[7empe]

I documented my method and linked it in my sig

Thank you! Just started the same routine with adjusting slopes and already have way better results with 4x8 4266 cl16! First of all it was crucial to set odt_read_duration and odt_write_duration to 3 with odt_read_offset and odt_write_offset to 1. This change got rid of q-code 55 while training. Now I am going through all the slopes. Initially I set them 7-1 to avoid auto changes during tests. Right now I am focusing on the first pair (data rising; data falling) and it looks like (8;7) gives me the best result of 11'12" (almost 600% of coverage) without an error. Thanks again for sharing this stuff!

 

[7empe]

Any ideas where to start with VRef? Should I start from 1/2 of VDIMM, just like for VTTDDR?

 

[7empe]

This is what I can achieve after setting correct slopes:

 

[Imprezzion]

After some of you guys saying bandwidth is quite important if latency doesn't suffer too much I went for some higher freq testing.
It's a start lol.

4533 17-19-19-37-370-2T
5.3Ghz all core AVX0
5.0Ghz cache
1.52v DRAM
1.40v SA
1.35v IO
80-40-40 ODT, PPD disabled, DLL 0, slopes not touched yet, vref 0.780, all manual RTL and such, very loose tertiary (except tRDRD_dr).

I'm pretty impressed with this board / IMC now that I delidded it and direct die cooled it. It's getting fed substantially more vCore then before (over 0.1v more lol) and that did wonders for the IMC / cache. Before it struggled hard to even do 4.9Ghz cache and usually had to stress test RAM at 4.7Ghz and the IMC would not handle anything over 4400 RAM at all. Now it somehow breezes through 4533 at 5Ghz cache like it's nothing lol.

Shame I can't seem to be able to get better primaries again on these DIMM's. They really really hate equal primaries or even +1.

Even totally untweaked like this is already has the same latency as my previous 4200 16-17-17-35-280-2T OC at way way more bandwidth.
AIDA test is done with a normal full Windows boot with full bloaty mc bloatface apps and such running in the background. I'm sure if I go full safe mode and disable a bunch of services and task kill a bunch of stuff I can get like 35.4 ish but yeah, that ain't real world lol.

 

[Gen.]

Settings for maining

 

[newls1]

Question fellas, ive been a few weeks with my TREFI Maxed @ 65535 and just wondering at what point temps might show instability with it being set this high? I do have fans blowing over ram (from radiator being directly below) and so far max temp on my B-Die modules is 36c-ish...... What temp does b-die start showing errors?

 

[Imprezzion]

Can be as low as 45-48c. I didn't really get any noticable errors under 53c tho.

Weird this.. I just posted a near 2 hour TM5 run of 4533C17 without errors but it won't even play a game 10 minutes without crashing to desktop.. went back to my 4200C16 profile, doesn't crash.. what could cause that..

 

[Nizzen]

Can be as low as 45-48c. I didn't really get any noticable errors under 53c tho.

Weird this.. I just posted a near 2 hour TM5 run of 4533C17 without errors but it won't even play a game 10 minutes without crashing to desktop.. went back to my 4200C16 profile, doesn't crash.. what could cause that..

Game is loading the whole chain of components. That's why Battlefield V is my choice of memory stability test. Testing cpu, cache/memory. Just everything

 

[itssladenlol]

Can be as low as 45-48c. I didn't really get any noticable errors under 53c tho.

Weird this.. I just posted a near 2 hour TM5 run of 4533C17 without errors but it won't even play a game 10 minutes without crashing to desktop.. went back to my 4200C16 profile, doesn't crash.. what could cause that..

You said you are using vref 0,78 thats like imc suicide lol.
Standard is 0.5 vref, 0,50 - 0,55 is Safe.
0,78 degrades your imc in minutes lol.

Hope you ment Vtt voltage and Not vref otherwise you could have damaged your imc permanently

 

[Imprezzion]

You said you are using vref 0,78 thats like imc suicide lol.
Standard is 0.5 vref, 0,50 - 0,55 is Safe.
0,78 degrades your imc in minutes lol.

Hope you ment Vtt voltage and Not vref otherwise you could have damaged your imc permanently

Yeah I did mean vtt got the terminology mixed up. The one that decides at what value it is a 0 and what is a 1. That is vtt right? I had to raise it slightly above 50% to stop it from erroring really bad in copy tests.

 

[itssladenlol]

Yeah I did mean vtt got the terminology mixed up. The one that decides at what value it is a 0 and what is a 1. That is vtt right? I had to raise it slightly above 50% to stop it from erroring really bad in copy tests.

I dont understand why you run 4533 with such loose timings, makes no sense.
Just look at my 4400 setting.
You should settle for lower Frequency with tight timings, just getting high Frequency with this Bad secondaries and thirds is not gonna benefit anything.

 

[Imprezzion]

I dont understand why you run 4533 with such loose timings, makes no sense.
Just look at my 4400 setting.
You should settle for lower Frequency with tight timings, just getting high Frequency with this Bad secondaries and thirds is not gonna benefit anything.

View attachment 2479971

I did about every tweak you guys ever recommended and it did get me to 4400 17-17-17-36-320-2T with reasonable secondary timings but my tRDWR will not budge lower then 16's no matter what I try and the DIMM's are just too low of a bin to sustain CAS 16 on 4400.

I guess I'm trying to compensate with frequency for these sticks just do not wanna run low timings on any frequency. At any frequency, even 3800 or 4000, these sticks just do not wanna run low enough timings to make it worth while. Efficiency wise the best I ever got was 4200 15-17-17-34-280-2T with tRDWR at 12's, tWRRD at 27/22 (tWTR 7/2), tCWL and tWR 14, tRRD_L+S 4, tFAW 16, RTL IO at 63/63/5/5. This did need a pretty crazy vDIMM of like 1.63v tho. Changing CAS to 16 lowers vDIMM to 1.49v. That is pretty much all these sticks have in them unless I run 4400 straight 17's with looser tertiary timings (16 tRDWR, 29/24 tWRRD) and slightly looser RTL IO (66/66/6/6) but they need very little vDIMM (1.48v) to do that so it might be more efficient in the long run.

There is no way in gods green earth these DIMM's will ever do the 4500+ C16 I see some people do here.

Poor 3600C16's GTZN's with a probably bad / fairly late production date as well.

Maybe I should upgrade to a proper kit... Like a 4400C19 kit or something..

 

[7empe]

Hi Guys, question regarding 2x16GB kit with the best OC potential (high frequency / tight timings). Which set do you recommend? Would 2x16GB Crucial Ballistix Max 4400 MHz CL19 be a good choice? Or maybe the same brand but 4000 MHz CL18 (identical price) would be better? Any other recommendations?

 

[YaqY]

Hi Guys, question regarding 2x16GB kit with the best OC potential (high frequency / tight timings). Which set do you recommend? Would 2x16GB Crucial Ballistix Max 4400 MHz CL19 be a good choice? Or maybe the same brand but 4000 MHz CL18 (identical price) would be better? Any other recommendations?

If you are looking for the best daily performance dual rank bdie is ideal. Ideally the 3600C14 and 3800C14 bins should be strong, 4000C16 is also a nice bin. I have the 3200C14 bin and these sticks can do 4600C16 so shouldn't have issues with them, lots of silicon lottery involved too. If you go for 2x16 4400 Crucial sticks these dimms will be single rank micron rev B as micron uses rev B on their 16Gb sticks above and including 4000, good for frequency but wont beat dual rank bdie for daily performance.

 

[7empe]

If you are looking for the best daily performance dual rank bdie is ideal. Ideally the 3600C14 and 3800C14 bins should be strong, 4000C16 is also a nice bin. I have the 3200C14 bin and these sticks can do 4600C16 so shouldn't have issues with them, lots of silicon lottery involved too. If you go for 2x16 4400 Crucial sticks these dimms will be single rank micron rev B as micron uses rev B on their 16Gb sticks above and including 4000, good for frequency but wont beat dual rank bdie for daily performance.

Thanks! Which brands have 3800C14 or 4000C16 bdies as dual rank?

 

[YaqY]

Thanks! Which brands have 3800C14 or 4000C16 bdies as dual rank?

G.Skill will have them, seems harder to find stock in them at the moment. I have had good experience with the 3200C14 kit which is quite cheaper too as well so i would still consider that a very good kit.

 

[MericaShotUrAss]

Thanks! Which brands have 3800C14 or 4000C16 bdies as dual rank?

Like YaqY said if you want best performance 2x16gb gskill or bust, 3200c14 kit, 3600c16 kit 1.35v will do 4600c16, along with the 3600/3800c14 kits, I’m running 4700c16 daily, and testing 1003 bios atm been running 0088, and gonna go for 4800-16-17-17-34 with almost if not exact same timings as 4600/4700c16
I’m running a 3200c14 kit, and also used a 3600c16 kit at 4500c16 while ago which did
4800-14-14-14-28-280-2t benching.

 

[Nizzen]

Hi Guys, question regarding 2x16GB kit with the best OC potential (high frequency / tight timings). Which set do you recommend? Would 2x16GB Crucial Ballistix Max 4400 MHz CL19 be a good choice? Or maybe the same brand but 4000 MHz CL18 (identical price) would be better? Any other recommendations?

Best dualrank 2x16 kit is g.skill 4266c17 2x16GB.

 

[munternet]

Hi Guys, question regarding 2x16GB kit with the best OC potential (high frequency / tight timings). Which set do you recommend? Would 2x16GB Crucial Ballistix Max 4400 MHz CL19 be a good choice? Or maybe the same brand but 4000 MHz CL18 (identical price) would be better? Any other recommendations?

I have 32GB F4-4266C17D-32GVKB which is quite pricey but a good chance of being a good bin. The 3200c14 which I think @MericaShotUrAss uses has excellent performance but may be more of a lottery if it's a good bin

 

[Imprezzion]

Well, negative Nancy here but my 3600C16 are.. rather mediocre at best. So no, avoid those IMHO.

I'm running a new test as I type this, results incoming. I don't think it's going to pass even 1 hour and it would really really surprise me if it passes 2 hours lol.