- Oct 30, 2004
- 32,252
- 14
- 99,960
We of see it posted that RAM speed doesn't matter and that RAM amount doesn't matter based upon some published but limited in scope reference. Fortunately, every once and a while, someone decides to delve a little deeper.
THG showed that when comparing F1 and Metro 2033 and saw 0% change in fps with Metro but 11% gains w/ F1 when going from 1600 to 2500 . A recent youtube video was discussed here where GTAV / Witcher 3 showed substantial gains from 8 GB to 16 GB.
In my observations:
a) Increases are game specific
b) Minimum fps is affected more than average
c) Multicard performance is affected more than single card performance
Most tests which come to the conclusion that RAM has no impact, usually fail to address the above items in their testing. Extremetech did an article earlier this months which does get into much of this; with much of what they are reporting taken from the 2nd linl
http://www.extremetech.com/gaming/222483-how-much-ram-do-you-need-should-you-upgrade-it-and-will-it-speed-up-your-pc
http://www.eurogamer.net/articles/digitalfoundry-2016-intel-skylake-core-i5-6500-review
The results (testing included 7 games):
1. Avg fps increased from 2 to 18% going from DDR4-2133 to 3066/3200
2. Min fps increased from 7 to 22% going from DDR4-2133 to 3066/3200
This confirms two of the three postulations above:
Another test was referenced with DDR3 with similar results. Fallout 4 increased 16% from 1600 to 2400 topping THGs test w/ F1. They also noted that Intel base builds do better in this respect than AMD. It also charts procing pretty much confirming what has been true since 2013 that cost wise, the jump to 2400 can be easily justified from 1600 to 2400 with just a $13 increase in system cost (<1% on typical gaming system). DDR4 is a bit steeper costing $25 from 2400 to 3200.
The also touched on another thread we have had here (8Gb or 16GB) writing:
As memory impact is a topic often ignored, we tend to still see many posts relying on test of inadequate scope from which it is easy to draw incorrect conclusions. More wide scope investigations like these, hopefully including the impact of multiple GFX cards, would put us all in a position to make better purchase decisions.
THG showed that when comparing F1 and Metro 2033 and saw 0% change in fps with Metro but 11% gains w/ F1 when going from 1600 to 2500 . A recent youtube video was discussed here where GTAV / Witcher 3 showed substantial gains from 8 GB to 16 GB.
In my observations:
a) Increases are game specific
b) Minimum fps is affected more than average
c) Multicard performance is affected more than single card performance
Most tests which come to the conclusion that RAM has no impact, usually fail to address the above items in their testing. Extremetech did an article earlier this months which does get into much of this; with much of what they are reporting taken from the 2nd linl
http://www.extremetech.com/gaming/222483-how-much-ram-do-you-need-should-you-upgrade-it-and-will-it-speed-up-your-pc
http://www.eurogamer.net/articles/digitalfoundry-2016-intel-skylake-core-i5-6500-review
The results (testing included 7 games):
1. Avg fps increased from 2 to 18% going from DDR4-2133 to 3066/3200
2. Min fps increased from 7 to 22% going from DDR4-2133 to 3066/3200
This confirms two of the three postulations above:
Another test was referenced with DDR3 with similar results. Fallout 4 increased 16% from 1600 to 2400 topping THGs test w/ F1. They also noted that Intel base builds do better in this respect than AMD. It also charts procing pretty much confirming what has been true since 2013 that cost wise, the jump to 2400 can be easily justified from 1600 to 2400 with just a $13 increase in system cost (<1% on typical gaming system). DDR4 is a bit steeper costing $25 from 2400 to 3200.
The also touched on another thread we have had here (8Gb or 16GB) writing:
As memory impact is a topic often ignored, we tend to still see many posts relying on test of inadequate scope from which it is easy to draw incorrect conclusions. More wide scope investigations like these, hopefully including the impact of multiple GFX cards, would put us all in a position to make better purchase decisions.
Facebook
Twitter
Instagram