Gigabyte Z170X-SOC Force - DDR4

[xxbassplayerxx]

Hey, as one last trial, set the 33.33 multi and 100 BCLK then change the BCLK using the buttons on the MB once you make it in to Windows. Test this for stability.

 

screen shot all of your timings (including 2nds and 3rds). From there, set the RTL and IOL values from the BIOS. See my other posts in this thread for how to format those. RTL and IOL are a big part of the training process and very possibly could be what is failing.

[Fact11]

Hey XBass, I am using the Asrock Timing Calculator, to view my timings when i'm in Windows, but when I have tried to manually enter the RTL and IOL, it always fails to boot. If I leave them on AUTO it boots up fine.

 

But then again I don't know if I'm doing it properly and whether or not you need to change all the timings from AUTO to a specific value and then start changing one timing at a time.

 

I have just noticed that every time I lower my CAS, the RTL and IOL numbers get tighter. But I'm not too sure what too set them too manually, in order to try achieve a lower CAS.

 

I have attached a screenshot of the asrock calculator, showing the timings I get, when I have set my primary timings to 16-18-18-28-1T and the RTL and IOL that get generated from being set to auto.

 

Based on that, what secondary and tertiary timings should I set manually instead of leaving on auto, so that if I make my primary settings 15-18-18-28-1T, that it might pass training and boot up?

 

Here is a clearer view of the attachment that is easier to read

 

http://i.imgur.com/BEcMN2P.jpg

Edited February 6, 2016 by Fact11

[xxbassplayerxx]

Are you entering the RTL and IOL as I explained earlier? It's very counterintuitive.

 

I'm out and about today but if you haven't found the post by later, I'll link you to it.

 

EDIT: post is here http://forum.hwbot.org/showpost.php?p=417742&postcount=26

Edited February 6, 2016 by xxbassplayerxx

[Fact11]

Ok thanks, i'll have a read of it and give it a try and see how I go.

[Wiggles]

It looks like E-Die, CoolHandLuke on OCN said that he had better success in 32M and with 2T and a higher CAS at similar timings.

[xxbassplayerxx]

Wiggles, is he using an SOC Force??

[phobosq]

I have on my benchtable GB Z170X SOC Force, had both OCF and Impact previously. I can compare directly OC of my kit of RipjawsV 3200C16, which do 3733C12 with 1.85 vmem with my 6600k on both OCF and Impact easily with tight secondary and third timings, while on GB they don't post higher than 3333C14 no matter what voltage or subtimings. Tried that on BIOSes F5 and F6h. I also tried AFR (Fury Black 2666C15) but they didn't go higher then 3100C12, higher frequencies = no boot.

Main question - is it me or the board?

[zeropluszero]

Try X08 bios

https://www.dropbox.com/s/of0kx0m7jm8qhpz/Z170XSOCForce.rar?dl=0

 

I'm not having any trouble with B die at 3866 12-12-12

[xxbassplayerxx]

Zero, have you been able to get any higher than 3866? Can you share how you've been able to get so high? I was having trouble getting the board to do 3600 12-12-12 with proven 4K C12 sticks last week.

[Guest george.kokovinis]

Folks, do you mind if I join the discussion ?

 

Have very recent experience with GA-Z170X-Gaming 7.

Latest bios F7R.

 

Five Z170 boards have passed from my hands, so I can make direct

comparisons with MFR, e-die and b-die memory modules.

 

1) Asus M8E. A board I will never sell.

2) Asus Hero. Gone.

3) Z170 OC Formula. Will sell only if M-Ocf becomes retail.

4) MSI-M5 Gaming. This was a wrong decision.

5) Gigabyte Z170X Gaming 7.

 

Surely inferior theoretically to SOC-Force, but here are my findings.

Keep in mind that all GA boards undervolt everything, even at LLC high.

 

Kingston Predator 3000 - 4x4gb.

3200 - 12-15-15-28-1T with 1.65V and 1.25VccIO and 1.30VccSA.

 

Ripjaws V ( high binned ). 3600. 4x4gb. e-die

3866 - 13-19-19-28-280-2T. 1.86Vdimm ( bios ), 1.275VccIO and 1.30VccSA.

Here the secret is to give PCH 1.20V, otherwise it does not boot or reboot.

 

Feel free to adjust subs as you like. From normal to ultra tight.

 

Trident Z - 3600. 2x8gb - b-die

3866- 12-12-12-28-220-2T.

Not easy to reach this goal.

Needs very fine tuning.

Exactly 1.88Vdimm.

Training and eventual voltage set manually.

Vccio 1.265V

VccSA 1.30V

PCH 1.20V

 

Subs :

Write recovery : 13

Read to precharge : 6

Ras to Ras 7 ( L+S )

Write to read ( L+s ) 5

tFaw : 12

tcwl : 9

trefi : 65535

tcke : auto

 

IOL offset - both channels -22

 

 

Rock solid with many reboots.

I do not know if all the above apply to SOC-Force.

Also 4000 on b-die which is easy both on M8E and OCF, is a no-no.

[zeropluszero]

Zero, have you been able to get any higher than 3866? Can you share how you've been able to get so high? I was having trouble getting the board to do 3600 12-12-12 with proven 4K C12 sticks last week.

Honestly nothing special.

When I first got the sticks I booted 12-12-12-28 1T twcl11, trfc280, all others auto at 1.8v, 1.25vccio and vccsa. Reboot, change to 3733, reboot 3866, 55 @ 4000. Then went back to 3866 and tightened. My sticks I was trying too high vdimm, i think they like the 1.8-1.9 range, so if you're higher than that, bring them down.

[xxbassplayerxx]

Interesting. My B-Die should be here tomorrow so I'll keep pushing. Do you have any screens with memory timings available?

[zeropluszero]

Nothing you cant find digging through others results.

I've found my board doesnt like twcl9, and thirds dont go quite as hard as asrock.

[xxbassplayerxx]

Just an update here. Got my B-Die in and tried a few things.

 

No code has to be inserted here.

 

That was about as far as I got last night. Not really looking good so far. I also tried random boots at 3733 and 3866 12-12-12.

 

EDIT: @zeropluszero, can you get some XTU or Pi runs in on this board and share results? You're the only one that has mentioned getting reasonable speeds so far.

 

EDIT 2: Found these results with your 12-12-12 timings. Wish it was as easy for me as it was for you

 

[hwbot=3133661]submission[/hwbot]

[hwbot=3133662]submission[/hwbot]

[hwbot=3133659]submission[/hwbot]

Edited March 2, 2016 by xxbassplayerxx

[zeropluszero]

Check my results in http://forums.overclockers.com.au/showthread.php?t=1173853 and http://forums.overclockers.com.au/showthread.php?t=1186910

I honestly havent benched b die under 3866 12-12-12

 

Why are you using XMP?

 

try a higher tRFC to start with, start at 280. tighten to 220 when you get the other stuff first.

I'm not in the habit of sharing timing configurator, but honestly, the seconds and thirds are barely any different than other brands.

[xxbassplayerxx]

I tried with XMP off as well.

 

Basically, even this would not boot:

 

XMP off, 1.26V SA & IO, 1.80V vDIMM, 3600 12-12-12 280 1T, all others auto

 

As you can see above, even 3000 12-12-12 wouldn't boot from the get-go. I'll keep trying tonight... Just getting frustrating.

[zeropluszero]

how about single channel? maybe one stick is shit?

or try swapping sticks in different slots.

[xxbassplayerxx]

Secret found. DRAM Initial Volts on auto is 1.50V no matter what you set the real volts at (or at least when I set 1.80V it was trying to train at 1.50V). Raising that to be equal or just below DRAM volts has allowed me great success! I'll post my results shortly.

Edited March 3, 2016 by xxbassplayerxx

[crustytheclown]

That's exactly what i said at post #24

[xxbassplayerxx]

That's exactly what i said at post #24

 

You are a god above men :ws:

 

I read that but then probably had to wait a week or so before I actually got the chance to mess with the board and completely forgot about it

[zeropluszero]

I guess i always set that. Didn't think it that important

[xxbassplayerxx]

I usually only set the obvious stuff when it comes to voltages. I generally leave the less familiar stuff on auto.

 

Anyway, I just submitted this:

[hwbot=3146773]submission[/hwbot]

 

 

Nothing touched beyond mains and tRFC. Will get to tweaking tonight if I get a moment. I was also able to get in and run XTU at 4100 12-12-12 28 220 2T but scores were really low (600 or so). Will dig in and see if I can fix it.

Edited March 3, 2016 by xxbassplayerxx

[zeropluszero]

yeh, heaps of seconds and thirds work to do there, but that doesnt sound too bad.

[xxbassplayerxx]

Have you run this board with an i3? Seems it rounds BCLK down to nearest 0.50. The highest I can go without a hard lock is 102.74, which ends up being 102.5 in OS. When opening GTL (as shown in my image above) it reflects 102.5 even though 102.74 was selected in BIOS and booted at.

 

102.75 would round up to 103.00 which is locked out on these chips, causing the hard lock.

 

This is the same issue that the Asus boards have, I think, and is why they can't do 742.

Edited March 3, 2016 by xxbassplayerxx

[zeropluszero]

maybe this weekend. chip should arrive today.