How much of a difference between CL30 and CL36 ram?

Posted by Pufferfisho@reddit | buildapc | View on Reddit | 16 comments

Hi, I'm building my first 1440p gaming pc with a Ryzen 5 7600x and RX 9070 xt.

I was looking at what ram to buy and found the Corsair vengeance ddr5 32gb (16x2) 6000, both CL30 an CL36. I know CL30 is better, but for me right now there is a 100€ difference in price.

So, is the difference in performance worth the price? Or am I good with the CL36? Also, would you suggest trying to wait for the price to go down (maybe for black friday) or buying right now?

I'm also open to suggestions if there is a better ram than Corsair vengeance.

[-]

XGreenDirtX@reddit

Latency = CAS latency * (2000 / frequency)

So 30(2000/6000) = 10ns 36(2000/6000) = 12ns

Is that worth 100 bucks? Not to me.

[-]

Plane-Produce-7820@reddit

Not much.

As an idea I have 5600 cl 36 ram and overclocked it to 6000 cl36 which real world is around 1/2 the difference between the same speed cl36 vs cl30.

I gained 1-3fps in minecraft only at 4K. Largest impact will be on cpu bound games.

[-]

Ishivara@reddit

You need to tighten up your timings, don't expect anything because the chips in your ram are trash

[-]

Plane-Produce-7820@reddit

You realise ram speed has a larger impact on gaming then timings do all else being equal.

Tightening the timings gets lower performance gains than pushing my ram speed.

[-]

Unless you have an Intel CPU then you are so wrong, you can't even imagine how wrong you are. And even if you have an Intel CPU going up with mts without going lower with timings has barely an impact compared to going up with mts and lower with timings

[-]

Plane-Produce-7820@reddit

Probably why I only look at AMD since that’s what I use.

[-]

Ishivara@reddit

Am5 CPUs have a imc that is barely considerable working, going 6400 C28 1:1 with extremely tight subtimings gives the same results in game and barely worse results in bench-like activities compared to 7800 C34 1:2 with extremely tight timings. As an example of one of the many CPUs I've tried, on a 7700 you'll get 53.306s on a yc 2.5b in 1:1 and 52.536s in 1:2 which translates in +0 fps going higher because the bandwidth is crap and the IMC is a joke

[-]

Beautiful-Musk-Ox@reddit

given the recent extreme price increases i would go with cl36 personally. You can see teh gaming difference here (at 1080p, ram is less to much less important at higher resolutions) https://youtu.be/Fr7Bfr-wPYw?t=640, cl36 will be inbetween the cl30 and cl40 examples

[-]

Ishivara@reddit

It's a bad video because it doesn't showcase the oc capabilities of hynix dies

[-]

Beautiful-Musk-Ox@reddit

they did apply buildzoid timings as one of the things they tested, but if you're going to OC then you should be watching other channels as hardware unboxed is for the average builder

[-]

Ishivara@reddit

No disrespect but buildzoid timings are way too conservative to be considerable proper oc. And it's obvious why: if you have to put them out in a video you need to minimize the amount of issues for people in order to avoid annoyances in the comments.

I've personally gone up to a 25% difference in cpu bound between Samsung and hynix m die (2x24) because I could just crank it with the manual oc

[-]

Ishivara@reddit

The difference between 6000 c36 and 6000 C30 is the chips ram has: the first one has Samsung or micron, the second one hynix. In terms of XMP very little changes, if you have a 3d cache CPU barely anything changes. However, if you want to do any manual oc you'll lose about 20% fps in CPU bound situations due to micron/Samsung not being able to go nearly as far as hynix dies. Manual ram oc is not relevant on 3d cache CPUs.

[-]