Gskill ripjaws b die 32gb cas14 3200mhz kit F4-3200C14D-32GVK overclocking

[Brando]

Just got this on sale and playing around with it. https://www.newegg.com/g-skill-32gb-288-pin-ddr4-sdram/p/N82E16820232218

So far it's going pretty strong at 3600mhz cas15 but it doesn't want to do 4000mhz cas17 or 4300mhz cas18 like my 16gb tridentZ kit. Is this an IMC thing or just lower binning for ripjaws or luck of the draw? I already tried bumping ram volts to 1.45v which is almost 1.5v on my mobo and bumped vccio/vccsa by .01v from 1.1 and 1.15 to 1.2 and 1.25v. Couldn't even get into windows at 4000mhz but 3600mhz cas15 passes a quick windows memory diagnostic. I haven't done deeper testing to verify stability but it's close. Kinda weird. Here's where I'm at now.

 

[Bustos]

Bought the same kit, subbed.

 

[Brando]

So after a little googling it seems that it's normal for the imc to work harder for the same speeds on 16gb sticks compared to 8gb sticks. I don't think it's gonna be worth cranking the vccio and vccsa to get more speed since it doesn't help me at the moment but it's nice to know that future cpu's with better imc's will probably handle it better. If I'm off on this let me know.

TLDR: higher capacity ram sticks are harder on the memory controller and I forgot

 

[Woomack]

16GB modules are harder to OC. Most of them won't run stable at 4000+ at reasonable voltages etc. It also depends on the motherboard and some other things. IMC will throw you a 23 error or something else from 2x if it can't run any higher (or will require higher voltage).

Here are some of my tests on 4x16GB Samsung B
https://www.overclockers.com/forums...llistix-Elite-4x16GB-DDR4-3000-BLE16G4D30AEEA

The same kit can make 4000 17-17-17 1.35V on Z390/X299 but can't make much more or it was a matter of my setup. For sure 7900X has a weak IMC which can't make more than ~4100 but on Z390 should run higher and I couldn't anyway.

I have 4x16GB Micron E kit which runs about the same so 4000 CL17-17-17 and ~4100 max but on that 7900X with weak IMC so there is a chance it will go higher. This one is a keeper as not many 8GB Micron E modules can do that.

 

[Brando]

Thanks for the info. I think i heard about this at some point but somehow i thought it was only an issue with more than 2 sticks. Oh well i wasnt going to run them like that anyhow. It was just curiousity. Btw i saw in your review where you said theres no software to check memory ic type. Check out thaiphoon burner.

 

[Woomack]

TB didn't work when I was testing my memory and it didn't work on any of my X299 mobos. That review has 2 years already ... btw I was in touch with TB creator, not so much recently as free version is enough for my needs. I won't hide I was tired of asking for a new license for every motherboard on which I was testing hardware (and I'm switching them way too often). The software is generating a different key based on your motherboard so can't really use the same license on more rigs.

 

[Brando]

So just to be clear are you saying its the memory itself thats hard to oc or are you saying its the strength of the memory controller?

 

[Woomack]

Memory itself is hard to OC past some clock. It's because of the higher density of the IC. You can see that single and dual rank (so 8 and 16GB modules) memory kits based on exactly the same IC have different max OC results. In stores top 2x16GB kits are at ~4200, top 2x8GB kits are at ~4800, top 4x8GB are at ~4600. It's easier to OC 4x8GB than 2x16GB what usually is caused by memory itself but IMC/motherboard is affecting that too.

There is also chipset/IMC factor. Not all motherboards will handle higher density IC above some clock and not all IMC will run above some clock. At least the latest series processors can handle 16GB modules at least up to 4200-4400.

 

[Brando]

I might grab a 9900ks when they come out so I guess I'll find out then if a better imc can get some more out of it but I think 3600 cas 15 is just about perfect so I'll be happy enough either way. 4000 is just such a nice round number.