RAM performance by frame latency [PCgameshardware.de]

Discussion in 'General Hardware' started by Carfax, May 12, 2016.

  1. Carfax

    Carfax Ancient Guru

    Messages:
    3,948
    Likes Received:
    1,442
    GPU:
    Zotac 4090 Extreme
    Here's a new article from PCgameshardware.de which verifies what I've always known..

    [​IMG]
    [​IMG]
    [​IMG]

    RAM capacity, and RAM frequency matter despite what some naysayers might believe. In the test, PCgameshardware.de show some very significant differences in frame time between 4GB/8GB/16GB configurations in BF4..

    If they had used the standard benchmark test method for RAM employed by almost every other hardware site which only uses the average frame rate, or if they're generous they include the minimum frame rate as well, it would likely have resulted in NO DIFFERENCES which would have reinforced the lies that have been circulating around the internet regarding RAM capacity/frequency and gaming.

    This review by Techspot is one such example, and because the testing methodology is flawed, there is no appreciable difference between the various RAM capacities that were tested.

    [​IMG]

    But by using frame time analysis like what PCgameshardware.de used, we get the full picture.

    As for frequency, it's even more salient, with large differences in frame rate (up to 23%) between the fastest and slowest kit tested.

    [​IMG]

    The impact of RAM frequency will always vary depending on the configuration and the setup. Dual core and pure quad core CPUs have the most to gain, as they have less cache and or lack HT, which helps to hide memory latency.

    But I personally can attest that even on a high end hexcore rig and very high graphical settings, the impact of RAM frequency is noticeable if you know what to look for.

    The Witcher 3 is such an example. The game has limited multithreading, and so shader compiling can cause mild stutters unless you have shader cache turned on in the GPU control panel.

    But I've noticed that at DDR4 3200, the game stutters far less than at say DDR4 2666 when entering (Geralt must be running and shader cache must be turned off) Novigrad from Portside gate fast travel sign after a fresh load. What I'm guessing is that the higher frequency RAM allows a quicker copy of those shaders into RAM from storage, and also allows the CPU to do the compiling much faster as well.
     
    Last edited: May 12, 2016
  2. fantaskarsef

    fantaskarsef Ancient Guru

    Messages:
    15,636
    Likes Received:
    9,512
    GPU:
    4090@H2O
    Hmmm, interesting. Thanks for the find.
     
  3. Valken

    Valken Ancient Guru

    Messages:
    2,922
    Likes Received:
    903
    GPU:
    Forsa 1060 3GB Temp GPU
    This affects ARMA 3 a lot too... Would have been cool to see CPU cache such as Broadwell affecting games with fast ram.
     
  4. thatguy91

    thatguy91 Guest

    Like I said the other day, I did a little research and it seems DDR4-3200 is the sweet spot for Skylake and probably Broadwell and the upcoming Kaby Lake as well. I doubt that they will tweak the memory controller to any great extent with what will likely be essentially a Skylake refresh. I should point out I don't have Skylake, this is just going by online reports, and experience from a Skylake user.
     

  5. Carfax

    Carfax Ancient Guru

    Messages:
    3,948
    Likes Received:
    1,442
    GPU:
    Zotac 4090 Extreme
    Most every game is affected to some extent, depending on the hardware configuration. And then some games are affected more than others by bandwidth and latency because of how the engine is programmed to handle assets..

    Fallout 4 is another great example. Anyway, I expect the trend of games being able to take advantage of high frequency RAM to increase as they become more complex, interactive, larger and more seamless.

    Also since Haswell, CPUs have much faster caches and more aggressive prefetching which plays very nicely with fast RAM..
     
  6. Carfax

    Carfax Ancient Guru

    Messages:
    3,948
    Likes Received:
    1,442
    GPU:
    Zotac 4090 Extreme
    And? The point of the benchmark was to show how RAM frequency can affect performance.

    Increasing the resolution shifts the burden from the CPU to the GPU, so of course RAM frequency's impact is going to lessen.

    It's much the same for CPU frequency as well, but does that mean overclocking the CPU is useless? Of course not.
     
  7. Carfax

    Carfax Ancient Guru

    Messages:
    3,948
    Likes Received:
    1,442
    GPU:
    Zotac 4090 Extreme
    Yeah, but the whole point of this post was to show that you can't simply rely on frame rates to judge the impact of both RAM capacity and frequency.

    You really have to look at frametime, but the vast majority of reviewers don't go to that length.

    The frametime graphs with BF4 were done at 1080p, a resolution that most people play at, plus ultra detail and MSAA 4x.

    Yet differences in frametime was still apparent between 4GB, 8GB and 16GB configurations.
     
  8. PrMinisterGR

    PrMinisterGR Ancient Guru

    Messages:
    8,125
    Likes Received:
    969
    GPU:
    Inno3D RTX 3090
    And to think that a lot of people in here thought I was crazy for telling them it matters :p
     
  9. EspHack

    EspHack Ancient Guru

    Messages:
    2,794
    Likes Received:
    188
    GPU:
    ATI/HD5770/1GB
    I've been enlighten! lol seriously, I remember a friend of mine telling me to show him how my ddr3 was faster than his ddr1, and I couldn't, and from the common benchies out there I was certain that the only thing that mattered about ram was capacity and how much you needed for your multitasking, now with this FCAT thing I see it does improve something, and it is something that drives everyone insane and a really hard thing to diagnose

    great find OP
     
  10. AsiJu

    AsiJu Ancient Guru

    Messages:
    8,774
    Likes Received:
    3,354
    GPU:
    KFA2 4070Ti EXG.v2
    Very nice find. I myself thought for a long time that RAM speed is more or less insignificant vs. RAM amount but this isn't the case anymore.
     

  11. Singleton99

    Singleton99 Maha Guru

    Messages:
    1,071
    Likes Received:
    125
    GPU:
    Gigabyte 3080 12gb

    It seems you're right :):):):) , faster ram does indeed produce more fps, and by the looks of those charts it more than most would think when you read articles around the web on this subject .

    I will now be running my ram O/C ed @ 2900mhz , when i get more time i will run some benchmarks to , its just when i did run it 2900 instead of 2666 i don't think i saw any increase in fps with the witcher 3 i will have to check again maybe i did something wrong at that time or maybe it was my min fps were better and i didn't notice :nerd:...


    Nice article by the way :)
     
  12. Carfax

    Carfax Ancient Guru

    Messages:
    3,948
    Likes Received:
    1,442
    GPU:
    Zotac 4090 Extreme
    Remember the Witcher 3 benchmark was done at 720p to shift the burden on the CPU, and also likely done in Novigrad, which is the most CPU intensive area in the game.

    So if you want to test it, do a run through Novigrad. But you really have to pay attention because whilst RAM frequency can definitely impact frame rate, it's main benefit is to help with frametime, so you have to rely on how the game FEELS to measure unless you have frametime measuring software.

    The rest of the game is GPU intensive, so it won't have much of an impact there until Blood and Wine come out. Blood and Wine by the looks of it, should be more CPU intensive, especially in Beauclair the capital city of Toussaint.
     
    Last edited: May 17, 2016
  13. PieEyedPiper

    PieEyedPiper Master Guru

    Messages:
    628
    Likes Received:
    10
    GPU:
    RX 580
    Of course it matters, it's pretty much a given that faster memory will net more fps, the question is whether it reaches statistical significance in more than just a handful of titles, and at what settings.

    Let us all not forget here that it is a very valid point to make; 720p is no gaming resolution. Please let us keep perspective and not just make broad statements like "..the point is it shows performance.." when someone provides a critical thinking path.

    I'm glad BF4 shows such a response, there was a another game too about a year ago that had graphs very similar; showing rather impressive gains with memory frequency increases. Until it's shown that at high(er) resolutions the gap is significant between "faster" and "slower" frequencies no one should feel urged to go replace their current gear.

    All in all, good stuff. :)
     
  14. Carfax

    Carfax Ancient Guru

    Messages:
    3,948
    Likes Received:
    1,442
    GPU:
    Zotac 4090 Extreme
    Here's another frametime graph comparison between 8GB and 16GB for Star Wars battlefront. People were wondering why DICE recommended 16GB for SWBF, this is why. 16GB offers a palpable improvement in frametime compared to 8GB..

    [​IMG]
    [​IMG]
     
  15. morbias

    morbias Don TazeMeBro

    Messages:
    13,444
    Likes Received:
    37
    GPU:
    -
    Would it not be more accurate to say increased RAM bandwidth (rather than speed/timings/quantity) reduces dropped frames?
     

  16. Carfax

    Carfax Ancient Guru

    Messages:
    3,948
    Likes Received:
    1,442
    GPU:
    Zotac 4090 Extreme
    Probably not, because we don't know whether it's the increased bandwidth that's responsible for that effect, or the reduced latency.

    It's likely a combination of both I'd wager, as increasing the speed of RAM results in an increase in bandwidth, and a reduction in latency.
     
  17. morbias

    morbias Don TazeMeBro

    Messages:
    13,444
    Likes Received:
    37
    GPU:
    -
    Ah well, I suppose it depends on how you view bandwidth :)

    Theoretically bandwidth is calculated using transfers per cycle x speed, but as you say it is both as in practice latency has a big effect on the bandwidth because it can result in no tranfers for multiple cycles so the theoretical calculation of bandwidth does not determine what happens in reality.
     
    Last edited: May 19, 2016
  18. dasa2

    dasa2 Guest

    Messages:
    38
    Likes Received:
    0
    GPU:
    XFX 290@1200 MCW60
    there is some games where faster ram will never mater like shadow of mordor where a pentium is bottlnecked by a gtx980 at 1080p
    at least in the section of game anandtech benchmarked here
    Intel Pentium G3258 99.92FPS
    Intel Core i7 6700K 99.6FPS
    http://www.anandtech.com/bench/CPU/1321

    ram benchmarks should have a comparison to cpu speed so that people can see the value in it
    [​IMG]
    http://www.eurogamer.net/articles/d...it-finally-time-to-upgrade-your-core-i5-2500k

    from what i have found latency is the most important thing for games but this may very from one game\system to the next
    keeping in mind that higher frequency reduces latency provided timings are not slackened to much
    [​IMG]
    http://www.hardware.fr/articles/940-5/cpu-ddr4-vs-ddr3-pratique.html

    with dragon age 3 my 290 could only manage ~30-40fps at 2560 but drop the res or upgrade to gtx1080 and the cpu would become the botlneck not able to stay over 60fps without overclocking ram
    [​IMG]
    http://forums.anandtech.com/showthread.php?t=2455994
    % increase
    ht on vs off ~5%
    3.9 vs 4.2ghz ~6%
    1333 vs 2400 ~17%
    1600c11 vs 2400 ~14%
    1600c9 vs 2400 ~11%

    its nice to see there is more info about these days showing faster ram can help
    back in early 2013 when i first posted some tests i ran on arma3 there was a lot of angry people thinking i was spreading false info or working for corsair despite using samsung oem ram
    [​IMG]
    https://forums.bistudio.com/topic/144563-cpu-vs-ram-performance-cpu-threading-benchmarked/
    [​IMG]
    [​IMG]
    https://forums.bistudio.com/topic/1...mance-comparison-1600-2133-up-to-15-fps-gain/
    [​IMG]
    http://forums.atomicmpc.com.au/index.php/topic/55771-cpuram-performance-in-thief/
     
    Last edited: May 23, 2016
  19. thatguy91

    thatguy91 Guest

    I won't quote the post since it's directly the previous one and it's big :). There's two big considerations to take into account with RAM performance, and that is whether the RAM is DDR3 or DDR4 and what platform it is on.

    Many tests compare the performance of RAM speed using Sandy Bridge platform (ix-2xxx CPU's and Z68 board). There seems to have been a memory controller change with Ivy Bridge, as Ivy Bridge can make use of much faster RAM. Ivy Bridge shows a significantly greater change between 1600 and 2133 than Sandy Bridge. The same controller was then carried through to Haswell, or at least it seems it was as the performance stats for Haswell relate almost exactly the same as with Ivy. Skylake has a DDR4 contoller with different performance aspects. This is important since DDR3-2133 performance signficantly better on Ivy than Sandy. Many people bought 1600 and even 1333 because they believed that there was no benefit spending an extra $10 on the faster RAM.

    Price is the biggest decider. You should really get the best value RAM when taking into account speed, latency, and price. For DDR3 platforms Ivy Bridge and later that is DDR3-2133, and for Skylake it is DDR4-3200. It is likely that Kaby lake will also be 3200. The figure for Zen will be different, it make be it can make use of even faster RAM
     
  20. dasa2

    dasa2 Guest

    Messages:
    38
    Likes Received:
    0
    GPU:
    XFX 290@1200 MCW60
    most the tests done by myself above were on sandy where they stop at 2133 and it seems to have continued scaling in performance from 1866 to 2133 fairly well
    i didnt compare read\write\copy speeds though

    i have found read\write speed didnt increase much with ivy moving from 2133-2400 latency continued to drop though and game performance improved

    haswell also hits a bit of a wall around 2400 as far as read\write\copy speed goes where performance seems to drop at 2666\2800
    [​IMG]
    http://www.corsair.com/en/blog/2015/september/ddr3_vs_ddr4_generational
     
    Last edited: May 23, 2016

Share This Page