I thought we might see an update to keep VSA and VIO from going so high when auto is used. Mine will go to or near 1.4 with auto.

brentsg wrote:
I thought we might see an update to keep VSA and VIO from going so high when auto is used. Mine will go to or near 1.4 with auto.

Some CPUs need it that high. As always, when you're pushing high memory speeds, manual tuning is needed if you want optimal settings for your CPU.

Raja@ASUS wrote:
Some CPUs need it that high. As always, when you're pushing high memory speeds, manual tuning is needed if you want optimal settings for your CPU.

My concern is with the MB pushing those voltages to unsafe levels while on auto. I can manually tune them to 1.2 or so and I can pass memory stress tests all day long, but the MB will set them close to 1.4 if I let it do auto.

brentsg wrote:
My concern is with the MB pushing those voltages to unsafe levels while on auto. I can manually tune them to 1.2 or so and I can pass memory stress tests all day long, but the MB will set them close to 1.4 if I let it do auto.

Like I said, some CPUs need that voltage for the speeds. If yours works with less, configure it manually.

I had set both to 1.2v (with the xmp settings active) and it seemed to fix the boot issue, but then my screens glitched out in Windows. Does that mean the voltages are too high? I noticed setting them to 1.2 (and saving and restarting) actually read a bit higher in bios, around 1.24 or so.

When I can get into BIOS from a cold boot, the voltages are:
VCCIO = 0.984
SA = 1.072

When I get into Windows and look HWiNFO, the voltages are:
VCCIO = 1.168 (steady at idle)
VCCSA = 1.144 (spikes at 1.152 idle)

When I run RealBench stress test and look at HWiNFO, the voltages are:
VCCIO = 1.176 (steady)
VCCSA = 1.152 (steady)

Would it make sense to set the voltages to 1.176 & 1.152 in bios? I don't know if settings those numbers exactly will keep those numbers there or end up higher.

While this error 55 code occurred occasionally before, it started happening every single time after I upgraded to a newer UPS battery backup. I wouldn't think this would cause any issues, considering both batteries just pass the outlet power directly to devices. I had my PC unplugged for most of the day when I swapped batteries.

I've had zero issues in Windows 10, no matter what I'm doing (gaming, rendering, stress testing), save for that screen glitch when I raised the voltages to 1.2. It went back to flawless stability when I put them back to auto. So I at least guess the RAM is good.
I just don't know what is causing the RAM to be undetected during boots and reboots UNLESS I perform a hard power down beforehand. Something to do with power? Fast Boot is disabled, so it should be a Normal Boot either way. I don't consider myself an expert at all, but I do think it's voltage related. That's why I was hoping for an update that regulates the voltages better.
If you can find an easy fix for me, I would appreciate it.

In the meantime, I may just put my PC to sleep at night instead of shutting down. It's not ideal, as I prefer a fresh boot every day, but it beats killing the power every morning.

FDSage wrote:
I had set both to 1.2v (with the xmp settings active) and it seemed to fix the boot issue, but then my screens glitched out in Windows. Does that mean the voltages are too high? I noticed setting them to 1.2 (and saving and restarting) actually read a bit higher in bios, around 1.24 or so.

When I can get into BIOS from a cold boot, the voltages are:
VCCIO = 0.984
SA = 1.072

When I get into Windows and look HWiNFO, the voltages are:
VCCIO = 1.168 (steady at idle)
VCCSA = 1.144 (spikes at 1.152 idle)


When I run RealBench stress test and look at HWiNFO, the voltages are:
VCCIO = 1.176 (steady)
VCCSA = 1.152 (steady)

Would it make sense to set the voltages to 1.176 & 1.152 in bios? I don't know if settings those numbers exactly will keep those numbers there or end up higher.

If the system failed to POST, the initial voltages displayed in UEFI will be default, hence won't match up with those of a successful POST at the overclocked memory config.

FDSage wrote:

While this error 55 code occurred occasionally before, it started happening every single time after I upgraded to a newer UPS battery backup. I wouldn't think this would cause any issues, considering both batteries just pass the outlet power directly to devices. I had my PC unplugged for most of the day when I swapped batteries.

Difficult to say if that's related. You'd need to debug thoroughly. I would think it's more of a coincidence than anything.

FDSage wrote:

I've had zero issues in Windows 10, no matter what I'm doing (gaming, rendering, stress testing), save for that screen glitch when I raised the voltages to 1.2. It went back to flawless stability when I put them back to auto. So I at least guess the RAM is good.

If it's failing to POST successfully, it means the memory isn't completely stable. You'd need to do more testing and apply more voltage to SA/IO to work out if it's making matters worse. I'd apply as much as 1.30V for both and then start thinking about it.

FDSage wrote:

I just don't know what is causing the RAM to be undetected during boots and reboots UNLESS I perform a hard power down beforehand. Something to do with power? Fast Boot is disabled, so it should be a Normal Boot either way. I don't consider myself an expert at all, but I do think it's voltage related. That's why I was hoping for an update that regulates the voltages better.

This has nothing to do with voltage regulation in the electrical sense, which is a different topic altogether. This is related to the interaction between the CPU's IMC and the memory modules, which in your case are failing to train consistently.

FDSage wrote:

If you can find an easy fix for me, I would appreciate it.

The easiest fix for someone who wants things made simple would be to tune SA/IO voltages and to increase DRAM voltage. Anything I do will be more related to memory timings and may require additional work from the end user.

I see. I guess I understand my options from here (raise voltages or adjust timings), but I guess I don't understand why it isn't stable without having to tune anything.

Would it have helped me if I had gotten different RAM? I could understand If I didn't do my part in picking out modules that were known to be compatible.

If I decide to upgrade my modules in the future, what should I look out for in terms of compatibility? I'll probably want something even faster if I'm upgrading.

I probably won't have any more questions on the subject after this unless you or someone happens to find some stable configuration. Otherwise I won't have any more comments on this issue unless I've found something that works.

FDSage wrote:
I see. I guess I understand my options from here (raise voltages or adjust timings), but I guess I don't understand why it isn't stable without having to tune anything.

Thats how overclocking works; its variable. Anything faster than stock speeds can require manual tuning. The further from stock speed you stray, the more chance that you'll need to put in some work.

FDSage wrote:

Would it have helped me if I had gotten different RAM? I could understand If I didn't do my part in picking out modules that were known to be compatible.

Something slower perhaps, yes. On four-DIMM boards that use T-Topology, running four single-sided memory modules can be easier than running two double-sided modules. Lots of variables, though.

FDSage wrote:


If I decide to upgrade my modules in the future, what should I look out for in terms of compatibility? I'll probably want something even faster if I'm upgrading.

For someone like you, I'd advise going with slower rather than faster kits. Going faster = potentially more work.


I have a 3600 CAS 17 kit here, which Im going to try today. I may find a few things, but I can't tune your system for you. Just something you need to be prepared for. Or you can simply try tuning/increasing voltages til it's stable.

-Raja