Does Memory Performance Bottleneck Your Games?

References
By Thomas Soderstrom
published

Not all games are held back by graphics performance. Some seem to be CPU-limited. However, we've even seen benchmark results that appeared to be affected by memory bandwidth. Today we compare quad-channel, dual-channel, and low-latency configurations.

Test Settings And Benchmarks

Swipe to scroll horizontally
Test System Configuration
CPUIntel Core i7-3960X (Sandy Bridge-E): 6C/12T 3.3 GHz, 12 MB Shared L3 Cache, LGA 2011 Overclocked to 4.4 GHz (44 x 100 MHz) at 1.325 V Core
MotherboardAsus P9X79, BIOS 3203 (11/26/2012)
CPU CoolerCoolink Corator DS 120 mm Tower w/Zalman ZM-STG1 Paste
RAMG.Skill F3-17600CL9Q-16GBXLD (16 GB) DDR3-2200 CAS 9-11-9-36 1.65 V
Graphics2 x MSI R7970-2PMD3GD5/OC: 1010 MHz GPU, GDDR5-5500
Hard DriveMushkin Chronos Deluxe DX 240 GB, SATA 6Gb/s SSD
SoundIntegrated HD Audio
NetworkIntegrated Gigabit Networking
PowerSeasonic X760 SS-760KM: ATX12V v2.3, EPS12V, 80 PLUS Gold
Software
OSMicrosoft Windows 8 Professional RTM x64
GraphicsAMD Catalyst 12.10

Intel’s LGA 2011 platform gives us the flexibility to test both dual- and quad-channel memory configurations. Asus’ P9X79 was retained from our previous case round-up, with an upgraded C2-stepping Core i7-3760X CPU.

Due to its 100 MHz base clock, the P9X79 motherboard sets G.Skill’s old DDR3-2200 CAS 9 to DDR3-2133 CAS 9-11-9-28. We used the same memory and timings for our DDR3-1600 C9 test. Our reduced-latency configuration uses this memory at DDR3-1600 CAS 7-9-7-21.

If you own a pair of Radeon HD 7970 graphics cards, there's a good chance you're looking to game at 2560x1600 or better. We started at 1920x1080 and worked our way up to 5760x1080.

Swipe to scroll horizontally
3D Gaming Benchmark Configuration
Aliens vs PredatorUsing AvP Tool v 1.03, SSAO/Tesselation/Shadows On Test Set 1: High Textures, No AA, 4x AF Test Set 2: Very High Textures, 4x AA, 16x AF
Battlefield 3Campaign Mode, "Going Hunting" 90-Second Fraps Test Set 1: Medium Quality Defaults (No AA, 4x AF) Test Set 2: Ultra Quality Defaults (4x AA, 16x AF)
F1 2012Steam version, in-game benchmark Test Set 1: High Quality Preset, No AA Test Set 2: Ultra Quality Preset, 8x AA
The Elder Scrolls V: SkyrimUpdate 1.7, Celedon Aethirborn Level 6, 25-Second Fraps Test Set 1: DX11, High Details No AA, 8x AF, FXAA enabled Test Set 2: DX11, Ultra Details, 8x AA, 16x AF, FXAA enabled
Metro 2033Full Game, Built-In Benchmark, "Frontline" Scene Test Set 1: DX11, High, AAA, 4x AF, No PhysX, No DoF Test Set 2: DX11, Very High, 4x AA, 16x AF, No PhysX, DoF On
Thomas Soderstrom
Thomas Soderstrom
Thomas Soderstrom is a Senior Staff Editor at Tom's Hardware US. He tests and reviews cases, cooling, memory and motherboards.
89 Comments Comment from the forums
  • SteelCity1981
    Well considering my Core i7 840QM cpu can only support a max of DDR3 1333 i'm pretty at a dead end in terms of upgrading to faster ram.
    Reply
  • cumi2k4
    so....bottom line: don't bother buying higher clock memory unless it cost nearly the same?
    Reply
  • Crashman
    cumi2k4so....bottom line: don't bother buying higher clock memory unless it cost nearly the same?The bottom line is that the price difference is ALREADY so small that anyone who can afford a pair of high-end GPUs should bother to get the good DRAM as well.
    Reply
  • mayankleoboy1
    if i was the owner, i would fire the programmers who developed the memory bottlecked game engine.
    Reply
  • esrever
    CrashmanWow, talk about not reading the article! Here's a hint, from the article you didn't read:Bottom line: Buy the fastest memory you can afford, AT LEAST DDR3-1866, unless you're certain that the slower memory you're buying can be overclocked.unless you care about 150 fps vs 120 fps I don't see the point
    Reply
  • hero1
    Nicely written. My next rig which is going to be in couple months will have the DDR3-1866MHz in it. That's as far as I will go with RAM.
    Reply
  • itzsnypah
    So the point of this article is with a $2k (8gb ram) computer spend an extra 1% ($20) for 1% gaming performance increase that you only get sometimes? Very linear scaling me thinks, sometimes.
    Reply
  • fkr
    Am I reading this correct when I see in metro that dual channel is faster than quad channel. Also crashman I am pretty sure that very few people care how many fps you get in f1 2012 since the game is pretty terrible, metacritic user score of 6.8/10. reading through the reviews of that game it seems to not even be updated since 2011 edition, at least not the physics. I also thought the article says unless your doing eyefinity and intel hd whatever it makes little difference. I also seem to remember about a year ago another article stating how little difference ram made in a system, ram speed that is.
    Reply
  • fkr
    If you are down to your last twenty dollars do you put it towards better ram or CPU. i5 3570 or a 3570k. I think the k edition is better money spent, but maybe somebody cares about f1 2012, crash"in my mothers basement"man.
    jk
    Reply
  • jase240
    I would like to see how faster RAM effects loading times, that's the ONLY reason I can imagine paying a little more. And even then 1866 would probably be fine, considering most will overclock to 2133 well.

    Although in reality 1600 can do the job just fine and it could overclock nicely too if you get the right RAM.
    Reply