Why did adding identical RAM slow my PC performance.
Posted by Anphonio@reddit | buildapc | View on Reddit | 36 comments
Hey guys, I've had my pc for 2.5 years now, currently running an i5 12600k and an RTX 3060 TI with 2x8 GB DDR4 RAM. I've had no issues whatsoever this entire time. My brother has recently upgraded his PC including his RAM so I took his old modules which were identical to mine. I've put in the 2 additional RAM units so I now have 4x8GB DDR4 and I was surprised to see my PC performing worse. I've alternated combinations of pairs and all the units are perfectly fine but something is going wrong when all 4 are connected. My motherboard and CPU are both graded to handle up to 128GB of RAM, and I have an 850w PSU so it can't be a power issue. I don't quite see what the problem is. Any Ideas?
My guess is it could be something to do with overheating due to lack of airflow between the ram modules.
liaminwales@reddit
It helps if you say what problems you have hit, screen shots for bonus points.
Then say what RAM kit's you have, extra points for full model numbers. A lot of brands will mix memory chips in the same line, it may look like the same ram but end up being two different types of ram chips.
Lastly it's harder to run 4 sticks of RAM, it's more of a problem with fast sticks\~
Anphonio@reddit (OP)
Modules are all Corsair Vengeance LPX 2400MHz C14. I'm not sure where I can find model numbers but I have this saved (cmk16gx4m2a2400c14)
It's all hitching is the problem. Max FPS is staying the same but it's hitching significantly every second in any game I play.
NerghaatTheUnliving@reddit
I really don't get why this is getting downvoted. They requested more info, you provided it. Some Redditors can be real shitstains sometimes.
kyronami@reddit
because mixing ram can lead to instability/bad performance if the ram sticks came from separate kits even if they are the same exact ram
Anphonio@reddit (OP)
Really, so I would've had to have bought a quad pack to start with?
thebeansoldier@reddit
Nope. Less sticks = more stability. Best to get a 2 stick set. They’re not expensive anymore. You can get a 32gb 3200 for just over 50$
ProjectManagerAMA@reddit
Holy smokes! I didn't realize prices had dropped THAT MUCH! Time to upgrade my RAM!
JETgroovy@reddit
I did a big upgrade last year and put in 2x32gb of ddr5 5600, cost me like $150. I know that doesn't sound cheap but I'm set on ram for a long while.
TGC_Karlsanada13@reddit
No, get the 2x16 variant would be better.
ForThePantz@reddit
Memory controller struggles to keep up with 4 DIMMS; 2x16GB will be faster than 4x8GB. Bazinga.
Scarabesque@reddit
If 4x8 works stably (it does for me) it'll be faster than 2x16 single rank.
2x16gb dual rank ddr4 is pretty rare and expensive. Fairly certain all 2x16gb 3200cl16 kits you can but today are single rank.
CMDR_Fritz_Adelman@reddit
Also check if XMP profile is turn on after installing new ram.
DillerDallas@reddit
Try this first OP
geniuslogitech@reddit
it rly depends on memory topology of the motherboard, while your statement is true for daisy-chain mobos it doesn't work like that for T-topology, there it will work the exact same, a bit worse than 2x16 on a daisy chain or dual slot motherboard because trace length there is shorter, on T-topology they made it longer so both slots in a channel get same length
kyronami@reddit
well when you buy ram kits that have multiple sticks they are guaranteed to work together perfectly by whoever made the ram, so in general yes a quad pack is the better option VS buying two 2x packs and trying to combine them
OneTutMan@reddit
And it just looks better having them all filled out
geniuslogitech@reddit
depends on your mobo memory topology, maybe 2x16 is better maybe 4x8 and 2x8 is the same but those 4x8 would need to all be exact same, you probably got different "lottery bin" kit that looks the same on the outside but it's not the same inside, different chips, only if you buy higher end stuff you always get the same known chips
Ditto_is_Lit@reddit
While what's being posted in replies holds 100% true and it does vary on which cpu mobo combo as well, it's still possible to use unpaired kits. I had two identical kits (was stuck with the 2nd kit because I needed to use my PC while I got the other kit replaced in an RMA). I did manage to successfully use both kits full clocked with XMP but took some extra work to do so.
First step check your bios and see if you're on the latest stable revision and flash the board if need be. I would also test the ram with memtest on both kits separately with a single pass and again once combined. If you find errors in either kit test each dimm individually for the ones that give errors and RMA the problematic kit, you'll need the memtest log to provide proof when requesting the RMA so be sure to answer yes when prompted after the test completes.
With the previous bios I was getting massive hitching, pages crashing, hard locks, etc. Even memtest was giving errors on tests like it was a faulty kit. The stable bios changed everything and memtest ran full passes and was able to run as good as using a single kit in games and better in CAD. While it may not be a sure thing it could be worth the effort to try out.
Faux_Grey@reddit
Two things would help here:
Model of Ram sticks:
Model of motherboard:
You've most likely got two different sets with different timings or speed, perhaps your original RAM is 3200 CL14 and the new memory is 2133 CL20,
You should probably run a memory benchmark while testing different arrangements of memory to see that the problem is actually around memory performance.
painefultruth76@reddit
Reseat the Modules.
Same installation?
Windows paging file?
Did you increase your swap file size?
Run memtest?
Run 3 at a time to make sure you don't have a bad stick?
So many TS options you haven't walked thru, and I'm kinda liking one bad stick...the more I think about it.
sharia1919@reddit
As the others mention it is because the RAM blocks are not actually the same.
Let's say you buy ddr4 1600mzh, Cl14 ram.
But they actually contain many more settings/numbers would be 14-16-16-20. Those are different latencies. The second group you have from your brother may be instead 14-18-18-22 or something. So yes, they are both CL14, but their subsequent numbers indicate different performance.
And even if those numbers are still identical, then their may still be even more "hidden" settings/performance numbers that are different. This is why you always need to buy paired matches of RAM blocks. They sell them in groups of 1, 2 or 4.
But as the others say, even with 4 matched pairs, then the system itself is not happy about it, because it still needs to have all of the blocks to communicate and work together. And this often results in instability.
So the usual advice is to simple use 2 sticks, since you bypass this risk element altogether.
Significant_Apple904@reddit
4 RAM slots run in dual channel meaning slot 1,2 share channel 1, slot 3,4 share channel 2, by occupying all 4 slots, you now have double the traffic going through the same 2 channels, thus slower than before.
2 slots generally have much better capabilities of reaching higher clock speed (XMP or EXPO) and 4 slots generally can only run at stock speed.
It's highly recommended to run 2 sticks of RAM to fully utilize the architecture
--Dolorem--@reddit
4 ram sticks is sometimes slower than 2 ram sticks
dfm503@reddit
It’s technically faster to have 2 dimms but it shouldn’t be notably different. I would ensure that XMP is still enabled when using all 4.
geniuslogitech@reddit
depends on memory topology of the board, on t-topology it's exact same, on daisy chain it is a bit slower but shouldn't be that big of a difference if it's same chips so I guess he didn't get same chips in his lottery bin kits
Gregardless@reddit
Pull them all out and post pics. Make sure that you have each set separated. Put one in slots A1 and A2 and the others in B1 and B2. These are every other slot not right next to each other. So the first DIMM from one set in the slot closest to the CPU, skip one, then the other DIMM from the same set in the slot third closest to the CPU. Then put the other set in the open spots.
You can also check using CPU-Z that the memory chips on each set are the same. And check all the timings to verify they line up.
aagejaeger@reddit
I think you went from dual channel to single channel. Your board probably doesn’t do quad channel. You basically gained capacity but lost bandwidth.
BeavisTheSixth@reddit
4 sticks is still dual channel.
jerryeight@reddit
Did you enable the xmp or xpo ram profiles for the new ram? If you don't, the system runs all ram at the low default speeds.
Jakunobi@reddit
It's like others are saying. Dual channel RAM is the best with two sticks from the same kit. Mix-matching stick (eventhough same capacity and brand), or adding sticks on the free slots, is going to introduce some instability if you're unlucky. You just gotta forget about the extra sticks you got and shelve them or sell them. Upgrade your ram if you want more capacity. And don't just get 1 stick of 32gb if that's the size you want. Get a kit of 2x half the ram size you want. That'll help you get dual channel.
ryo4ever@reddit
What’s your motherboard model?
TimmmyTurner@reddit
check xmp settings in bios, they most probably aren't running at max speed
Anphonio@reddit (OP)
They were, I had XMP enabled.
Vizkos@reddit
Define "something is going wrong" and "slow my PC performance". Are you losing a few FPS, or does your PC have a lot of lag, for example?
Generally speaking, you'd want to scale up the memory size per stick and stay with 2x instead of having 4x. 4x adds more load to the CPU needing to manage 2x dual channel module sets.
Anphonio@reddit (OP)
Significant loss in performance when gaming. My FPS maxes at the same values as before but it hitches an absolute ton. I had XMP enabled on the new configuration too.
Soccera1@reddit
Send pictures