Newb ram question/confusion

I've recently built a computer that I've intended to be as future proof as reasonably possible (for me). Specifically its an i7 6700k, with an asrock z170 extreme 6 motherboard. Originally I wasn't too picky on ram, just picking one from the logical increments that happened to be on sale, Gskill Aegis ddr4 2400 ram (2x8 gig). Assembled my build, all works well. Run pcmark8 home conventional 3.0 benchmarking. Get a score of 4550. Looks good to me.

I then read some online discussions saying basically that ram speed is going to be more important perhaps in future games. I then order and receive some corsair vengeance 3200 (2x8 gig) and install those in place of the gskill. I then rerun the same test, expecting a better score. Instead my score has slightly DECREASED. I run it many times and the best I can get is 4544. I did turn on xmp in the bios and the ram is running as advertised.

The gskill ram is G.SKILL 16GB (2 x 8GB) Aegis DDR4 2400MHZ (PC4-19200) 15 15 15 35 at 1.2 volts
The corair ram is Corsair Vengeance LPX 16GB (2x8GB) DDR4 DRAM 3200MHz 16 18 18 36 at 1.35 volts.

I then tested with 3dmark firestrike and got the following results, basically after running several tests and taking the average to report.
Gskill ram at 2400 15 15 15 35 firestrike overall 10682
Corsair ram at 2400 15 15 15 36 (xmp not enabled in bios) firestrike overall 10670
corsair ram at 3200 16 18 18 36 (xmp enabled in bios) firestrike overall 10449
ALL 4 sticks in 32 gigs total 2400 15 15 15 35 firestrike overall 9632

I guess it is apparent that I have no idea what I am doing. Any ideas on what I should do? Are these numbers meaningless? Should I return some ram or run with them all in? I have no idea why with the 32 gigs in all together the benchmark went down...

download cpuz and see what the ram speed is running at under the memory tab. this is to make sure your new kit is running at the faster 3200 speed, if it isnt then you need to go into the bios and set it to that speed.

where exactly did you read this about ram speed and games?

Typically you see very little to no gains with memory speeds. There is a title or so that does respond well, however. That said, memory isn't typically a limiting factor in most anything...I'm also curious where you read that.

Seems like you enabled xmp....the 3200 mhz sticks have a higher CAS at 16...but do post up cpuz screenshot of SPD and memory tabs.

You should not be mixing and matching sticks. You will have to, as you have done, run the faster sticks at slower speeds... kind of a waste.

When I talk about measurable performance gains from faster ram I am speaking about this article, referenced by another poster a few posts down, and maybe other articles that I can't find...
http://www.techspot.com/article/1171-ddr4-4000-mhz-performance/page3.html

I know it is for only a few titles, and in most evident for a dual graphics card system, etc, but it appears that the effect is real. That said, I'm trying to figure out why MY results from the firestrike benchmark get worse, apparently , when I run with the faster ram, at its faster speeds. I'll post the results of cpuz and the benchmark as soon as I'm able to get results and swap things out, though I'm not sure how to post pics etc.

If you want to compare the memory module you would do better to test with a memory specific test like the ram test in aida64.

With that being said the latency of the corsair kit is probably why you are seeing these results. I would try manually setting the timings to 15 15 15 36(if you can stabalize it) and then rerun the tests to see your results.

gkorjax said:

When I talk about measurable performance gains from faster ram I am speaking about this article, referenced by another poster a few posts down, and maybe other articles that I can't find...
http://www.techspot.com/article/1171-ddr4-4000-mhz-performance/page3.html

I know it is for only a few titles, and in most evident for a dual graphics card system, etc, but it appears that the effect is real. That said, I'm trying to figure out why MY results from the firestrike benchmark get worse, apparently , when I run with the faster ram, at its faster speeds. I'll post the results of cpuz and the benchmark as soon as I'm able to get results and swap things out, though I'm not sure how to post pics etc.

Click to expand...

Because you are testing a synthetic benchmark and not a real game... Yes, ARMA is one of those games. For every one of those, there are 99 others that don't budge past margin of error.

So, thank you for your time. I reran my tests and basically saw NO difference in any of the configurations. I was more careful and ran ccleaner each time. The variation between tests withing a configuration was the same as between configurations. I still have idea as to why yesterday I got firestrike results 1000 points less with the 32 gigs of mixed ram. Today the results were the same as for any other setup...perhaps my computer was running something in the background that i missed...

Long story short, I think I am going to return the lower speed Gskill ram and just run with the 16 gigs of corsair. I'm not sure if I should mess with tightening the timings. I don't think it is so important for me.

Thanks again.

Put in the sticks... set XMP profile... ENJOY!

Timings dont effect performance on ddr4 like with ddr3.

It didn't do anything for DDR3 either really...

+1 Just set X.M.P and don't worry about it.