DDR3 RAM Upgrade - ANSWERED

[[☥] - CJ -]

Alright so.. I currently have Corsair Vengeance 4x4GB 1600mhz for 16GB and had it for a number of years now and since then ive known they werent the same revision, 2 are XMP 1.2 and 2 are XMP 1.3, which i assume has affected my performance to one extent or another, even if it is a tiny amount my plan is to get 2x8GB for 3 reasons, pretty cheap these days, less strain on the memory controller if any, and easier to troubleshoot.

So my question is
Should i get 1600mhz again or 1866mhz?

https://valid.x86.fr/nljhgs

[emoticorpse]

What are the prices/brand and any other info for each?

[Raoul]

Try to get 2133 or 2400 at C9 or C10

[[☥] - CJ -]

https://www.amazon.com/dp/B006EWUO22/

https://www.amazon.com/Patriot-1600MHz-Desktop-Memory-Heatsink/dp/B008LTJJJ0/

https://www.amazon.com/Patriot-1866MHz-Desktop-Memory-Heatsink/dp/B00453R90W/

My problem aside from not being able to choose between 1600 or 1866 is that i would PREFER to stick with Corsair as i have no experience with any other RAM brand and that i can get the Corsair set for $26 a month for 3 months through Amazon, but then the Patriot is cheaper as a whole at its current sale price..

and then theres G.Skill which i know is a good brand
https://www.amazon.com/G-Skill-Ripjaws-240-Pin-Desktop-F3-12800CL10D-16GBXL/dp/B0064DQR9U/

i'm also trying to make sure its a set my board supports

[It's Chase]

Just get whatever is cheaper. I can't imagine you would notice a difference between 1600mhz and 1866mhz. Both are brutally slow by todays standards.

Ideally, a whole platform upgrade would be the better choice instead of spending money on such old hardware that could possibly give out at any time.

[Bullwark]

Автор сообщения: ☥ - CJ -

i'm also trying to make sure its a set my board supports

Your best bet is to look at your board's and chosen RAM's brand QVL and take what's the best that is supported. That way you have a 100% chance of them working, because they're tested!
There you'll find all sorts of infos about speed, etc...

[Raoul]

Ahh I misread your OP thought you wanted to upgrade to 16GB, checking your validator you have lost no performance the revision doesn't matter and there is absolutely no reason to get another kit.

[kingjames488]

3 important points here...

1. if you have different ram they'll sync to the lowest spec, so if it's 1.2 and 1.3 they'll both run at 1.2 speed.
2. ram clock speed is not an important factor for anything but the most demanding applications, and since you're running old ram it's of 0 importance.
3. DDR3 is cheap now, just get whatever you can afford/your system can support at the base clock speed. capacity is more important than speed for you.

[emoticorpse]

I guess I'd go with the 1600 one (if I had to choose which looks like you do, lol).

Just because I'm looking for reasons to differentiate the two, I suppose that ram being a dollar cheaper doesn't hurt, plus the date first available on the specs says 2017 as opposed to the other one that says 2012 (not sure if those specs are even worth a damn, but I'll take it since I'm looking for anything) so maybe there's a chance it's performance has been refined or fine tuned/improved over the older released models. Plus 1600 to me looks more like a standardized speed to deal with where any mention of your ram might derail into a "how'd you land on the 1866 dude?" kind of talk.

I don't know. I'm just kind of talking I think.

[Illusion of Progress]

It's a pointless change.

Unless you actually have stability problems, or you are having to manually set the RAM lower than it is rated for, there is nothing being lost.

The different XMP versions don't matter unless they straight up don't work together. Are they working together now? If yes, then it's a non-issue. XMP is just a standard (with different revisions), not a performance measure itself. The performance is in the profile itself. 1600 MHz is 1600 MHz... so If you're doing 1600 MHz now, then it's working to full performance.

Less strain on the memory controller, again, non-issue if it's not unstable.

Extra reason for thought...

Going from four modules to two, you're going to go from at least dual rank, if not porabbly quad rank, down to possibly single rank most likely (or maybe 8 GB modules on DDR3 are still dual rank?). You will go down at worst, stay the same at best though. You won't notice the difference, and even measuring it might only show margin of error differences, but if we're splitting hairs over non-issues, this is technically something that will, on average, stand a chance of being decreased.

I had 4x 4GB[www.gskill.com] when I had LGA 1155 (bonus point; I had the same board [minus the "Gen 3" part] and CPU as you with that RAM). I still have 4 modules on my current platform now (and it's all dual rank modules so it's a quad rank configuration overall). I would never straight up make a free trade from that to 2x xxGB (especially if losing ranks), let alone pay for it. It's a downgrade, even if it's an inconsequential one. I'd only do that if the current configuration was barely running, or if I needed to go up in individual module capacity in order to upgrade total RAM amount.

So if you're not going to 32 GB and expecting that to enable you to stick with the platform for many more years, then save your money. This is borderline throwing it away.

[_I_]

the higher xmp just has more values that are changed when selecting the xmp profile
1600 cl9 is good enough for the 2500k, no reason to upgrade it unless you need more ram, but then a newer cpu/mobo would be better anyway

[Bad 💀 Motha]

Having multiple revisions of XMP won't cause performance issues.

For example in one of our systems here; ASUS 990FX Motherboard w/ FX8350
It had 4x 4GB. 1x Black Kit of 2x 4GB from Amazon and 1x Red Kit of 2x 4GB from Newegg; all of them are DDR3 1866 Corsair Vengeance. But one of the two sets is slightly newer, thus in CPU-Z is shows those having a slightly newer revision of XMP. However they do in fact have the exact same specs; i.e. Frequency + Timings + Voltage. So no it's not a problem at all. This system with all of that RAM has been running daily since beginning of 2013. It still runs like new, very stable. The CPU has also been running at 4.8Ghz the entire time. One of the RAM kits was installed first, then the other kit was added about 1 year later when I wipe the Win7 clean and installed the extra RAM and an SSD, clean installing the OS to SSD. It was wiped clean again around 2019 with clean install of Win10. Two family members use it from time to time and I myself use it sometimes as a means to test various games and drivers on it.

I also don't see any reason to use above DDR3 1600 on Intel 2nd or 3rd Gen. Most of those older boards were very picky about RAM to begin with. So safe bet to just stick with 1600Mhz RAM, which are dirt cheap. No point paying more for say 1866, 2133, 2400; even if the Motherboard says it supports it, it might not actually be stable when you populate all 4 Slots.

[[☥] - CJ -]

Автор сообщения: Illusion of Progress

It's a pointless change.

Unless you actually have stability problems, or you are having to manually set the RAM lower than it is rated for, there is nothing being lost.

The different XMP versions don't matter unless they straight up don't work together. Are they working together now? If yes, then it's a non-issue. XMP is just a standard (with different revisions), not a performance measure itself. The performance is in the profile itself. 1600 MHz is 1600 MHz... so If you're doing 1600 MHz now, then it's working to full performance.

Less strain on the memory controller, again, non-issue if it's not unstable.

Extra reason for thought...

Going from four modules to two, you're going to go from at least dual rank, if not porabbly quad rank, down to possibly single rank most likely (or maybe 8 GB modules on DDR3 are still dual rank?). You will go down at worst, stay the same at best though. You won't notice the difference, and even measuring it might only show margin of error differences, but if we're splitting hairs over non-issues, this is technically something that will, on average, stand a chance of being decreased.

I had 4x 4GB[www.gskill.com] when I had LGA 1155 (bonus point; I had the same board [minus the "Gen 3" part] and CPU as you with that RAM). I still have 4 modules on my current platform now (and it's all dual rank modules so it's a quad rank configuration overall). I would never straight up make a free trade from that to 2x xxGB (especially if losing ranks), let alone pay for it. It's a downgrade, even if it's an inconsequential one. I'd only do that if the current configuration was barely running, or if I needed to go up in individual module capacity in order to upgrade total RAM amount.

So if you're not going to 32 GB and expecting that to enable you to stick with the platform for many more years, then save your money. This is borderline throwing it away.

See thats the thing actually, on CPU Z 1 set is shown as dual rank and the other is shown as single

[Illusion of Progress]

Автор сообщения: ☥ - CJ -

[
See thats the thing actually, on CPU Z 1 set is shown as dual rank and the other is shown as single

That's actually possible, and gives you a rather uncommon configuration; your RAM is running in a triple rank configuration like that (dual rank in one channel and single rank in the other channel).

You would for sure go down in rank with a change to two modules then, because the most you could have with two modules is dual rank, and if the individual modules are single rank (not sure if they are with DDR3 but in DDR4 8 GB capacity has been single rank for a while now; I think even 16 GB is too now) then you'd go down to single rank in that case.

Ranks really aren't something you'd notice... but it's there.

I'd stay with your current RAM as it is since you haven't mentioned anything that seems to signify a reason that it's not working well.

XMP speeds and timings are working, it's 1,600 MHz which is decent for Sandy Bridge era platforms (same for the 9-9-9-24 timings at that speed), and it isn't unstable. You can't ask for more, and you don't have much to improve upon. It's even triple rank. The ONLY somewhat "low" thing I see is the "2N" at the end and if that refers to the "command rate" like I think it does, you'd really rather have that running at 1T. That might be a consequence of needing to run it there to have the two sets cooperate (?), but I've always had every RAM configuration I've used be able to do 1T. If that is a command rate running at 2T though it is not something major or worth changing the RAM over IMO.

If you need 32 GB then yes it would be worth considering these things as you'd be changing it anyway, but otherwise I don't see any issues.

[[☥] - CJ -]

Well i mean, the rank stuff and 1 set being newer than the other set kinda annoys me a little bit.. and there seems to be (seemingly) very minor timing differences between the 2 @ XMP 1600, tRC 41 vs tRC 33, not that i know what that means.. But the older set says its running Dual Ranks while the newer set says its running Single.

The older RAM set appears to be the one with tRC 41..the older RAM set is the one i got when i built the PC, the newer RAM set shows a date code of 2015, so its a couple years newer than the original which makes sense as i got it to help with performance on ARK at the time.

i dont know
my main thing is just to have everything be the same, and i thought using 2 sticks instead of 4 would be better overall, especially with my systems age/architecture etc etc..

as far as the command rate goes, thats only showing up when using XMP which i am, according to CPU-Z thats the only time 2T comes into play, other RAM speeds (JEDEC) dont have it.

[xSOSxHawkens]

As someone who has actually tested this allot in modern titles.

You want 1866 or better with tight timings.