Memory War: Single VS Dual Channel for GAMING | Does it Matter?

Share
Embed
  • Published on Apr 26, 2019
  • Tested on a Legion Gaming PC: www.lenovo.com/us/en/legion/
    Thanks to Lenovo Legion and Intel.
    RAM Used- Patriot Viper 3600MHz: amzn.to/2Vc03yp
    Store: epicpants.com, Music: bit.ly/Trk2ik
    Fenek on Amazon: www.amazon.com/s/keywords=fenek
    Fanatical Game Deals: www.fanatical.com?ref=teksyndicate
    Patreon: www.patreon.com/teksyndicate
    Website: teksyndicate.com/
    Forum: forum.teksyndicate.com/
    -----------------------------
    Discord: discord.gg/vsCxNSD
    Facebook: TekSyndicate/
    Instagram: teksyndicate
    Reddit: www.reddit.com/r/teksyndicate/
    Twitch: www.twitch.tv/MidnightDojo/
    Twitter: teksyndicate/
    Want to know what we use to make these videos? Check out our Kit: kit.com/TekSyndicate
    For marketing (sponsorship opportunities) inquiries email info@teksyndicate.com
    Ask us LITERALLY ANYTHING: Inbox@teksyndicate.com
    Follow the team:
    Logan: Logan_RTW
    Pistol: pistolsteph
    Justin: DatAwesomeRobot
  • Science & TechnologyScience & Technology

Comments • 92

  • Stephen Weston
    Stephen Weston 4 months ago

    How do you guys have over 600k subs yet only get a few thousand views per video?

  • Michael Stewart
    Michael Stewart 7 months ago

    cant locate page on site 404

  • Andrea Andrea
    Andrea Andrea 7 months ago +1

    lol what happened to your hair?

  • Gigi Berlogea
    Gigi Berlogea 7 months ago

    Haha so many close numbers... why would you put this in a video?

  • Jo3yization
    Jo3yization 7 months ago

    You can see a youtuber called 'wolfgang' single vs dual tests across a range of games proving it makes a huge difference, CPU intensive titles take a much larger fps hit than GPU intensive games/benchmarks. Test Battlefield 1/CSGO/pubg single vs dual 2666 or use any of the games in other single vs dual comparisons that can be found on YT: thexvid.com/video/-k5wA7EFwpo/video.html

  • Neurot1c Studios
    Neurot1c Studios 7 months ago +1

    Well within the margin of error between the 16GB configurations... Proof of that when the single channel 16GB stick pulled ahead by 1 frame. Not too surprised as this has been the case for gaming for several generations of HW. However, you should test scenarios where you're using an IGP like Vega or where there's not enough VRAM and the program has to fall back on system RAM.

  • aks
    aks 7 months ago +1

    love the oscar wilde hair style!

  • Dmitry Granicin
    Dmitry Granicin 7 months ago

    I tried to run my home server with only 1 stick of RAM, it exploded and nothing worked.

  • F Hilario
    F Hilario 7 months ago

    And the loading time on these games? Also not big difference ? 🤔 It could save a cup of coffe or two.

  • Devon Crain
    Devon Crain 7 months ago

    Frame time & frame rate over time missing

  • Ross Mclaughlin
    Ross Mclaughlin 7 months ago

    Awesome video would have loved to see 8gb dual channel though but honestly your content is honest n I also liked FX not bad for budget video this is why I fallow yous yous post truth with out regards to popular opinion keep up good work 👍

  • Juan Gonzalez
    Juan Gonzalez 7 months ago

    All mighty Kung fu Master, please teach me your Kung fu...

  • f4z0
    f4z0 7 months ago +2

    This look reminds me of Travolta in Swordfish.
    Nice video btw

    • Mshojat
      Mshojat 6 months ago

      I don't like it. I don't know what it is exactly, but it's very odd. I had a similar sort of hairstyle when I was younger (middle-parted hair of ear length- either top or bottom of ear), but I didn't have huge hair recession then like Logan has here. I suppose that's what makes it look strange; too much forehead for that kind of hair style.

  • HypnoticSuggestion
    HypnoticSuggestion 7 months ago

    I think this is largely to do with just how much faster memory has gotten. Years ago it definitely mattered, but now days memory is so damn fast that a single channel is enough, at least, for gaming. I have quad channel, and I tested with just two channels and made zero difference in games. In single channel it actually slowed down, but that's because it's slow 2133mhz ddr4.

  • ditto
    ditto 7 months ago

    Get the nail! 🔨

  • ISO Inebriation
    ISO Inebriation 7 months ago

    I really wish more reviewers would at least attempt to overclock ram instead of recommending purchasing more expensive "faster" memory kits. I've had a ton of sets rated at 2133, and I've never had an issue getting them to at least 3000mhz with decent timings. Literally set dram voltage to 1.3 and frequency to 3000mhz, that's passed in every system. I've even bought a corsair vengeance kit rated at 4000mhz just to compare to, the systems are more likely limited by the motherboard or IMC.

  • Ear Rapist
    Ear Rapist 7 months ago +1

    Whats up with the apex legends fps?

  • chrcoluk
    chrcoluk 7 months ago +2

    I think this is because latency matters more than bandwidth.

    • Secret Sauce is overrated
      Secret Sauce is overrated 7 months ago

      Alot of games are developed for consoles, and these games are specifically developed to utilise low amounts of bandwidth for the cpu. Both the leaked PS4 and Xbox One devkit docs recommend devs minimize the cpu's access to the main memory pool because unified memory has a big drawback that cpu bandwidth usage disproportionately robs the gpu of bandwidth needed to render graphics. There is an interesting video on youtube where a pair Naughty Dog's programmers also gave a university console programming presentation for students a couple years ago that you want to always/constantly focus on keeping cpu code small enough that it can fit inside the Jaguar's 2x2MB L2 cache and you minimize cpu's use of main memory bandwidth. There are also government research studies at Oak Ridge National Laboratories that show unifying memory regardless if its IBM, Intel or AMD archictectures causes memory contention and will disproportionately rob each other of memory bandwidth.
      Some games designed without the limitations of current gen consoles in mind may be alot more cpu bandwidth intensive.

  • Grigor Antonov
    Grigor Antonov 7 months ago +1

    Pointless video. Didn't show 1% and 0.1% lows. Ignored Ryzen, which benefits massively from fast RAM in dual-channel configuration.

  • Stefan Krunic
    Stefan Krunic 7 months ago +6

    whats with the hair.. you look like a lunatic

  • Jasper Bell
    Jasper Bell 7 months ago

    Please do another test on ryzen

  • Nicholas Erwin
    Nicholas Erwin 7 months ago

    Also it you gotta count price - performance ratio. Sure you can buy crazy expensive ram with fastest clock speed and lower timings but how much performance are you actually getting that is worth the money that could be spent on another component such as a video card that would give more FPS? Or a faster M.2? Or maybe even a faster CPU?

  • Hugo
    Hugo 7 months ago +1

    Why can you not just show a graph instead of talking about every result on it's own? Oh wait, 10:08 video length, i see now

    • Stefan Krunic
      Stefan Krunic 7 months ago

      he's cocksucker, wants to make more money

  • mossimossi2
    mossimossi2 7 months ago +3

    Can we see a Ryzen (preferrably 2000 series) with 3000mhz DDR4 onwards comparison? Also it might be good if you guys are showing the 2 systems running side by side and rivatuner stats in real time. There are so many other reviewers out there that do that and they look way more accomplished than what you're doing.. which is a shame really as I really liked your channel in the past if you catch my drift..

  • Meteor Media
    Meteor Media 7 months ago +2

    8700k paired to a GTX 1060. What a joke.

  • jawnTem
    jawnTem 7 months ago

    I'd like to see a comparison using 16 vs 32. On my system I think it actually slowed stuff down; but, I'm not that tech-oriented nor do I have the equipment to run such comparison tests.

    • Secret Sauce is overrated
      Secret Sauce is overrated 7 months ago

      Alot of games are developed for consoles, and these games are specifically developed to utilise low amounts of bandwidth for the cpu. Both the leaked PS4 and Xbox One devkit docs recommend devs minimize the cpu's access to the main memory pool because unified memory has a big drawback that cpu bandwidth usage disproportionately robs the gpu of bandwidth needed to render graphics. There is an interesting video on youtube where a pair Naughty Dog's programmers also gave a university console programming presentation for students a couple years ago that you want to always/constantly focus on keeping cpu code small enough that it can fit inside the Jaguar's 2x2MB L2 cache and you minimize cpu's use of main memory bandwidth. There are also government research studies at Oak Ridge National Laboratories that show unifying memory regardless if its IBM, Intel or AMD archictectures causes memory contention and will disproportionately rob each other of memory bandwidth.
      Some games designed without the limitations of current gen consoles in mind may be alot more cpu bandwidth intensive.

  • BassRacerx
    BassRacerx 7 months ago

    again for ryzen please!

  • HOkayson
    HOkayson 7 months ago

    Checkout Overwatch single vs dual, that was a hoooge difference when I upgraded my laptop!

  • StaySic4Ever
    StaySic4Ever 7 months ago

    In certain games it will certainly show difference, more on Ryzen too. So really today for a new rig the best pick is really a 16GB 3200MHz CL14 kit.

  • DualPerformance
    DualPerformance 7 months ago

    Thank me later thexvid.com/video/-k5wA7EFwpo/video.html

  • arse
    arse 7 months ago

    i'd rather have a single stick of faster ram than slower dual, certain games have huge fps boosts like csgo get massive gains for fast ram

  • Chris B
    Chris B 7 months ago

    Hm, the results of Tek Syndicate are somewhat counter intuitiv.
    Other videos show significant difference in performance (more than +/- 10%) between single- and dual-channel. Here example video (intel box with 1080Ti) thexvid.com/video/9w1U-kForAU/video.html

    There a number of differences between Legion box here on Tek Syndicate and intel 1080Ti box form example box. Also the games are different...

    but after comparing those two videos alone my conclusion:
    1) Legion T730 with nvidia 1060 does not need dual-channel for the games shown here. (A impact with a nvidia 1080 or better would be nice)
    2) After checking other videos, it is clear that testing on singel system is not enough to draw generalized conclusion, its complex discussion. -> More benchmarks are needed.
    3) Should you buy dual-channel? It depends... you have to do more research to get results that are useful :-\

    Side note: Digital Foundry and Linus Tech Tips have also made videos about this topic, you should watch them to get clearer understanding of the whole picture. The topic single- vs dual-channel is complex.

  • conyo985
    conyo985 7 months ago +1

    LOL! 1% lows isn't the same as min. FPS.

  • RepsUp100
    RepsUp100 7 months ago +3

    Dual channel master race

  • giorx5
    giorx5 7 months ago

    Average FPS isn't affected as much as minimum FPS

  • xXiloveXx
    xXiloveXx 7 months ago

    Okay but the hair looks like a wig

  • Zonther
    Zonther 7 months ago +13

    Decided to try this with my R7 2700X with Dual 16GB Cas 14 3200 MHz and RX 580 8GB 1430Mhz core 8200Mhz memory. Only did Final Fantasy XIV: Stormblood Benchmark Max settings, Full screen, 1080p.
    Dual
    Score: 11011
    Average Frame Rate: 74.858
    Total Loading time: 46.048 seconds

    Single
    Score: 10917
    Average Frame Rate: 74.425
    Total Loading Time: 47.604 seconds

    Hope this helps

  • Moka hless
    Moka hless 7 months ago +4

    The problem with this video isn't the conclusion - your conclusion is something I agree with in many circumstances. The problems are as follows:
    -Basically my biggest problem is your overarching general conclusions when what you did was only test a small subset of circumstances. Balancing systems can be complex and in some cases, dual channel memory can make a difference these days.
    -Methodology.
    -Not accounting for system components and testing bottlenecking
    -Testing the RAM at a single speed. Testing at multiple speeds down to 2133 and up to 3200 would have been one way to account for bottlenecking
    -Some of your own tests showed rather large differences even in the circumstances you gave the system but you dismissed them because they were within the same sort of range and the rest of the games at your settings didn't show any difference. What if that 41% dip had been the difference between 42FPS and 60FPS minimums? May well have been at different settings but the tests weren't very thorough.


    Below are my accompanying initial reactions:

    1:31 Oh yeah. Great idea there. Let's get right into the benchmarks. Forget your testing methodology and system specifications. *checks description* - nope, not there either.

    2:41 "nearly identical"
    shows a 41% difference in minimum frames.

    Why are you including decimal places? Are you trying to make us think there's no variance between tests?

    7:00 seriously? That's like saying you expect a 6-core CPU to be faster than a 4-core in all games. Why did I subscribe to this channel again? It certainly wasn't for the testing.

    7:06 oh yeah NOW we get the specs. Wait. A 1060? There's more than just additional testing with AMD CPUs you need to do.

    7:25 this runs contrary to the tests you just ran. Dual channel is a way to indirectly have faster RAM. Why would faster RAM be better than dual channel? You didn't even test this before stating it like fact.

  • sidd98110
    sidd98110 7 months ago +4

    I recently Switched to dual channel from Single 8GB to 2x8Gb and the difference in the games i play was a lot. Try Running BFV on 8GB single channel. its literally unplayable. You need to benchmark games which are CPU intensive rather than Gpu Intensive. I have an intel 6600k.

  • Emerson Silva
    Emerson Silva 7 months ago +4

    Dude c’mon, these test where pretty invalid. You didn’t mention what hardware you where using. As people have mentioned, it does look like it’s being gpu bottleneck. Aswell as not showing stuff like 1% or .1% lows where stuttering would happen by not having the proper memory bandwidth. That farquad hair of yours has gotten to you...

    • Em's Tek Tube
      Em's Tek Tube 7 months ago

      Emerson Silva And no mentions to how it greatly affects ryzen systems!

  • Thuan Tran
    Thuan Tran 7 months ago

    gpu seems to be the bottleneck here. interested in seeing the same test with a 2080/2080ti

  • revilo304
    revilo304 7 months ago

    but single channel looks dumb

  • ROFLCOPTER
    ROFLCOPTER 7 months ago

    DUAL CHANNAL!1!1!1!!1

  • Léo Bitencourt
    Léo Bitencourt 7 months ago +8

    I don't mean to give off the impression I'm a hater or something, but it's painfully obvious there was a GPU bottleneck. If you had a budget CPU or a stronger GPU I bet it would make a difference. Maybe the difference wouldn't be big, but it would be there. If the CPU is just fooling around, it doesn't really matter if it's a bit weaker or a bit stronger. Maybe what you tested can be called a "more realistic scenario", but not many people would pair an i7 8700k with a 1060.

    • Léo Bitencourt
      Léo Bitencourt 6 months ago

      @XInfinity2013 It doesn't depend as much as ryzen, but it still makes a difference. And I believe the difference between single vs dual channel ram is bigger than the difference between 2400mhz vs 3200mhz ram.

  • the one you dont see
    the one you dont see 7 months ago +2

    its not the tests showing about 1% its the meta of dual-channel for enthusiast. the tests show 1% but full analysis time frame consistency and using all the monitor software for that dont forget inputs too. the synergy of the whole system of me at my desk playing doom keyboard under my left hand mouse in my right with more of a mid-range pc thats not hitting 100fps on medium 1080p but around 50fps test still shows 1% yet the smoothness monitors are showing closer to 2% not much bigger but let me tell you in dual-channel i got that doom slayer grin so i can tell if its dual-channel.

  • zerphase
    zerphase 7 months ago +2

    The answer is quad channel?

    • Major Epeen
      Major Epeen 7 months ago

      The task being performed did not come anywhere close to saturating the memory bandwith. All dual channel memory does is double the bandwidth for the memory. (To put it simply you have increased the size of the pipe allowing more data to flow through (although it is technically divided into 2 pipes)) All quad channel will do is double this bandwidth again (albeit this will be split into 4 pipes). The test results were all within the margin of error which shows that the task being performed did not come anywhere close to filling this "pipe". If the test did come close to saturating the memory then enabling dual channel mode would have yeilded better results. You need to run memory intensive tasks in order to see the benefit of enabling dual channel mode. The type of test being performed ( Graphical) will only be affected by memory clock speed not bandwidth. Increasing CPU clockspeed and GPU clockspeed and memory speed were the only variables that would have yeilded better results.

    • pacsmile
      pacsmile 7 months ago

      Farquaad channel.

  • Grzegorz P.
    Grzegorz P. 7 months ago +6

    Your benchmarks are too easy on the CPU and very graphics-intensive. Your GPU is clearly the bottleneck here and system RAM speed doesn't affect it much, since GPUs have their own (GDDR/HBM) RAM memory. Try something CPU and RAM intensive like VeraCrypt, Handbrake etc. + try to find some more CPU-intensive games with lots of physics and less fancy eye-candy.

  • cee128d
    cee128d 7 months ago +4

    Dual Channel affects INTEGRATED graphics where they share the system ram. Once you put an actual graphics card into play it becomes margin of error.
    To sum it up, you have to perform tests accordingly to get the proper results, unless you aim is to back up a predetermined outcome.

  • Deon Spates
    Deon Spates 7 months ago +1

    Dual Channel RAM is more important for AMD Ryzen! Trust me I had experience in this case.

  • John Nada
    John Nada 7 months ago

    Who knew that build from the verge was so good?

  • E Hiebert
    E Hiebert 7 months ago +1

    Thanks for the video.
    Would like to see this type of video using AMD.

  • Introvert Plays
    Introvert Plays 7 months ago +4

    Pity that Tek Syndicate is now a dedicated Intel Ambassador. Hoped for some statements on Ryzen 3xxx as soon as it is released. I guess that won't happen then.

  • D Jaquith
    D Jaquith 7 months ago +1

    Pretty misleading video, you must have scoured the web to find a few games that it doesn't matter have Dual Channel.
    Here's a couple videos live that show otherwise Dual Channel is superior. In addition, micro stuttering is much higher with Single Channel.
    thexvid.com/video/kCg7eVN6N9w/video.html
    thexvid.com/video/V5YGIs0OFbs/video.html

  • DJ SALTYNUTZ
    DJ SALTYNUTZ 7 months ago

    sauce on track id

  • TJ Rooger
    TJ Rooger 7 months ago +1

    Heresy. Thou shalt fetch the cross of most holy crucification immediately, and place thyself upon it, for I am entirety too far away to do it myself.

  • Soap A
    Soap A 7 months ago +23

    Showing averages and min doesn't tell the whole story. We need to see a frame time chart, and frame rate over time.

    • Jo3yization
      Jo3yization 7 months ago

      @Soap A Yep,, they didnt use CPU intensive titles which have much higher overall memory bandwidth requirements than GPU based benchmarks.
      Shadow of war is not CPU heavy and scales well all the way down to 2cores4threads.
      Same with DOOM, not a CPU intensive title and scales well across most CPUs.
      Then they used a bunch of GPU intensive benchmarks, again, not CPU heavy.

    • Soap A
      Soap A 7 months ago

      This video also illustrate the MASSIVE difference, thexvid.com/video/o5nDQ_ufpMY/video.html so something is not right with the testing methodology of this tek syndicate video.

  • Peter Jansen
    Peter Jansen 7 months ago +35

    Big caveat: this only applies to Intel!!
    For Ryzen you will lose a lot of performance with one stick of RAM.

    • flabbergast
      flabbergast 7 months ago

      On a ryzen 5 1600 my ex got slightly better fps than myself with single channel.
      I think it's mainly placebo because it used to make a huge difference.
      A faster single channel is better than 2 slower in dual channel.
      As I said, with the same cpu and the same graphics card my ex got a higher score in Passmark despite doing the same test several times and me having dual channel memory.

    • HypnoticSuggestion
      HypnoticSuggestion 7 months ago

      Absolutely. It can make a gigantic difference because of the chiplet design and infinity fabric. On mainstream Intel it makes almost no difference.

    • Forever Computing
      Forever Computing 7 months ago

      Stick?
      My computer has zero sticks of RAM....

    • AlphaZeroOmega
      AlphaZeroOmega 7 months ago

      @One4speed Actually he didn't. Even mentions at the end he's willing to try it out on an AMD system.

    • Bumble
      Bumble 7 months ago

      @Ryan Kober Most new gaming mid range to high end builds are ryzen and nvdia, I think this is a sponsored video

  • dazzlerweb
    dazzlerweb 7 months ago +2

    I did some testing on this a couple of years ago after another channel did a similar investigation and came to the same conclusion, often results are within margin of error.

  • Bumble
    Bumble 7 months ago +2

    Isn't ryzen much more sensitive to dual channel for gaming? Ideally you would have tested both intel and amd platforms.

    • Piotr Kozaczka
      Piotr Kozaczka 7 months ago

      @Bumble yeah. Would be nice if be tested ryzen as well

    • Bumble
      Bumble 7 months ago

      @Piotr Kozaczka But still - Ryzen cpu's (which are selling more than intel last 3-6 months) should be tested. Who is going high end gpu with an intel cpu in 2019 for 4k gaming?

    • Piotr Kozaczka
      Piotr Kozaczka 7 months ago

      @Bumble if you don't have a GPU that pushes cpu to it's limits single channnel is fine ;]

    • Bumble
      Bumble 7 months ago

      @Piotr Kozaczka Thanks for agreeing with me, I was being polite in pointing out the problems with the testing.

    • Piotr Kozaczka
      Piotr Kozaczka 7 months ago

      ryzen is generaly memory bound platform

  • adamsrealm
    adamsrealm 7 months ago +1

    Dual channel is a no-brainer. Regardless of the tests, dual channel is ALWAYS better.

    • Jeremy
      Jeremy 7 months ago

      @T. because you're not getting the full story.

    • T.
      T. 7 months ago +1

      It's always better, except when it's not, like in these gaming tests