From BIOS 1601 to 1701 produces RAM compatibility issues

[kaese]

I have a RAM Kit (type 32GB HyperX FURY DDR4-2666, Quad-Kit) that can boot and run flawlessly on BIOS 1601, while in contrary BIOS 1701 fails to boot or run the same kit at 2666 MHz (even after intense tweaking).
So, RAM compatibility has decreased (at least for my case).

Now, especially because 1701 has a microcode update, I dont want to be stuck to BIOS 1601 forever!
Does anybody have some insight, onto what has changed from 1601 to 1701 regarding DDR4 control?


Note that I made the same observation with two CPUs (5820K - different batch: one Costa Rica and one Malaysia) and two identical RAM-Kits.

[GoNz0-]

Welcome to Asus, they will tell you that you need to tweak a ram kit that they once said is compatible as they broke something. Not that they will admit to that!

Have you tried running it at 100blk instead of the 125 it tends to set to when you enable XMP?

Also did you play with the system agent voltage as it seems to point to that as well.

[kaese]

Yes, I tried both 100 and 125 blk but there was not much of a difference (in general, I noticed a tendency to more stability at 100).

I tried some V_SA tuning but not excessively - I will do that on the weekend....

The weird thing is with 1701 it doesnt even work out-of-the box with only one module... At least the board has a lot of features as Retry and MemOK!

[Chino]

A lot goes on behind BIOS updates and the automatic rules gets changed a lot. That's one of the reason why I discourage users to update their BIOS if their system is perfectly stable. With that said, your settings on one revision will not carry over to a newer revision. For example if you used a value of 1.12V for System Agent voltage on one revision, you might need to tweak it again.

On a sidenote I am using that exact kit on the Maximus VIII Extreme. Let me install it on the Rampage V Extreme when I get home and I'll let you know what I find. 🙂

[mdzcpa]

Chino wrote:
A lot goes on behind BIOS updates and the automatic rules gets changed a lot. That's one of the reason why I discourage users to update their BIOS if their system is perfectly stable.

THIS ^

Unless there is a new feature you absolutely must have, updating the BIOS is never a good idea if your system is already stable. Sure, the first few BIOS revisions can bring some tweaks to added stability and maybe overclockability, but after that there is no reason to update unless you need to.

I'm still on 0802 😮

[kaese]

Hi Chino,
I would be very much interested in hearing from your tests with the very same Kit.
Let me remark, that if auto-settings change a lot from one version to another, the settings dont seem to be well under control by the manufacturer. If that is the case (and I believe there may be many reasons, such as huge variance of available RAM Kits in combination with "individual" CPUs), sharing of detailed information for an enthusiast mainboard would always be highly appreciated.

By the way, BIOS V1601 runs the kit @2666MHz with V_SA on auto, setting a quite low voltage of 0.85V (HWinfo64). My RAM modules are manufactured week 34/2015 (both kits).

[Chino]

kaese wrote:
If that is the case (and I believe there may be many reasons, such as huge variance of available RAM Kits in combination with "individual" CPUs), sharing of detailed information for an enthusiast mainboard would always be highly appreciated.

You've said so yourself. With so many CPUs out there, it's impossible to sample them all to define a more accurate scaling routine. What they do is sample a few hundred of them.

Anyway, here are few benchmarks with the FURY DDR4-2666 32GB kit. I'll drop some HCI Memtest or Google Stress App runs later on when I get the chance.

[Techotic]

Just reflash back to 1601 and save yourself the headache. The new microcode won't do anything for you at the moment. I'm still on 1502, and everything is running flawlessly.

[Korth]

The simplest solution would be to stick with BIOS 1601 "forever". No need to flash new firmware until it actually does something good for your system. This isn't really that bad an option if no new BIOS updates ever do anything useful for you anyhow.

Or hope that whatever compatibility issue the BIOS 1701 microcode update introduced will be fixed in a subsequent BIOS release. Maybe your feedback will actually provoke Asus into fixing this new compatibility issue. Good luck.

Or just use different RAM. Costs money but solves the problem forever - or at least solves the problem until another Asus BIOS breaks compatibility again.

[kaese]

Actually, I appreciate a lot, Asus continuously publishes new versions of BIOS for the RVE. By the way, being told to stuck forever to a prior BIOS version does not feel like an adequate solution to me for an high-end OC flagship board, especially for tech enthusiasts.

Thanks for testing, Chino, can you confirm, BIOS V1701 worked with XMP auto settings without further tweaking?

Anyway, I am now very happy, I found stable settings with BIOS 1701 - even running two 32Gb kits at the same time (what is a completely different story of course)! I guess GoNz0- was on the right track with the 100 Mhz Blck. For anybody interested, please find the detailed settings in the images. Possibly, some of the voltages could still be optimized but I can work with it for now at 24/7.


There are a few things I wondered:

1. It is necessary to give the cache voltage a slight offset, otherwise the CPU seems to have problems keeping up RAM clock in or switching to idle (= C1/C2-states). I do not fully understand that: isnt the IMC dependent rather on system agent voltage? Does it make sense?

2. Regarding the V_Cache Offset mode: According to the BIOS-text message (Total Adaptive Mode CPU Cache Voltage, it is set to 1.07V+0.1V=1.17V. Is this correct? Because HWInfo64 (v5.06-2640) shows only 1.05 V.

3. I have to turn of memory check of at startup - however I believe it is stable (after using the settings for 72 h with several stress tests) otherwise I get a bd q-code error.

4. RAM-Voltage is set to 1.285V in BIOS but HWInfo and BIOS it shows higher values (typically 1.303 V, never less than 1.297V). Event though I left the RAM Eventual Voltage on auto, it is weird as the eventual voltage should if only apply during startups, isnt it?

Any suggestions very welcome.