gaming and what timing should I try to change?

[Shimejii]

19 minutes ago, rcarlos243 said:

I have a 13600K with Z690 Unify X and 64GB (2x32GB) G.Skill Trident Z5 DDR5 6400 CL32 (F5-6400J3239G32GX2-TZ5RK).

 

I do not plan on adding active cooling or raise the voltage. It is currently running at 1.4v through XMP and hits up to 80c when stress tested through TM5 anta777 extreme

 

It is a Hynix according to Thaiphoon Burner

 

 

 

 

 

which timing would give me better performance in gaming and what timing should I try to change?

There wont be one. There just isnt any real gains here to do that you will ACTUALLY notice. 1-3 fps is not noticable, thats the margin of improvement you are going to see with tinkering with this type of ram beyond 6400 is such a huge diminishing return atm. 

 

Depending on the game and resolution, you may not even see an improvement of FPS. Quite a few games dont see much improvement. On the otherhand you may have stability issues, most of the newer cpus have shown stability going to 7200ish but again its not going to impact gaming at all beyond a few fps.

 

Timings wise its going to depend, see below

[RONOTHAN##]

3 minutes ago, rcarlos243 said:

which timing would give me better performance in gaming and what timing should I try to change?

To be fair, the main timings you've adjusted (tREFI and tRCD) are the most important. Other timings that are good to set though would be:

1. tRCD_W - This can almost always be set to 8
2. tRAS - At 6400 this can almost always be set to 28, though I have seen some reports that with dual rank memory it can't go quite as low. 
3. tRTP - This depends on the sticks. Every A die stick should be able to do 16, with the good ones being able to do as low as 6. My kit at 6400 should do 8, though your mileage may vary. 
4. tRRD/tRRD_L - This also varies per stick. At worst, these should be set to 8/8 as that should work on every A die kit and there isn't a ton of benefit going lower. That said, most kits should do 4 on tRRD and sometimes a bit lower on tRRD_L (my kit does 7, though not every kit does below 8). 
5. tFAW - Set this to 16
6. tCKE - Set this to 4, though this occasionally doesn't work. 
7. tCWL - Most A die should be able to set this crazy low. My sticks will do as low as 8, though your mileage may vary. 
8. tRDRD_SG - This can usually be set to 12 at 6400, though you might need to loosen it out to 14
9. tRDWR_SG/tRDWR_DG - At 6400 I'd expect to be able to set these to 18, though you might need to stick to 20 depending on how good your sticks are
10. tWRWR_SG - This can almost always do 12, and most G.Skill sticks I've seen can do this at 10. 
11. tWRRD_SG/tWRRD_DG - This depends on the kit, though I've yet to see 64/48 not work, and I've seen as low as 58/44 work. These can take a while to error though, so keep that in mind.
12. DR/DD timings - These do have a fair bit of headroom in them, though since I don't really have any experience with dual rank A die on Intel I don't know quite how you'd have to set them. 
13. tRDPDEN/tWRPDEN/tCPDED - These can be set to 4. This disables the power down mode effectively. 
14. tWRPRE - This is what actually controls the tWR timing, and this does get affected by some other timings (IIRC tCL and tCWL, though I forget exactly which). This does have some variance in what works and doesn't, though I'd be surprised if 80 didn't work, and if you're lucky you might be able to get as low as 32 working. 

Most of those timings don't have massive affects on performance, though all together they should get a bit of an uplift. 

[rcarlos243]

13 minutes ago, Shimejii said:

There wont be one. There just isnt any real gains here to do that you will ACTUALLY notice. 1-3 fps is not noticable, thats the margin of improvement you are going to see with tinkering with this type of ram beyond 6400 is such a huge diminishing return atm.

 

HWUnboxed did a test with a tuned DDR5 and on CPU bound scenarios 1% lows got around 10FPS boost

 

Why do you bother posting blatant false information?

 

[Shimejii]

1 hour ago, rcarlos243 said:

 

HWUnboxed did a test with a tuned DDR5 and on CPU bound scenarios 1% lows got around 10FPS boost

 

Why do you bother posting blatant false information?

 

Because 10 fps in the BEST of cases makes very little difference going from 165 FPS to 181 fps in a game like spiderman, 268 fps to 277 fps in Calisto oh such a huge gain and noticeable. Plague tale 150 to 161 FPS. You do realize that this kind of difference is not what most people who deal with this kind of thing consider a big improvement, its minor stuff that at the end of the day take. And that is at 1080p. 

 

AT the end of doing all of that, you are talking about marginal gains that can affect stability to those that dont know what they are doing. Its nice to see that they are learning how to do it, but there needs to be tempered expectation. If it was comparing 4800 to 6000 tuned, yes there would be a huge difference, but comparing a 6400 kit with already decent timing? No there really wont be any noticeable improvement to the actual gameplay. 

 

Other tasks or games that are much more RAM heavy? i can see a slight improvement, but they are already using a very good kit of memory and again it should not make a difference. 

 

Its something fun to do and try, but not much results. Usually the big thing i see with tinkering subtimings and such is stability, if the ram kits arent stable and you want to tinker to fix it that tends to be what you have to do when going beyond 6400+. Diminishing returns are very real

[rcarlos243]

15 minutes ago, Shimejii said:

Because 10 fps in the BEST of cases makes very little difference going from 165 FPS to 181 fps in a game like spiderman, 268 fps to 277 fps in Calisto oh such a huge gain and noticeable. Plague tale 150 to 161 FPS. You do realize that this kind of difference is not what most people who deal with this kind of thing consider a big improvement, its minor stuff that at the end of the day take. And that is at 1080p. 

 

AT the end of doing all of that, you are talking about marginal gains that can affect stability to those that dont know what they are doing. Its nice to see that they are learning how to do it, but there needs to be tempered expectation. If it was comparing 4800 to 6000 tuned, yes there would be a huge difference, but comparing a 6400 kit with already decent timing? No there really wont be any noticeable improvement to the actual gameplay. 

 

Other tasks or games that are much more RAM heavy? i can see a slight improvement, but they are already using a very good kit of memory and again it should not make a difference. 

 

Its something fun to do and try, but not much results. Usually the big thing i see with tinkering subtimings and such is stability, if the ram kits arent stable and you want to tinker to fix it that tends to be what you have to do when going beyond 6400+. Diminishing returns are very real

 

Whether or not 10FPS worth of performance is noticeable to you or not is irrelevant to me.

 

All you rambling is your subjective opinion that was never asked in the first place.