Does having two sets of different frequency RAM DIMMS cause instability?

[birdflyer]

Greetings community,

I recently have been reaching a bottleneck with my current 32GBs of RAM (I do video editing and animation). Currently I have Corsair Vengeance RGB PRO 32GB (2x16) DDR4 3200mhz C16 memory. I upgraded this from what I had originally, Corsair Vengeance LPX 16GB (2x8GB) DDR4 3000MHz C15.

 

If I where to add these older dimms into the system, would it cause stability issues?

 

I assume I would have to clock back the newer memory to 3000mhz but would I have to change something else as well being that one set is C16 and the other set C15?

 

A little more info on the system:

CPU: Ryzen 7 2700X

GPU: AMD Radeon RX590

MB: ASUS ROG Crosshair VII Hero X470

 

My editing Suite:

DaVinci Resolve 17.1

Affinity Photo

[Hairless Monkey Boy]

You can try running all of the RAM together at the slowest common speed, which would be DDR4-3000 CL15. It may work, and it may not. You'll just have to try it to see.

 

Important to note DDR4-3000 CL15 is equivalent in terms of latency to DDR4-3200 CL16 because:

 

(CL15 / 3000MHz) * 2000 = (CL16 / 3200MHz) * 2000 = 10ns

 

But do pay attention to the other timings, though. You can reverse engineer the formula above to find the latency in ns for each one, and set the timings accordingly.

[birdflyer]

1 minute ago, HairlessMonkeyBoy said:

You can try running all of the RAM together at the slowest common speed, which would be DDR4-3000 CL15. It may work, and it may not. You'll just have to try it to see.

 

Important to note DDR4-3000 CL15 is equivalent in terms of latency to DDR4-3200 CL16 because:

 

(CL15 / 3000MHz) * 2000 = (CL16 / 3200MHz) * 2000 = 10ns

 

But do pay attention to the other timings, though. You can reverse engineer the formula above to find the latency in ns for each one, and set the timings accordingly.

Wow awesome! So setting it to 3000mhz cl15 should be perfectly fine with no stability issues? Also, when trying the different speeds should I run something like memtest to ensure there are no errors?

[Hairless Monkey Boy]

2 minutes ago, birdflyer said:

So setting it to 3000mhz cl15 should be perfectly fine with no stability issues?

That is not what I said.

Quote

It may work, and it may not. You'll just have to try it to see.

 

2 minutes ago, birdflyer said:

Also, when trying the different speeds should I run something like memtest to ensure there are no errors?

Yes. Run memtest86 through all 4 passes. Then, if you want to be more confident, you should run the HCI Design memtest, too. With the HCI Design memtest, you should run 16 instances (one for each CPU thread) each testing 2.25GB (adding up to 75% of your 48GB system RAM) up to at least 300% completion.

[birdflyer]

4 minutes ago, HairlessMonkeyBoy said:

That is not what I said.

 

Yes. Run memtest86 through all 4 passes. Then, if you want to be more confident, you should run the HCI Design memtest, too. With the HCI Design memtest, you should run 16 instances (one for each CPU thread) each testing 2.25GB (adding up to 75% of your 48GB system RAM) up to at least 300% completion.

Ok, will do, thanks. I am a bit of a noob in the RAM department

4 minutes ago, HairlessMonkeyBoy said:

It may work, and it may not. You'll just have to try it to see.

Got it.

[birdflyer]

5 minutes ago, HairlessMonkeyBoy said:

That is not what I said.

 

Yes. Run memtest86 through all 4 passes. Then, if you want to be more confident, you should run the HCI Design memtest, too. With the HCI Design memtest, you should run 16 instances (one for each CPU thread) each testing 2.25GB (adding up to 75% of your 48GB system RAM) up to at least 300% completion.

I do have one more probably dumb question:

Will doing this have a chance of hurting anything in the system or no more than usual?

[Hairless Monkey Boy]

10 minutes ago, birdflyer said:

Will doing this have a chance of hurting anything in the system or no more than usual?

Not physically, no. But if you run very unstable RAM in your system, there is a chance you can corrupt your OS. So you should try not to boot into your OS until you verify stability with memtest86. But don't stress about it too much if you boot into Windows by accident. It's probably fine.

[birdflyer]

Just now, HairlessMonkeyBoy said:

Not physically, no. But if you run very unstable RAM in your system, there is a chance you can corrupt your OS. So you should try not to boot into your OS until you verify stability with memtest86. But don't stress about it too much if you boot into Windows by accident. It's probably fine.

Awesome, thanks once again

[birdflyer]

17 hours ago, HauntedBob said:

As hairlessmonkeyboy said it should be fine if you run the faster memory identical to the slower, but I'd like to add you may also need to adjust sub timings to get that 1 to 1.. Hopefully your bios has those options. It's possible the slower memory has tighter sub timings than the faster so you need to see the full SPD of both and find the common low points.

I assume that because I have an ASUS ROG Crosshair VII Hero that it would have those options in the bios. What they are labeled, called, and where they are found is the issue I might have as I don't often mess with ram settings except for initially setting XMP settings (or AMD equivalent to XMP, whatever they call it).

[birdflyer]

@HauntedBob and @HairlessMonkeyBoy, I may be able to trade with someone for the exact same RAM kit that I have but 3200mhz C16 instead of the 3000mhz C15. It would still be Vengeance LPX that would be paired with the current Vengeance RGB Pro, but at least would be the same speed and latency. Should I consider trading for that or just try with what I've got? Also if I did trade for that, I should probably still run all the tests to ensure that both sets of DIMMS play nice with each other correct?

[Hairless Monkey Boy]

9 minutes ago, birdflyer said:

Should I consider trading for that or just try with what I've got?

Neither both being Corsair Vengeance, nor both having the same XMP rated speed and CAS latency makes them any more likely to work nicely together. The RAM you already have may be more similar than the trade. I'd try with what you have first, and if it doesn't work then you have nothing to lose.

9 minutes ago, birdflyer said:

Also if I did trade for that, I should probably still run all the tests to ensure that both sets of DIMMS play nice with each other correct?

Yup. Like I said, they are no more likely to work well together based on the information we have.