ASRock X99 OC Formula/3.1 BIOS and Software issues

I posted this over at XS.org last week without any reply up to now. So I'm trying the official ASRock-Forum for this.

Xtremesystems.org wrote: I'm running an OC Formula/3.1 for about two weeks now in my main rig. ASRock X99 OC Formula/3.1(BIOS 1.20) Intel Core i7-5960X various memory-kits 2x NVIDIA GeForce Titan X Enermax MaxRevo 1350w Crucial C300 256 GB (Windows) 2x Crucial M500 960GB @ Raid0 Custom Watercooling I have a few issues which I like to get sorted out. First there is the DMI PLL Voltage. Left to this option in bios there is a value of 1.500 shown. On the right the description says "Chipset 1.05 Voltage". Now I'm unsure which information is right. Trying to set the voltage manually leads to the fact that the lowest setting is 1.100 volt and the 1.500-setting is already colored red. I searched a few Intel-docs but I couldn't find anything about DMI PLL Voltage. So what is the right value for this and if  it is the 1.500-setting, why is is colored in red? Am I taking a risk here? The next thing is the missing hardware monitoring in Windows 10 64 Bit. I tried two installations of AFD but I only have the tab for OC Tweaker - nothing else. No Tools, no System Info, no Live Update - nothing. Since I saw a screenshot over at Tweaktown where everything is fine I think it's an issue with Windows 10 (Tweaktown used Windows 7 for their review). I tried different compatibility-settings without success. So please give me a compatible version for Windows 10. Next one. The most annoying issue is memory overclocking. I have three sets of memory here at the moment: 4x 4 GB G.Skill Ripjaws4 F4-2666C15Q-16GRR 4x 8 GB G.Skill RipjawsV F4-3200C16Q-32GVKB 4x 16 GB Samsung DIMM (2x 2133, 2x 2400) Everytime I try to reach memory clocks of 3000 mhz and higher I am loosing channel D. I already reseated the cpu, I swapped memory kits, swapped sticks between channels and played a few hours with RTL and IOL - everything without success. Yesterday I found out that I can boot with channel D when I increase VDimm heavily. With VDimm higher than 1.55 volt I can boot both Ripjaws-kits with 3200. For the RipjawsV I used completely stock settings (timings and voltages) and only channels A, B and C work - with +200 mv D is initialized correctly, too. I didn't try with the 16-gb-sticks but I think I will have a benefit of high vdimm, too. But that setting is way to high for a 24/7-setting - especially when regarding the fact that 1.55 is much higher than the stock VDimm for the RipjawsV. There are no faulty sticks since they boot fine with stock settings if I put them in channel A, B or C. The memory controller shouldn't be a problem here since this cpu is able to handle memory up to DDR4-3333 with tighter timings. What's the deal here? Then I discovered another "bug". Since I am running the rig without energy saving-settings I am using override voltage and fixed multipliers. I set the cache multiplier and the minimum cache multiplier both to the same value (doesn't matter which value it is). But when I'm doing this the Boot Failure Guart kicks in at nearly every reboot. This problem disappears as long as I set a fixed multiplier and leave the minimum multiplier at AUTO. That's it for now. Would be cool if some of the issues could be sorted out because otherwise I don't know why this board would be called "OC Formula". Don't get me wrong - I like challenges. But up to now that's to much issues for a board in this price-range. ;)

Meanwhile I have an additional information regarding DMI PLL Voltage.

Setting this voltage to the lowest possible value of 1.100 helped not to loose channel D at DDR4-3000. Now I can use 4x 16 GB at 3000 with "low" volts and channel D is only missing sometimes at first boot after I change memory timings. A reboot solves this. Do I set DMI PLL Voltage to 1.15 channel D is missing again frequently. With 1.1 volt I could also run 3200 with 4x 8 GB G.Skill RipjawsV F4-3600C17D-16GVK (2 Kits). I had issues with memory bandwith (was very low) but all four channels were recognized.

So why is DMI PLL Voltage set to 1.5 volt by default? It is colored red (so I assume it is kind of an "unhealthy" value) and it seems to hurt memory overclocking. Is there any reason for this?

Xaltar wrote: Firstly, that is a beast of a build

Thank you, it's quite ok. It still isn't enough power to play Rise Of The Tomb Raider in UHD with maximum graphic details. So I think I have to upgrade soon....



Yesterday I had to realize that my configuration with 4x 16 GB Samsung was not stable at 3000. Now I switched to 4x 8 GB G.Skill RipjawsV F4-3600C17D-16GVK (2 Kits, single sided, Samsung B-Die) as my final configuration. I am running them currently at 3000 with 13-13-13 at 1.35 volt. DMI PLL Voltage is still set to 1.1 volt without any hicup up to now.

xrzhang wrote: I read your post and I have a question for you.  I have X99 OC Formula and 5820K. I would like to upgrade memory to 4X16G Corsair Vengeance LPX 64GB (4x16GB) DDR4 DRAM 2800MHz (PC4-22400) C14 Memory Kit Since I do not find any information about if the motherboard supports 4X16G,  I want to know: 1. 4X16G works fine on your board? 2. 64G is the maximum for this board and 5960? Thanks!

With BIOS 1.10 I wasn't able to boot with 16-gb-sticks installed. I had to flash 1.20 or 1.21 beta to get these sticks to work. But my goal was to reach DDR4-3000 and this setting wasn't stable. So I changed the sticks.

I never tried lower clocks with theses 16-gb-sticks so I can't say anything about stability. But I think they should be stable at 2666 or in worst case at 2133 (default memory clock for a S2011-3-cpu).

For your 2nd question I don't have an answer. Since 16-gb-sticks are supported by the newer bios-versions it should be possible to put 8x 16 gb in. But I didn't test this so this is an assumption only. But for such a configuration I am pretty sure that you have to deal with lower memory clocks. DDR4-2133 would be my guess for a perfect stable system with an OC Formula/3.1.