• Happy holidays, folks! Thanks to each and every one of you for being part of the Tom's Hardware community!

[SOLVED] My mixed RAMs are running at the FASTER speed?!

pouria19

Honorable
Jan 16, 2015
4
2
10,515
Hello everyone.
I have two 8GB RAM modules and I'm trying to test if I am able to use them together to get 16GB of RAM.
I've done my research and according to it I know that:
1- The best option is to buy RAMs that come in pairs from factory because mixing even identical RAMs is not guaranteed to work
2- You can mix RAMs with different speeds and timings and sizes, and if there isn't any stability issue you are good to go
3- If you mix RAMs with different speeds they will run at the speed of the slowest RAM
4- If you mix RAMs with different timings they will run at the slowest timing

The RAMs I'm using are these:
RAM #1: G Skill Aegis F4-2400C15-8GIS (15 latency)
RAM #2: G Skill Aegis F4-2400C17-8GIS (17 latency)

So I put these RAMs together to see what happens, and what happened was surprising. They were running at 2400MHz speed with 15-15-15-36 latency! The second RAM does not support this latency. It has a 17 latency. How is this possible?

Here are the screenshots of the different configurations I tried and the corresponding CPU-Z results:

-- Configuration 1 [ RAM#1 in slot 2 ]:

ceOwaQP.png

PTILPhL.png


Read benchmark:

XcVzUrF.png


Latency benchmark:

Evc9Ung.png


-- Configuration 2 [ RAM#2 in slot 2 ]:

ERHAr3z.png

NEZ41dl.png


Read benchmark:

b0pWtKa.png


Latency benchmark:

qgn6zzO.png


-- Configuration 3 [ RAM#1 in slot 2, RAM#2 in slot 4]:

l9V69Xn.png


Read benchmark:

GYbZP6a.png


Latency benchmark:

V3O3KgQ.png


-- Configuration 4 [ RAM#2 in slot 2, RAM#1 in slot 4]:

K447yPZ.png


Read benchmark:

CjN50JN.png


Latency benchmark:

eehREK1.png


As you can see the RAMs are actually working in dual channel. And as you can see when I put them both together their latency is 15, which is the latency of the faster RAM. According to my research they should run at 17 latency which is the latency of the slower RAM.
Another thing I don't understand is when I put the second RAM in slot#4 the total speed is 2400, but when I put the first RAM in slot#4 the total speed is 2133.

I'm just confused and want to make sure nothing is going to get damaged if I use these rams together(CPU, mobo, RAMs).
I have run tests with HCI design memtest and Memetestx86 and both reported 0 errors.
I did some gaming with no issues.
I did a Heaven GPU benchmark with no issues.
And I've been using the system since yesterday with no stability/hang issues.
 
  • Like
Reactions: Quanticriver
Solution
Thanks for your answers.
I'm not a fan of overclocking. If my RAM is being technically overclocked is it possible that it will have reduced life? Or get too hot?
Unfortunately the one with the slower speed doesn't have a thermal sensor. But I physically checked it after memtest and it wasn't too hot.

Technically, XMP is an overclock profile. So most memory you can buy is overclocked.
Heat and degradation? Most RAM can handle 1.50 volts without active cooling (a fan over them) except C-die. That can only take 1.35v. If you go higher, it should die.
RAM has an operating span of 0-85 C according to the spec. RAM gets unstable way before that. Somewhere between 45-55 C. So your computer should crash way before RAM overheats or...
You are completely good. It's call overclocking even if it is just tightening the timings a bit. You've run both memtests I would so you're solid. :) Enjoy!
i have learnt today!

prolly you get the mhz of the slowest one, but timings are a different matter
That is interesting
I wasn't expecting it to work based on what I read on matching RAMs. Perhaps I just got lucky?
btw is there a chance that CPU-Z is reporting wrong timings?
I used AIDA64 too but it reports the same results as CPU-Z.
Or maybe the 17 latency one is also capable of running at 15 latency @ 2400 but they aren't reporting it for marketing purposes?
 
A lot of times memory isn't tested to certain specs, but that doesn't mean it can't run it--this is why overclocking is a good thing as it can squeeze some extra performance out of the parts that are 'better' than most. Which I believe your ram fits that description. :)
Thanks for your answers.
I'm not a fan of overclocking. If my RAM is being technically overclocked is it possible that it will have reduced life? Or get too hot?
Unfortunately the one with the slower speed doesn't have a thermal sensor. But I physically checked it after memtest and it wasn't too hot.
 
  • Like
Reactions: SamirD
Thanks for your answers.
I'm not a fan of overclocking. If my RAM is being technically overclocked is it possible that it will have reduced life? Or get too hot?
Unfortunately the one with the slower speed doesn't have a thermal sensor. But I physically checked it after memtest and it wasn't too hot.

Technically, XMP is an overclock profile. So most memory you can buy is overclocked.
Heat and degradation? Most RAM can handle 1.50 volts without active cooling (a fan over them) except C-die. That can only take 1.35v. If you go higher, it should die.
RAM has an operating span of 0-85 C according to the spec. RAM gets unstable way before that. Somewhere between 45-55 C. So your computer should crash way before RAM overheats or degrades. You CAN kill RAM outright, if you run at lot of voltage, like 1.8-2.0 volts with zero cooling.

Air-cooling the CPU usually helps with RAM cooling too. I have an AIO so there is just about zero air moving over RAM. Front fans are too low down the case so I have rearfan blowing in towards the RAM. Radiator on top, sidedoor has a 120mm fan sucking air out right next to GPU.
 
  • Like
Reactions: SamirD and pouria19
Solution
Thanks for your answers.
I'm not a fan of overclocking. If my RAM is being technically overclocked is it possible that it will have reduced life? Or get too hot?
Unfortunately the one with the slower speed doesn't have a thermal sensor. But I physically checked it after memtest and it wasn't too hot.
Generally I don't do it either, but something like what you ran across is something I would accept without hesitation since it isn't that significant. Generally when overclocking, if something is running faster without setting modifications like in your case, there's absolutely no ill effects long term since it's basically running like stock. Checking the temp is a good thing as memory will usually start to have problems when it's over temp, and if it feels the same as the other module, you have no worries.
 
  • Like
Reactions: pouria19