[truehighroller1]

@Larkonian It is doubt it is my IMC. I have had no issues with booting for weeks running 2x16 b die at 4133c15 on any of the bios.

When I put in 2x8 it has the issue with booting. Looks to me like a bios/compatiility issue with 2x8. I can pass memtest and tm5 and y crucher with my 2x16. I am not messing with 2x8. Ill stick to 2x16.

Here's your file you requested with my settings for my sr 2x8gb samsung b die memory settings. The highest I can run 100% stable is 15-16-16-32 2t @4000MHz. Delete the .txt

 

[shrimpmaster]

Edit 2 after all that rambling I put in 2x8 kit of gskill 4000c15-16-16 2x8 with the 4000c15 cmo I posted and no boot. c16-16-16- boots so there is something the board doesnt like. Ill check it out and post an update. Strange 2x8 is harder to boot than 2x16 dr b die. Trying another reboot with 4000c16 with the 2x8 failed. Looks like it need some other timings adjusted also.


What ram kit do you have? It could be needing more sa or vddq. I left them on auto, which sets them @1.35v for me. You could try setting sa from 1.4 to 1.45v and vddq from 1.45 to 1.5v to see if it boots.

The other thing you could try is setting the main 3 timings from 15-15-15- to 16-16-16. If that doesnt boot it might be your IMC. You have a 2x8 kit of b die that should run 4000c16 easy.

I am using the gskil 4000c14-15-15 2x16 kit and they are my best kits out of a dozen or so I have had in the last year. I did try 2x8 b die a little and even with my best kit of 2x8 which is a gskill 4000c15-16-16 kit it was harder to get stabe than this 4000c14 DR kit. Ill try messing with 2x8 and see what I can find.

Also before loading a saved profile turn off PSU and hold clear CMOS for 30 secs. It wont probably wont fix anything but I have had it help before because sometimes setting will "stick" and clear cmos can help. But even before you could only run 3900 so it seems like it is something else(mem, IMC?).

Edit saw someone post on another board when first loaing up a new cmo to clear cmos as I said above and the first time rebooting the boot will fail.

"When you do cmos clear, loading the Profile and get a Safeboot .. just go to Bios and F10+Enter again"

source ROG STRIX Z690-A GAMING WIFI D4

I now start to think it's not my IMC. I can post my 2x8gb gskill b-die up to 4100mhz (fails to load windows). But it's stable only up to 3900mhz no matter the voltages i try. For 3900mhz it passes tm5 and linpack xtreme with only 1.2v SA.
4000mhz is almost stable.

Strange thing is on gear 2, it works stable up to 4600mhz, soo not memory either.

Running 3900mhz 15-15-15-36 1t on 12700k + z690 tuf.

 

[IIISLIDEIII]

Edit 2 after all that rambling I put in 2x8 kit of gskill 4000c15-16-16 2x8 with the 4000c15 cmo I posted and no boot. c16-16-16- boots so there is something the board doesnt like. Ill check it out and post an update. Strange 2x8 is harder to boot than 2x16 dr b die. Trying another reboot with 4000c16 with the 2x8 failed. Looks like it need some other timings adjusted also.


What ram kit do you have? It could be needing more sa or vddq. I left them on auto, which sets them @1.35v for me. You could try setting sa from 1.4 to 1.45v and vddq from 1.45 to 1.5v to see if it boots.

The other thing you could try is setting the main 3 timings from 15-15-15- to 16-16-16. If that doesnt boot it might be your IMC. You have a 2x8 kit of b die that should run 4000c16 easy.

I am using the gskil 4000c14-15-15 2x16 kit and they are my best kits out of a dozen or so I have had in the last year. I did try 2x8 b die a little and even with my best kit of 2x8 which is a gskill 4000c15-16-16 kit it was harder to get stabe than this 4000c14 DR kit. Ill try messing with 2x8 and see what I can find.

Also before loading a saved profile turn off PSU and hold clear CMOS for 30 secs. It wont probably wont fix anything but I have had it help before because sometimes setting will "stick" and clear cmos can help. But even before you could only run 3900 so it seems like it is something else(mem, IMC?).

Edit saw someone post on another board when first loaing up a new cmo to clear cmos as I said above and the first time rebooting the boot will fail.

"When you do cmos clear, loading the Profile and get a Safeboot .. just go to Bios and F10+Enter again"

source ROG STRIX Z690-A GAMING WIFI D4

You should try 15-16-16-35 instead of 15-15-15-35. There are lots of B-Die that can't do tRCD 15 at 4000 speeds.

Also set DRAM Write Latency (tCWL) to 15, again it is only the best B-Die bins that can handle that.

my kit is this:

F4-4000C14D-16GTRS-(EOL) - Overview - G.SKILL International Enterprise Co., Ltd.

Trident Z Royal is the latest addition to the Trident Z flagship family and features a crown jewel design. Meticulously crafted to display just the right amount of light refraction, the patented crystalline light bar scatters the RGB colors in a magnificent display of LED lighting. Mirroring the...

www.gskill.com

I currently have the following voltages:
System Agent Voltage: 1.42v
VDDQ Voltage: 1.5v
DRAM Voltage: 1.55
100.133
gear1
4000mhz with xmp profile (14-15-15-35 at 1.55v) everything works, (no errors after 2 hours of extreme on testmem5), but if I try to enter those values you gave me it does not start.

the memory (with xmp) is stable even if I lower the voltages of the system agent and vddq, now I put them on:
-sa 1.34
-vddq 1.35
-dram 1.55


shar net

 

[truehighroller1]

I now start to think it's not my IMC. I can post my 2x8gb gskill b-die up to 4100mhz (fails to load windows). But it's stable only up to 3900mhz no matter the voltages i try. For 3900mhz it passes tm5 and linpack xtreme with only 1.2v SA.
4000mhz is almost stable.

Strange thing is on gear 2, it works stable up to 4600mhz, soo not memory either.

Running 3900mhz 15-15-15-36 1t on 12700k + z690 tuf.

Here's my max stable right now with my memory kit but I can boot to windows at 4133 as well just not stable.

I have the following memory kit.

G.SKILL TridentZ RGB Series 32GB (4 x 8GB) DDR4 4000 F4-4000C17Q-32GTZR

 

[neteng101]

New bios 121

Gave this a go - I've been tuning my OC reducing voltages on 115 (Vcore 1.32 Auto LLQ VCCSA 1.25 VDDQ 1.35 DRAM 1.40). 121 still doesn't train my memory correctly for just 3600 G1 (lower/higher works). Can't say I see any real difference 115 > 121, maybe margin of error improvements...

115 Pre Update

121 Post Update

 

[ssgwright]

Here's my new DDR 5 with a slight 200mhz bump... need to see if I can work on that latency

 

[IIISLIDEIII]

@Larkonian It is doubt it is my IMC. I have had no issues with booting for weeks running 2x16 b die at 4133c15 on any of the bios.

When I put in 2x8 it has the issue with booting. Looks to me like a bios/compatiility issue with 2x8. I can pass memtest and tm5 and y crucher with my 2x16. I am not messing with 2x8. Ill stick to 2x16.

let's say that I would be curious if the ram kit is "unlucky" because I still have time to make a return of this and I can have another one sent.
It seems strange to me that the ram comes with an xmp profile already at the limit, there should be some margin or am I wrong?
Another strange thing is that with the timing of xmp the ram is stable in gear1 at 4000 even with sa and vddq voltage at 1.35, I don't need to get close to 1.4, so in theory I should be able to tighten the times a bit.

Should the e-cores be deactivated or does it have nothing to do with it?

 

[bscool]

my kit is this:

F4-4000C14D-16GTRS-(EOL) - Overview - G.SKILL International Enterprise Co., Ltd.

Trident Z Royal is the latest addition to the Trident Z flagship family and features a crown jewel design. Meticulously crafted to display just the right amount of light refraction, the patented crystalline light bar scatters the RGB colors in a magnificent display of LED lighting. Mirroring the...

www.gskill.com

I currently have the following voltages:
System Agent Voltage: 1.42v
VDDQ Voltage: 1.5v
DRAM Voltage: 1.55
100.133
gear1
4000mhz with xmp profile (14-15-15-35 at 1.55v) everything works, (no errors after 2 hours of extreme on testmem5), but if I try to enter those values you gave me it does not start.

the memory (with xmp) is stable even if I lower the voltages of the system agent and vddq, now I put them on:
-sa 1.34
-vddq 1.35
-dram 1.55


shar net

I dont know what to tell you. I messed with bios 815 a little and it is a mess for me using 2x8. I dont feel like spending hour trying to figure it you. Maybe try the timings in the txt in another bios version or in the bios you found stable @3900. They will still give you a good boost in bandwidth and decrease latency.

The memory kit you have is top notch. Message me if you want more help. It will take a lot trial and error and I dont want clutter up this thread. ill send you a pm and you can reply if you want.

 

[truehighroller1]

LOOK WHAT I FOUND IN THE OCTOOL!

 

[bscool]

let's say that I would be curious if the ram kit is "unlucky" because I still have time to make a return of this and I can have another one sent.
It seems strange to me that the ram comes with an xmp profile already at the limit, there should be some margin or am I wrong?
Another strange thing is that with the timing of xmp the ram is stable in gear1 at 4000 even with sa and vddq voltage at 1.35, I don't need to get close to 1.4, so in theory I should be able to tighten the times a bit.

Should the e-cores be deactivated or does it have nothing to do with it?

Just from what I can see I think you should go with 2x16. But that is twice the price. Just testing 2x8 it looks strange on z690 Strix 812. But I dont want to stir you wrong. That is my 2 cents.

Maybe an older bios is better?.

There shouldnt be any reason 2x8 b die doesnt work well? unless it z690 is somehow optimized for 16gb DR dims being most will use that for ddr4 and ddr5 on z690. I really do not know.

 

[bscool]

Maybe 2x8 needs odts set manually?

 

[bscool]

LOOK WHAT I FOUND IN THE OCTOOL!

View attachment 2538732

That is a good find. Interesting.

 

[PolRoger]

Here's my new DDR 5 with a slight 200mhz bump... need to see if I can work on that latency

View attachment 2538730

Which brand/model kit of DDR5 6000 did you get in the NewEgg shuffle? I'm also kind of curious as to how the Egg was pricing kits of DDR5 now?

 

[ssgwright]

Are you a human?

www.newegg.com

 

[Freshair]

Hello from Germany, i am new here.
Very interessting Thread, with lots of info. My Board is MSI z690 edge ddr4. I have been trying to get my two sticks of 8GB b-dies stable at different settings for 4 days now.
I can boot into Win10 with several combinations up two 4000+ in gear 1 and up to 4800 in gear 2. No problems in normal use of Win, but always errors in tests and gaming.
I always played with timings/subs and voltages to become my favorite setting 3866 14-14-14-32 1T stable, but always error on tests. Today i testet lowering sa and vddq Voltage,
and i cant believe it, this thing was stable at 1.25 vddq and 1.256 sa . So i think these voltages have much effect on overclocking, but since yet i didnt see on what it depends.
So now the first thing for me is, how low can i get with voltage and then trying higher frequenz. I always used voltages above 1.3 and on auto the board set sa to 1.45..jesus.
High sa produces lot more heat on my ram, that can also be a reason for instability. There for these setting are very fast and i realy didnt need more.

 

[grey.clock]

Idk how you do it lol. I have some cheap Oloy Blades 4x16Gb 3600 14-14-14-34 @1.45v, XMP doesn't even get close to working on an Asus Strix Z690-A. I've been manually tuning it for about 10 days now and the closest settings I could get to work are 1.329v SA, 1.205v VDDQ, 1.45625v DRAM. Anything more than 1.21 VDDQ seems to make it really unstable. I'm just trying to run them at stock 3600 CL14.

They run fine with just 2 sticks though.

how do you like the oloy blade cl14 3600's with two sticks? I am running the same kit myself. I am really new to tweaking memory..... been overclocking since the amd xp days but alder lake is the first time I have gotten into tweaking the timings since the rest of my system is... maxxed out without options for new hardware due to shortages. I am using a 12700k, ASUS TUF and have it at around 55ns

 

[Revv23]

Plenty of proof DR @4000+ with tight timings is not happening on MSI. If so post links. To memtest stable 4000 to 4133+ tight timing DR.

ive posted a few myself over 4000; i havent had time to optimize timings yet. All I was implying was that I had no wall, i literally got over 4000mhz with auto settings on 1.00 bios.

I also don't see many with memtest stable tight timings over 4000mhz - but i know it's possible. I think we have an issue of people seeing some really aggressive overclocks running an aida64 latency test and then wondering why they cant run 24/7 at those speeds with a weaker IMC.

I also think there is a much bigger IMC difference than we first thought between chips, buildzoid cant run 4000+ stable on any of the 3 brands he has tried so far.

 

[bscool]

ive posted a few myself over 4000; i havent had time to optimize timings yet. All I was implying was that I had no wall, i literally got over 4000mhz with auto settings on 1.00 bios.

I also don't see many with memtest stable tight timings over 4000mhz - but i know it's possible. I think we have an issue of people seeing some really aggressive overclocks running an aida64 latency test and then wondering why they cant run 24/7 at those speeds with a weaker IMC.

I also think there is a much bigger IMC difference than we first thought between chips, buildzoid cant run 4000+ stable on any of the 3 brands he has tried so far.

After having issues running 2x8 b die on the Strix bios 812 today I don't know about anything right now

 

[Revv23]

After having issues running 2x8 b die on the Strix today I dont know about anything right now

Yeah thats a new one too LOL. I should get my 2*8 out but im having too much fun playing halo right now to tweak LOL.

I'd like to give my asrock board another shot too now that they have a new bios but again having too much fun gaming. my same kit that auto'd over 4000 couldnt even get to 3000 on asrock. It XMP'd 3600 14-14-14 on the EDGE 1st boot.

 

[zhrooms]

Just from what I can see I think you should go with 2x16. But that is twice the price. Just testing 2x8 it looks strange on z690 Strix 812. But I dont want to stir you wrong. That is my 2 cents. Maybe an older bios is better?. There shouldnt be any reason 2x8 b die doesnt work well? unless it z690 is somehow optimized for 16gb DR dims being most will use that for ddr4 and ddr5 on z690. I really do not know.

2x8GB Single Rank ran like a dream for me (and many others in this thread), I ran 4133 16-16-16 1.45V without an issue on the Strix with 0707, before I returned it, the exact kit was the 2x8GB Trident Z RGB 3600 16-16-16 1.35V from 2017. From my personal experience, using both 2x8GB Single Rank and 2x16GB Dual Rank, on TUF/Strix with 0707 on both, there is no difference between the boards or the sticks in terms of compatibility, the boards also overclock the same, which isn't a surprise as they're the same board more or less, and very similar sticks (both hits a wall at 4133 15-15-15 unless you push voltage), the only issue I've had on either board/kit is my IMC, but I think I've proven that the board itself (TUF) is already overkill from my 42.4ns overclock, by that I mean that it should run just the same on the cheaper ASUS Z690-P D4, really wish I could test a MSI Z690-A Pro, after having heard so many people have issues with it, but at the same time there are just as many that show screenshots of it being great, so it's still unclear to me if it's user error or something else, now that I know my exact IMC and RAM limit, would be very apparent if a new board I tried was better or worse, sadly I don't feel like putting any more time or effort into testing more boards, after the whole fiasco of purchasing a defective 12900K and having to return two Gigabyte boards because of trash BIOSes on launch, and I have everything working perfectly right now so would be nice to keep that going.