Gadgetory


All Cool Mind-blowing Gadgets You Love in One Place

Does Ryzen Really Need Fast Memory? Guide for Gamers

2017-04-20
hey guys welcome back to harbor unboxed today's video has nothing to do with risin hmm that doesn't see what it could be right no that's wrong today's video we're actually looking at rising memory performance again now you might be thinking hang on a minute Steve didn't you make me sit through a 13-minute video not two weeks ago that disgust Rises memory performance what more needs to be said well firstly thanks for bolstering the channels watch time YouTube might actually start serving me some good ads now and I'll be able to eat three meals a day so the reason we're doing this all over again is because the previous video looked at risings memory performance from a completely different angle the idea behind that video was to investigate claims that rise and scales better with higher clocked memory than say KB Lake does it was suggested that using ddr4 3,600 memory would see risin 7 overtake the core i7 77 okay in most games anyway this turn out to be completely false as it was discovered in every single game tested except for ashes at the singularity escalation that the 7700 K under clock to 4 gigahertz while still faster than the 1800 X clocked at the same frequency now the point isn't to make Rison 7 look bad in fact given the choice of the core i7 7700 K or the rise in 7 1700 I would go with horizon every time instead the point of all this rise in testing from day one has been to try and learn what the strengths and weaknesses of this new processor architecture are how it works and what the consumer can do to get the most out of their setup anyway so we've established that it's perfectly fine to bench mark Rison at the same memory frequency of an intel competitor and this is because for the most part the margins gonna be much the same whether your test with ddr4 3200 or 2133 for example in the previous video we saw pretty consistent scaling and almost all the games tested right up to ddr4 3200 using the gtx 1080 TI as a result the takeaway from most seem to be they would need to invest an expensive high frequency memory to get the most out of their rising CPU and well given the results I showed I can certainly see how you'd come to that conclusion the problem though being that most probably won't buy a GeForce GTX to ATI graphics card statistically something like the GTX 1060 or rx 480 is far more likely this is a problem because with these graphics cards gamers are heavily GPU bound in almost every game they play providing to have a decent quad-core CPU of better so the advantage higher clocked memory hands the CPU isn't realized with a slower graphics card for gamers what this means is purchasing the 175 dollar u.s. g.skill memory that we use might not actually translate into any additional performance when compared to the 100 dollar version rated at ddr4 2400 that money then would be far better invested in a faster GPU or maybe a CPU or perhaps even more memory capacity okay so at this point I've waffled on quite a bit you probably now realize that you might not need the fastest memory available but you probably still don't know what kind of memory you actually need to answer that we need to jump into the benchmarks but before we do that here are a few quick details all testing was conducted using the horizon 7 1800 X processor clocked at 4 gigahertz but the findings will be true for any Rison 7 or verizon 5 processor for that matter in total I have tested 6 demanding titles with the geforce gtx 1060 gtx 1070 and gtx 1080i the memory speeds will be comparing our ddr4 2133 2666 and 3200 ok that's everything let's check out what we found Mass Effect Andromeda is primarily a GPU bound game and we find this when testing with the gtx 1080i that said the upgrade from ddr4 2133 to 2666 did boost the minimum or the 0.1% result by 6 percent so that's a reasonable gain however when you re test with the gtx 1070 we see no such gains just one FPS separates the fastest configuration from the slowest and that's margin of error stuff the same is true when using the gtx 1060 here we see the memory frequency has no noticeable impact on the performance at all heman is a very CPU intensive game and right now it throws up some interesting results or rather slightly confusing results with Rison for whatever reason rising heavily limits the performance of nvidia is higher in GPUs in this title it's either a display driver issue or a lack of optimization on the games we have but whatever the case the average framerate is a bit misleading here it's not often you'll see the GTX 1060 trailing the gtx 980ti by little over ten percent the 1080i should be well over twice as fast when fully Unleashed anyway if we look at the minimum reported frame rate we see that Rison does benefit from the higher clocked memory even with the gtx 1070 though ddr4 2666 is coming close to extracting the maximum performance here when it comes to the gtx 1060 the minimum is much the same regardless of the memory speed though we do see a slight bump in the average frame rate still the variance between 21 33 and 3,200 memory is just 8% with the 1060 opposed to 17% with the 1070 mafia 3 is another good example of a game that's not just heavy on the GPU but also the CPU previously when testing with the gtx 1080i we saw massive performance gains from using the higher clocked memory the boost from 21 33 to 3,200 notice almost 40% more frames which is quite incredible however before you race out and spend big on high-speed memory stop and think what kind of graphics card you will be using here we see when using the gtx 1070 that previously seen gain of almost 40% evaporates to virtually nothing the minimum frame rate was improved by almost 10% when going from 2133 to 3200 though there was virtually nothing to be gained when upgrading from 2666 now with an even slower graphics card when using the gtx 1060 we see there are no gains to be had at all the gtx 1016 ir x 480 are hardly what I would consider slow and many gamers will be using less capable graphics cards so for these users memory frequency really isn't a big deal but let's move on and check out a few more games before making our final conclusion Deus Ex mankind divided isn't a heavy CPU user but even so we found when using the GTX 10 atti memory speed matters the difference between ddr4 2133 and 3200 was noticeable as we saw a 14% increase in the average frame rate and the all-important minimum was also boosted by 13% now with the gtx 1070 those glorious gains are gone the average is within the margin of error and the same is true for the 0.1% frame time when comparing ddr4 2666 and 32 100 once we filtered down to the GTX 1060 there's no real gains worth talking about and everything is again within the margin of error battlefield 1 is another game that saw huge performance gains when using the higher clocked memory with the GTX 10 atti however once we dropped down to the gtx 1070 those gains are nowhere to be seen as the rise in CP was able to attract the maximum performance out of this graphics card using ddr4 2133 that being the case we obviously found the same thing with the gtx 1060 wrapping up the testing we have ashes of the singularity escalation which has been tested using the high quality preset as you can see we are heavily cpu-bound in this test when using the gtx 1080 TTI as a result the gtx 1070 provided the exact same performance when using ddr4 2133 and 2666 memory it wasn't until we moved to 3,200 memory that the 1080 TI configuration started to pull away because this is such an extreme example of a CPU bound game memory frequency makes a huge difference even when using the gtx 1070 that said though dropping down to the gtx 1060 we did see little to no difference between the various memory frequencies for the average framerate we see that ddr4 2666 was 4 percent faster than 2133 though moving from 2666 to 3200 Netta's just two extra frames so even with an extreme example such as ashes of the singularity memory speed makes little difference for those using a mid-range graphics card so as you've just seen depending on the graphics card used and the games you intend on playing investing in premium high frequency memory might not be the best use of your hard-earned cash let's take a look at the average figures across five of the six games tests that I am excluding ashes of the singularity here as that was measured a bit differently this reflects pretty well what we just saw using the gtx 1080 TI arisin allowed for around 17% more performance meanwhile that margin was reduced to almost nothing with the gtx 1070 as we see a negligible 5% performance boost then with the gtx 1060 we're working well within the margin of error at most a 3% gain can be seen when boosting the memory frequency by 50% right so those rocking a mid-range graphics card don't really need to worry about memory frequency you're better off saving your money and expanding the memory capacity or perhaps even upgrading another component as it stands ddr4 2133 and 2400 memory are fairly evenly priced in fact in a number of cases the slower memory actually costs a little more looking at G skills flair X range the 16 gigabyte dual channel 2400 kits currently cost around a hundred and fifteen dollars the same price is the 2133 stuff while the 3200 version we use is priced at one hundred and seventy-five dollars u.s. of course it's possible to overclock ddr4 memory so stuff that's rated at 2400 can often be pushed to 20 666 or 2933 or even higher by adjusting the memory timings and voltages that said as always with overclocking your mileage will vary depending on the type of memory you have your motherboard and even the quality of your CPU right now based on pricing and of course the performance just seen in this video ddr4 2402 2666 looks to be the sweet spot and this is certainly the memory of frequencies I would recommend for anyone running a gtx 1070 or slower likewise these memory recommendations also apply for anyone using a verizon 7 or aizen 5 processor as well as anything on the intel side such as the core i3 Core i5 or even core i7 well I hope that clears up the memory confusion for consumers and as always if you have any other questions feel free to drop them below and I'll do my best to answer them as quickly as possible I'm your host Steve thanks for watching and I'll catch you again soon you
We are a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for us to earn fees by linking to Amazon.com and affiliated sites.