Question ASUS M5A99X EVO R2.0 RAM configuration

[graveller39]

I'm pretty much a complete novice when it comes to RAM configuration, but I'm hoping someone can either confirm I've got things right or offer some suggestions for my setup.

I put together my system a few years back and have had no real issues, but I recently stumbled on the fact that my RAM has been running at 1333MHz as opposed to its rated speed of 1866MHz - I realize (now) this is normal, but I'd like to get the best performance I can reasonably achieve, bearing in mind that stability and reliability are still the priority.

This is what I've got:
ASUS M5A99X EVO R2.0
ASUS HD7750-1GD5-V2
AMD FX-8350
2x8GB G. Skill Sniper F3-1866C9-8GSR
SanDisk SSDSXPS240G
Windows 7 Pro
Several other internal & external HDDs for storage

I'm not a gamer and don't likely need or want to overclock the system - I use my PC primarily for music and photo editing and occasionally video editing and of course browsing.

So, after reading various posts here and elsewhere, I've changed the Ai Overclock Tuner from 'Auto' to 'D.O.C.P.' in the BIOS and tried both Profile 0 and Profile 1. Both profiles bump the RAM speed up to 1866 and change the timings to their rated values. Its not clear what the difference is between the two profiles - maybe some of the more advanced RAM settings?

I've left everything else on 'Auto' or its default setting. Is this sufficient? Or should I be tweaking any voltages or other settings? The system appears to be stable, although I've done no testing and not much real "work" yet either.

Any thoughts or suggestions are most welcome.

 

[Darkbreeze]

The difference could be as simple as whether it chooses a command rate of 1 or 2, or it could have completely different primary or secondary timings altogether.

Usually whatever the first profile listed is, is the one that you want to use, but not always. I'd probably recommend NOT using the DOCP profile in your scenario, but instead setting all of the primary timings, speed and voltage to what you KNOW the sticks are supposed to be at based on their advertised specifications, and leave everything else on Auto.

Also, make sure you have them installed in the slots which are in the second and fourth positions over from the CPU socket. Running them at their advertised specifications, while "technically" an overclock, isn't really overclocking them. Those are the specifications and settings that the manufacturer has determined them stable at, they are simply outside what the platform specifies as a default range, so they are labeled as an OC. If you go PAST those settings, then you are overclocking, especially if you try increasing the speed/frequency which requires extensive testing for stability.

 

[graveller39]

Thanks for the quick reply Darkbreeze, much appreciated.

So, yes, I've got the sticks installed in the 2nd & 4th slots and the D.O.C.P. settings appear to set everything to to their rated values - your suggestion to set the primary timings, speed and voltage manually is just to ensure they're running at their advertised specs?

I read elsewhere about adjusting the CPU-NB voltage +0.05v to +0.1v when setting the RAM - would this be worth trying? Or will the Auto setting take care of it?

 

[Darkbreeze]

Here's the thing. AM3+ OFTEN simply won't run at 1866mhz. This is known, and has been since shortly after the release of the platform. AMD has admitted as much. Those of us in the enthusiast and review sectors have verified it through the years. AM3+ has trouble sometimes with high speed memory, which 1866mhz IS.

So the fact that you are able to get these to run at 1866mhz WITHOUT an overclock on the CPU is a good deal. I'd probably do just exactly what I said.

Set the timings manually, leave the rest of the memory configuration on auto. I'd probably bump up the DRAM (memory) voltage by .005v above the stock voltage, whatever that is for your sticks, probably 1.5v (So run them at 1.505-1.520v range or thereabout depending on what increments your board allows you to adjust DRAM voltage by. One tiny bump up should do the trick in terms of ensuring stability, but testing is always a good idea using Memtest86, Windows memory diagnostic and Prime95 Blend mode with AVX and AVX2 turned off, or a custom setting specifically for testing of memory in Prime) to make sure they are good. They are probably good at the stock voltage anyhow but if you keep them at 1.5v I'd definitely want to test them using Memtest86 for four passes and Windows memory diagnostic.

Memory instability, which you cannot always see or know is even occurring, is real, and it's no joke. It will corrupt ALL of your data in weeks or a few short months depending on the level of usage.

You should read this:

How To - Troubleshooting problems with PC memory (RAM) and XMP profile configurations

In the beginning, there was......BIOS Before you go ANY further, go to the manufacturer product page for your specific motherboard model AND revision. Revision is an important aspect because for any given motherboard there may be more than one version of that model which will be identified...

forums.tomshardware.com

 

[gskill support]

You mentioned the system was stable, so if you have CPUz, check Memory tab to see if it is running at the rated specifications. If it shows DRAM Frequency ~933MHz, everything is good and you have no worries.

Enjoy

 

[Darkbreeze]

Generally speaking, I'd agree, however, in ANY case where you are running outside the JEDEC specifications in a configuration considered to be an overclock, as noted on the motherboard specifications as an "OC", it is a good idea to test stability. For obvious reasons, memory manufacturers can't test every memory module or kit, on every motherboard, with every possible CPU configuration that's out there.

So while they can definitively say that the probability is very high that a high speed configuration is stable based on the extensive testing they've done with that module on a variety of test platfoms, there are NEVER any guarantees. For that matter, it's a good idea to do your own testing anyhow, at any speed or timing configuration, just to be sure there is nothing whatsoever wrong with the sticks themselves. Not all problems show up as blue screens, black screen restarts or hard errors.

 

[graveller39]

Thanks again for the good info - who'd a thought computers could be so complicated...? go figure...

It may take me a bit to wade through your tutorial, but it sounds like it'd be a good idea to have a better understanding of things and to give the system a test before going much further. I'll do that as soon as I can and let you know how things look.

 

[Darkbreeze]

So, to expand on that a little and clarify let me say this. The G.Skill rep is correct when they say that if you are running AT or below the speed and timings (Well, above in the case of timings, since looser timings are actually (Usually, although there may be one or two secondary or tertiary timing where lower is looser) higher numbers while lower number indicate tighter, higher performance timings) dictated by the advertised profile, whether by manual configuration or by XMP/DOCP/AMP profiles, and it is running without obvious errors, showing as the correct speed and number of sticks as well as showing dual channel operation when you look at the CPU-Z memory and SPD tabs, then it's fine, at least temporarily, for use.

I simply recommend that you DO do the testing anyhow because it's always better to find out SOONER rather than later, if there is a problem with the memory physically or some minor timing or configuration issue exists so it can be corrected and made stable before any pattern of corruption to your data can occur.

 

[graveller39]

I know its been 4 years, but I guess I've been busy. Really though, some unexpected things came up and I just recently got back to this. In the meantime, I had left the RAM at the default JEDEC specs.

My OS and/or SDD had been giving me issues, so I bought a new SSD and thought it'd be a good time to finally upgrade to Windows 10. I also thought it'd be a good time to revisit this and do the testing Darkbreeze suggested.

I used DOCP Profile 0 and tested with Memtest86 and Prime95 as per the "How to" and all seems good.

Thanks for the advice.