Does RAM Speed Matter? — DDR4-3200 vs DDR4-3600 — Game & Productivity Testing on R5 5600X

hello and welcome to tech deals does ram speed matter one of the most frequently asked questions during our live streams is should i pay extra for high speed ram in today's video we're going to answer that question today's video is brought to you by urcd keys the best source for windows 10 and office professional product keys at deeply discounted prices more details at the end of the video ddr4 3200 vs ddr4 3600 is on the list of the top 10 most often asked questions the price difference used to be larger but as we enter december of 2020 the difference has dropped to below 10 so the short answer is if you're buying new ddr4 3600 probably makes the most sense given how little price difference there is however before you all run off there is more nuance to the answer than just that if you have a first or second gen ryzen cpu you are unlikely to get 3600 to work properly the support just isn't there 3200 is a much safer choice even with updated bios's and improved ram support in newer board releases this is also true of third generation ryzen cpus installed in 300 or 400 series motherboards while the memory controller is in fact on the cpu the trace designs and bios's were not really meant for faster ram and support can be hit and miss now you can check your motherboard's qvl or qualified vendors list to see if the specific ram you want to install has been tested however those lists are rarely updated years later so the qvl of a three-year-old motherboard installed with a one-year-old cpu is not likely to be all that useful a third or fourth generation ryzen cpu 3000 and 5000 series respectively installed on a 500 series board is likely to accept most 3600 ram kits without any trouble so aim for that speed if you want the best performance now intel makes all this much much easier intel's ram support is generally better than amd's given they designed the xmp or extreme memory profile system that we all use today and most ram was tuned to work on intel cpus if you own a 7th gen or newer intel cpu on a 200 series or newer intel board any 3600 ram kit is almost guaranteed to work now exceptions might exist but they will be rare ones 100 series boards and six gen chips are just old enough it might be a hit or miss affair they were designed at the very start of ddr4 and you're probably safer with 3200 on those for 9th gen and newer you can go faster if you like ddr4 4000 and even 4400 are possible on the new cpus and tend to work without a lot of fuss if you have a premium motherboard intel also doesn't have the infinity fabric sync issues to deal with so there isn't a penalty to going that fast on team blue that brings us as to why we're testing 3200 versus 3600 and not including 4 thousand on zen 2 the infinity fabric runs at a one-to-one ratio up to 3733 on the ram speed but decouples beyond that while 4000 is indeed fast it introduces latency to the cpu that makes it not worth it 4000 also costs more today so we're not testing that here we also aren't testing 14 different cpus here to show every minor 1 difference that might be possible now if this video does well i will follow up with some additional configurations if there is interest let me know what you would like to see tested down in the video description below however i think today's test shows the bulk of the answer without overkilling it now our test bench today is my amd x570 setup featuring an msi x570 ace motherboard with the newest bios that enabled faster ram support on zen 3. those of you who saw my launch video may remember that the 3600 refused to work at launch that was a motherboard issue it was addressed a few weeks later with a bios update to keep things as absolutely fair as possible everything is exactly the same between all these runs except the ram speed the clock speed i'm even using the exact same ram kit just turned down on the 3200 run now this is g skill trident z rgb neoram it's amd designed and certified ram and it is in a four by eight gigabyte configuration so all four slots and all four ranks are filled it is sold as ddr4 3600 cl 16 not 18 ram so it's the good stuff and the specific timings are 16 19 19 39 and xmp and the motherboard's bios was enabled for the 3200 test the timings are exactly the same xmp was turned on everything was left the same but the ram speed was manually turned down to 3 200. now you can do this as well if you have a problem running high speed ram kit on your particular system turn on xmp leave it as normal but set the actual clock speed of the ram one or two notches down and you might find your blue screens and stability issues go away if you're just on the edge of compatibility and support the exact same cpu gpu and everything else was used for both sets of testing we are using a ryzen 5 5600 x 6 score 12 thread zen 3 chip with precision boost overdrive enabled but otherwise everything else is left in stock configuration besides pbo and xmp our video card is an rtx 3084 the win3 from evga the cooler is a knock to a nhu 12a very nice cooler the power supply is a wonderful sea sonic 750 watt 80 plus titanium prime our boot drive is the samsung 970 evo 512 gig drive and our games are all installed on an adata sx 8100 2 terabyte premium nvme drive the footage you're seeing in this video was captured on an avramedia live gamer 4k capture card installed on another pc so the test bench doesn't even know it was being recorded msi afterburner was used for the on-screen footage and the benchmark results that you're about to see finally affiliate links to amazon newegg and ebay for everything talked about in this video will be down in the video description below those make checking prices easy and using them while shopping for anything is a great way to support us at no extra cost to you even if you just need a jar of peanut butter our first game in this test is assassin's creed odyssey normally this is where i make a joke about ubisoft's wonderful game optimization but instead i want to comment on how uneven some games one percent low numbers can be and this one is a good example now i suspect i got lucky with an unusually good result on the 3200 test because i ran this many times and i couldn't break 70 on the one percent low on 3600 as you can see here there is no real difference in average performance apart from the odd one percent low number given more time i'd go back and retest the 3200 but sadly time is always working against me here anno 1800 is next and finally we have a victory for the 3600 clock speed test if you can't hear the sarcasm in my voice well it's there this really is a tie there is no difference here that isn't accounted for in the margin of error in testing here you can see the results 260 frames per second versus 262 on the average and 154 versus 156 on the one percent low well this is exciting borderlands 3 is up next and this will be the last game i show you the footage from because it's boring as can be so we'll put all the footage into a video over on rogue tech gaming if you want to see all of it and just run through the charts now 173 versus 174 and 137 versus 136 i almost can't bring myself to read out the numbers it's so exciting f1 2020 is likewise the same allow me to remind you we're not even using slower ddr4 3600 cl-18 ram both kits are cl 16 so the 3600 should be faster right far cry new dawn shows us the truth it really isn't fast i can show you a memory transfer benchmark that will show the difference but the truth is for most situations there really isn't one ghost recon breakpoint has one of the biggest differences here 198 to 206 which if you're just looking at numbers and looking at a chart sounds like a lot but let's be honest this is a four percent difference that you will never ever see outside of a chart horizon zero dawn is more of the same as is the division two a whole bunch of nothing exciting watchdogs legion does have a three percent difference i guess that's technically something here is the nine game average chart the most underwhelming chart ever the blunt honest truth is that ram speed is generally not a bottleneck on modern systems and as such increasing the speed of the ram does nothing for most not all but most applications here is the nine game percentage difference chart this might actually beat the last one as the most underwhelming chart ever now what about non-gaming you ask that's where ram is really needed right well here is the 7-zip test showing once again that ram speed isn't the issue here cpu-z bench and cinebench r20 both also show the same thing the cpu's on-chip cache combined with its prefetch ability renders ram speed mostly moot finally that brings us to a real test blender this is a long detailed complex test that does not let me repeat does not fit into the cpu's on-chip cache but it doesn't matter the limit is compute power not ram speed okay give me a show of hands how many of you are surprised right now i'd be shocked if none of you had your hands up with all the hype over ram speeds and other tests showing at least some difference at least some of you must have figured there would be more improvement than that a few thoughts on why this is so underwhelming first ram speed is only going to improve performance when ram speed is the limit to performance sounds obvious but i think that point gets missed in many conversations about performance if cpu compute or gpu speed or network bandwidth or something else is the limit to performance to whatever it is that you're doing then more ram speed won't make any difference think of it this way if the current program you're running is using 50 percent of your current available ram speed will doubling your ram speed help all that would really happen is that your ram use would drop to 25 percent but no change that you'll notice would actually happen i think it's worth noting that big expensive servers tend to use much slower ram than pc desktops do and they function just fine if the use of ddr4 2400 or 2666 was an issue for them if upgrading to 3200 or even 3600 would make big 56 or 112 core xeon servers run faster don't you think server companies would have pushed for that long ago they could sell big bucks upgrades with all the data they move around the fact that they don't bother is telling if you want to make your computer run faster more ram will help long before faster ram will one more thing to consider this is a limited test of one cpu on one motherboard results will vary on older cpus and sometimes different motherboards depending if the company overrides the xmp settings and does little behind the scenes tweaking as has happened on the amd am4 platform in the past i have done a previous video testing ram on the ryzen 7 1700 and i found that going from 2666 to 3 000 did make a difference not overly dramatic but it did make a difference but going from 3000 to 3200 did not dropping down to 2133 or 2400 hertz a bit more but honestly not as much as some may think regarding intel cpus they really really don't care all that much 3200 3600 4000 whatever unless you're doing custom ram timings and tuning high-end overclocking and competitive benchmarking this is really all a bunch of fuss over nothing to make a very long story short by ddr4 3200 cl 16 or ddr4 3600 cl 18. the 16 is not worth the premium both are almost the same price these days and ignore the fancy expensive stuff that makes no real world difference urcd keys is the best source for genuine windows 10 and office professional product keys that work the first time every time for the month of december get a bonus of 28 off the normal prices using our discount code td20 in the link in the video description below less than 15 gets you a windows 10 professional oem key that is a real product key and activates directly with microsoft use it forever as it links to your microsoft account and it works through reinstalls get a full copy of office 2019 professional plus for under fifty dollars that redeems at setup.office.com using your microsoft account also works forever through windows reinstalls we have been using urcd keys for almost three years and recommend you do as well thank you all so much for watching this entire very exciting video two gold stars to all of you for making it to this point like this video if you like it share it with your friends if you love it remember to subscribe to the channel with a big huge red button directly below questions comments thoughts feedback suggestions you know where the comment section is let me know what else in terms of ram speeds you would like to see tested i suppose i could do cl 16 versus 18 on 3600 i can just tell you now it'll be the most boring test ever it'll look like this one but there's enough of you want to see it done i guess i could do a maybe a short five minute video just to make the point i guess we could test 4 000 speed ram on the news n3 chips and see how that works of all the things however you can spend money on i'm not exaggerating in this video it really is the last place you should be spending money buy more ram not faster ram links in the video description below as i mentioned before great way to support the channel at no extra cost to you if you'd like to support us directly smash that join button become a member of the deal nation join the tech deals discord get access to exclusive videos early access videos and the private chat channels over there thank you all so much for watching i will see all of you next time [Music] you

Channel: Tech Deals

Views: 736,383

Keywords: DDR4-3200 vs DDR4-3600, ddr4 3200, ddr43200vsddr43600, Ram 3600 vs 3200, RamSpeedMatter, ddr4 3200 vs 3600, 3200 vs 3600 ddr4, Ram speed matter, ddr4 3200 vs 3600 gaming, ram 3200 vs 3600, 3200 vs 3600, 5600x 3200 vs 3600, doesramspeedmatter, Ram speed test, does ram speed matter, ddr4 3200 vs 3600 intel, 5600x ram speed, ram speed comparison, 5600x vs 3600, ram speed, does RAM Speed Matter? — DDR4-3200 vs DDR4-3600 — Game & Productivity Testing on R5 5600X, tech deals

Id: uy6dgB24frg

Length: 16min 28sec (988 seconds)

Published: Mon Dec 07 2020