• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

RYZEN DDR4 MEMORY, WHAT YOU NEED TO KNOW!

I've got 32gb (4x8gb) of G.SKILL Ripjaws V Series F4-2400C15D-16GVR 16 GB (8 GBx2) DDR4 2400 MHz C15 1.2 V that I'm thinking of re-using for my ryzen build.

Worth sticking with?
 
anyone know where I can get an AM3 backplate that'll let me mount my corsair H100?

edit: Ok the H100 uses the clip mounting mechanism, which is apparently already compatible with AM4, so it looks like im good.
 
Last edited:
If you can't afford the extra for faster 32gb I'd rather spend less and have faster 16gb if it's mainly for gaming. Those would be better even if for whatever reason you can't get 3200 working properly, you're still going to get 3000 with tighter timings.

Indeed i can't afford 32gb anyways so it has to be 16 gb should be enough for gaming, can always get another 16 gb to make it 32gb but won't be dual channel as it is 4 sticks of ram.
 
So first support for higher memory speeds as standard on the motherboard is only available on higher end motherboards, I show they are consistently available on cheaper motherboards, now it's well if you overclock well beyond the rated specs... so which is it.

The simple fact is, I've proven that the kinds of speeds like 3733Mhz are the max the high end boards support and the lower end boards support the same memory speeds. So if the £250 boards support 3200Mhz memory it's very likely that £100 boards will do as well.


Sorry to single you out here DM, this is aimed towards a few people involved in discussions on the impact of your motherboard on DDR4 performance.

The memory controller (on CPU die) and motherboard are the two critical factors in achievable DDR4 clock rates. Usually one will determine the limit of achievable clock, but keep in mind someones you get into complex emergent behaviours involving both.

If you assume the memory controller is not the limiting factor, the motherboard makes a significant difference. The reasons for the motherboard making a difference are summarised below:
  • The trace geometries are important for impedance matching, but this is basic design that all motherboards will have.
  • The geometry of your board layering is important for impedance matching, especially that of the dielectric layers. The actual dielectric used is also important. Variations in composition and thickness are not good.
  • The geometry of the ground/power planes is important, where any breaks or cuts close to your traces will change your impedance. These can be caused by simple things like vias or component leads.
  • Trace length matching is extremely important for these parallel interfaces. The higher the clock rate, the more important this becomes.
  • The power and decoupling provided on the motherboard.
  • Inter-trace geometry will effect your cross-talk and error rates.
Now some of these factors are unlikely to change between an expensive and cheap motherboard, but some will absolutely change.

Actually quantifying the cost vs. achievable clock rate is nearly impossible though. Some cheap motherboards will be designed by a magician and will be capable of 3Ghz (6Ghz DDR) clock rates, where us mere mortals working on expensive boards might achieve half that...

I would expect the level of verification to be higher in high cost boards, hence providing more consistent high clock rates.

EDIT: This is all to complex without a nice diagram! Look at the "eye diagram" below. The bigger the eye, the better the signal. Notice the top vs. bottom, aka good vs. bad board design!
http://m.eet.com/media/1166737/295166-tmw_eye_freescale_fig3.jpg
 
Last edited:
Back
Top Bottom