DDR3 RAM Gaming Analysis

RAM frequency is only going to play into the equation in a significant manner at extremely high BClocks where slower RAM could be an OCing bottleneck. I found it odd the reviewer chose 4GB of memory as opposed to 3GB or 6GB. RAM capacity will be a bigger factor than latencies and overall RAM frequency under the testing done here and especially so in actual game play.
 
RAM frequency is only going to play into the equation in a significant manner at extremely high BClocks where slower RAM could be an OCing bottleneck. I found it odd the reviewer chose 4GB of memory as opposed to 3GB or 6GB. RAM capacity will be a bigger factor than latencies and overall RAM frequency under the testing done here and especially so in actual game play.
Click to expand...

It is true that RAM frequency is more critical than timings and frequency but that is not the point of the article. The point is to see if RAM and or timings can become the bottleneck in real world scenarios, and if yes then by how much and in what games.
 
CH33TAH said:
The point is to see if RAM and or timings can become the bottleneck in real world scenarios, and if yes then by how much and in what games.
Click to expand...
Running only benchmarks are not real world scenarios. I'm not criticizing the person doing the review as I know first hand how tedious this type of testing is. But, to say "real world" is not really accurate.

Also, you said specifically :
CH33TAH said:
The point is to see if RAM and or timings can become the bottleneck in real world scenarios,
Click to expand...
The testing was only done with a 4GB set up on a system optimally designed to run 3GB, 6GB, or 12GB. So, it's impossible to determine if more or less memory would effect the results. Using a tool like Process Explorer could have helped clarify the results a great deal.
 
Running only benchmarks are not real world scenarios. I'm not criticizing the person doing the review as I know first hand how tedious this type of testing is. But, to say "real world" is not really accurate
Click to expand...

I agree with that. Only the test that were performed with FRAPS can be considered real world.

The testing was only done with a 4GB set up on a system optimally designed to run 3GB, 6GB, or 12GB. So, it's impossible to determine if more or less memory would effect the results. Using a tool like Process Explorer could have helped clarify the results a great deal.
Click to expand...

I only had 4 GB of RAM. If I had a set of 6 Gigs of RAM then I would have used that :)

Actually I would do a 4 GB vs 6 GB RAM test at the same timings and frequency to see how RAM amount plays into the equation.
 
CH33TAH said:
I agree with that. Only the test that were performed with FRAPS can be considered real world.



I only had 4 GB of RAM. If I had a set of 6 Gigs of RAM then I would have used that :)

Actually I would do a 4 GB vs 6 GB RAM test at the same timings and frequency to see how RAM amount plays into the equation.
Click to expand...

Hey, I did not mean to come across as raining on your parade. One thing I discovered the hard way is that there is no perfect methodology. And, when you put something out on the net, there are a gazillion people there to dissect it for you. And, many of them forget to be civil about it.

I was actually a little surprised by some of the increases you showed that seem to arise from timings and RAM frequency. My guess is that if I tried to present the same info you posted, I'd get crucified for it.

Good work IMO. :cool:
 
those charts are grossly scaled with the differenece in measurements easily being attributed to uncertianty margines. And as Mike pointed out, hes running dual channel on a system capable of triple channel.

I'm calling shins on the whole review.
 
those charts are grossly scaled with the differenece in measurements easily being attributed to uncertianty margines. And as Mike pointed out, hes running dual channel on a system capable of triple channel.

I'm calling shins on the whole review
Click to expand...

The graph scaling has nothing to do with the fact that Far Cry 2 is showing more than 15 % improvement. And 15 % is not "uncertainty margin" by any stretch of the imagination. True, that it is Dual channel and ideally you would want to be running tripple channel with I7 and X58 but that takes nothing away from the results. And besides, Far Cry 2 and Fall Out 3 are the only titles that I would consider showing "noticeable change". I actually intend to go tripple channel and re test some Far Cry 2 and Fall Out 3 but I doubt the results will be much different.

Hey, I did not mean to come across as raining on your parade. One thing I discovered the hard way is that there is no perfect methodology. And, when you put something out on the net, there are a gazillion people there to dissect it for you. And, many of them forget to be civil about it.
Click to expand...

I hear you. I take all criticism as constructive :)

My guess is that if I tried to present the same info you posted, I'd get crucified for it.
Click to expand...

I know. I have always been a believer that RAM frequency and timings hardly play a role just like most other benchers and gamers. But Far Cry 2 is one game that sort of changed my mind regarding this. When running Tri SLI and highly over clocked I7, it is not impossible to run into a memory bottleneck in some gaming titles. I had noticed this a while back and wanted to run some tests. That is all I have done and presented my honest results ;)
 
You must log in or register to reply here.
Back
Top