Is this CPU/MOBO combo ideal for this RAM?

Toggle sidebar Toggle sidebar
C

count_rugen

Distinguished
Dec 4, 2008
43
0
18,530
CPU: http://www.newegg.com/Product/Product.aspx?Item=N82E16819116991

MOBO: http://www.newegg.com/Product/Product.aspx?Item=N82E16813128715

RAM: http://www.newegg.com/Product/Product.aspx?Item=N82E16820231476

RAM timing, voltage, and latency numbers confound me. I understand (I think) that for a given motherboard and CPU, having RAM beyond a certain "speed" is pointless as the motherboard and/or CPU can't utilize the capabilities of the RAM. Using the above CPU/MOBO as an example, could one of you intelligent folk please explain how the RAM listed will/will not perform to it's maximum potential and why or why not? Will it actually operate at 2133?



P.S.--yes, I have googled the hell out of memory timings and so on. I think I just need someone to explain this to me in layman's terms. :)

Thanks!
 
Solution
JackNaylorPE
You understanding is based upon misinformation. The myth that RAM speed has no effect comes from misinterpretation of original DDR3 reviews where reviewers said that the relatively small increase in speed was not worth the large increase in price. With that price difference no longer relevant so many years later, the logic no longer applies but is continually reposted out of context. RAM speed has significant effect in CAD, rendering, animation, large spreadsheets / databases, graphics editing .... in gaming it's affect is harder to classify as it varies by game..

In gaming, the system bottleneck is most often the GFX card.... so faster RAM makes no difference if you have a budget - lower midlevel card. Move up to a 2nd GFX card...
Sort by date Sort by votes
timeconsumer

timeconsumer

Splendid
Nov 28, 2011
3,130
0
22,960
Yes that board can run that memory at the rated speed. However, there's really no point in getting memory above 1600Mhz Cas 9, it will give you almost no real world performance gain. And a Z97 board won't serve a lot of purpose on a locked processor either. You might as well save the $30.
 
Upvote 0 Downvote
C

count_rugen

Distinguished
Dec 4, 2008
43
0
18,530


Why would it provide no real world performance gain? If there's an article you can send my way, that'd be awesome. RE: the mobo. You'd recommend an H97 then? Surprisingly, there aren't as many H97 mobo's on sale (that support RAM at 2133).
 
Upvote 0 Downvote
JackNaylorPE

JackNaylorPE

Titan
BANNED
Oct 30, 2004
32,255
14
99,960
You understanding is based upon misinformation. The myth that RAM speed has no effect comes from misinterpretation of original DDR3 reviews where reviewers said that the relatively small increase in speed was not worth the large increase in price. With that price difference no longer relevant so many years later, the logic no longer applies but is continually reposted out of context. RAM speed has significant effect in CAD, rendering, animation, large spreadsheets / databases, graphics editing .... in gaming it's affect is harder to classify as it varies by game..

In gaming, the system bottleneck is most often the GFX card.... so faster RAM makes no difference if you have a budget - lower midlevel card. Move up to a 2nd GFX card or 3...4, and RAM plays a more significant role. Also, while most testing is done recording average fps, when we look at minimum fps, we see much more of an impact. The STALKER series was a set of games so affected.

In your average 1 GFX card gaming box, the effect of faster RAM can range from minimal to large.

Here we see a 0.08% increase in Metro LL going from 1600 to 2400.

http://www.tomshardware.com/reviews/32-gb-ddr3-ram,3790-10.html

Yet in F1 we see a whopping 11+% increase in F1 going from 159 fps to 177

image006.png


Now keep in mind that an 11% increase is about what you get moving from a 970 to a 980 at a cost of $230

Here's a comparison between 1600 CAS 9 and 2400 CAS 10 2x4GB sets

1600 Ripjaws $60 - http://pcpartpicker.com/part/gskill-memory-f312800cl9d8gbrl
2400 Ripjaws $65 - http://pcpartpicker.com/part/gskill-memory-f32400c10d8gzh

So here ya get roughly the same performance increase in F1 for $5 as you would moving from a 970 to 980 .... spending 4230 is something to think about...spending $5 , not so much.

It's no longer the dawning days of DDR3 when high spec RAM was rare because of low yields and the laws of supply and demand kept prices high.... repeating the "It's not worth the performance increase due to the large difference in price" mantra is outdated. Due to production line improvements over the years, the yield om high spec RAM is very good.

Back in the day, if a set didn't pass at one speed, it was tested at a lower speed till it passed and it was sold. Now so much of it is passing at the higher speeds that manufacturers have a glut and seek only to eke out a small price premium

At today's prices, the $5 increase is what I would call the proverbial "no brainer". I spent $10 more cause I wanted the better CAS timings of the Mushkin Redlines as well as the lower profile and Hynix modules.

http://pcpartpicker.com/part/mushkin-memory-997083

As far as the rest, I see no logic in spending $200 for a 4590 when a 4690k is just $20 more
http://pcpartpicker.com/part/intel-cpu-bx80646i54690k

Given that the 2400 set is same price as the 2133 set, the 2400 has slightly lower latency

CAS x 1000 / DDR speed = nanoseconds

9 x 1000 / 2133 = 4.22 ns
10 x 1000 / 2400 = 4.17 ns

 
Upvote 1 Downvote
Solution
timeconsumer

timeconsumer

Splendid
Nov 28, 2011
3,130
0
22,960
http://www.tomshardware.com/reviews/memory-bandwidth-latency-gaming,3409-10.html

"Getting back to the games that were affected by memory performance, only one title exhibited differences significant enough to be noticeable during real-world play. Even then, the average frame rates were so high that your eyes (and displays) would need to be about twice as fast as ours to realize the real-world benefits of faster RAM."

The one game that really showed the big benefit in their tests was the game you linked above. Most other games had no benefit, and that was dual-channel 1600 vs quad-channel 2133.
 
Upvote 0 Downvote
JackNaylorPE

JackNaylorPE

Titan
BANNED
Oct 30, 2004
32,255
14
99,960
Again, not what was determined nor what was said. 40% of the games tested showed significant improvements'

Two out of five game tests, F1 2012 and The Elder Scrolls V: Skyrim, showed us that that both bandwidth and latency ca n influence frame rates significantly.
Click to expand...

Also ....let's see what else they said that was relevant ... seems we can discount another 2 games

The performance of the first two titles is most consistently associated with the speed of a given machine's graphics subsystem, so it makes sense that we don't see a big impact from memory bandwidth or timings.
Click to expand...

Every game is different, some are CPU bound, most are GFX bound and some are RAM bound. Come up short in any area, and you will be constrained in some games.

But, again, this tested was limited:

-Only 5 games tested
-No 2x, 3x, 4x SLI / CF tested
-No minimum frame rates tested
-No 144 Hz screens used

Nothing's changed here, this has been with us well before DDR3

http://www.anandtech.com/show/2792/12

22.3 % (SLI) increase in minimum frame rates w/ C6 instead of C8 in Far Cry 2
18% (single card) / 5% (SLI) increase in minimum frame rates w/ C6 instead of C8 in Dawn of War
15% (single card) / 5% (SLI) increase in minimum frame rates w/ C6 instead of C8 in World in Conflict

Also see http://www.bit-tech.net/hardware/memory/2011/01/11/the-best-memory-for-sandy-bridge/1

3.6% increase in Crysis
9.5% increase in min frame rates in Civ 5 (only 5.2% in avg frame rates)
 
Upvote 0 Downvote
timeconsumer

timeconsumer

Splendid
Nov 28, 2011
3,130
0
22,960
For the Anandtech article, we all know that Cas latency affects games far more than clock speed. But buying 2133 or 2400 speed memory doesn't mean you'll be able to get it to 1600 C6. Is it more likely than with 1600 Cas 9? Yes. Is it guaranteed? No way.

As far as the bit-tech article. The Crysis numbers were gains of less than 1 FPS and the Civ V numbers within 2 FPS. That's within the margin of error in testing as far as I'm concerned.

All things being equal I'll take faster clocks anyday. But I wouldn't pay a $10 premium for margin-of-error gains.
 
Upvote 0 Downvote
Tradesman1

Tradesman1

Legenda in Aeternum
Jun 5, 2013
67,883
40
126,960
I too would suggest going to the 4690K, as for the DRAM, the higher data rate can and does improve performance every action uses more DRAM so more is done at once. And while people can argue about gaming, yes, gaming can and most often does increases in performance with higher data rates, they may be small and can be larger, though I see nothing from Count_Rugen indicating gaming is the primary purpose of the rig ;) it may well be. You will see much greater gains performance wise with the faster DRAM in things like multi-tasking (which may involve gaming), video, imaging, CAD, GIS, VMs, etc - apps that use lare data sets or are in of themselves DRAM intensive.

With todays prices you can often pick up 2400 or 2133 at the same price as 1600 so it really makes to buy entry level DRAM when you can have faster (as long as your CPU can run it. If you stick with the 4590 I'd suggest 1866, 2133 is a maybe and can always be run at 1866 if need be.
 
Upvote 1 Downvote
JackNaylorPE

JackNaylorPE

Titan
BANNED
Oct 30, 2004
32,255
14
99,960


It's funny that you talk about small differences being margin-of-error level.... what about the price difference ? What level would you call that ?.... $5 on even a low budget $800 build is 0.6% .... How is 5%, 7%, 10, 11% all insignificant and yet 0.6% price increase is a major thing ? What about a more typical $1500 build.... 0.3% increase in cost. That's 1/15th of what you labeled an "acceptable margin-of-error".

What we have here is a price difference that is certainly much more miniscule than the consistently observed performance advantage.... or what is routinely referred to as a "Strong Positive ROI". The return is certainly bigger than the investment.

Now let's get to some accuracy issues:

1. As THG stated.... both timings and speed / bandwidth can influence performance. No need to sacrifice one to get the other. I wouldn't think of taking 2400 down to 1600 to lower CAS. As for CS 6, I have never seen any 2x4GB CAS 6.... Mushkin is the only ones partpicker shows as selling CAS 7 but take 2400 CAS 10 down to 1600 CAS 7 ? Why ?

7 x 1000 / 1600 = 4.38 ns
10 x 1000 / 2400 = 4.22 ns

2. Read the rest of the anandtech article, not just the cherrypicked parts. Anandtech showd 4.5 fps in average fps from 1066 to 1866 with CAS 7, 6 fps in minimum fps.... that still below the margin-of-error ? Interesting also that all those years later bit-tech's 5% aligns nicely with anandtech's 7%.

fc2na.jpg


Although the differences are minor, we do notice a linear scaling in frame rates as we improve both bandwidth and latencies. Average frame rates are up 7% and minimum frame rates by 10% when moving from 1066 C7 to 1866 C7.
Click to expand...

I don't know that everyone would consider 7% and 10% minor but i think we can get a consensus that 0.6% and 0.3% is very minor. Not all games did that much but FC2 showed 10% increase in min fps and HAWX showed 15% !

3. It's a $5 premium, why did you feel the need to double it ? I recommend the $65 Gskill 2400s over the $65 Gskill 2133's over the $60 Gskill 1600s

I didn't spend $10 to just get better timings, I very clearly stated the additional reasons why I chose to spend the extra $5 ($10 total) were numerous reasons:

a) My box runs as much as 10-12 hours a day running CAD, maybe 1 or 2 gaming if I'm not too tired. CAD certainly benefits from the faster RAM
b) The extra $5 got me better aesthetics and lower profile
c) The extra $5 got me Hynix modules and nothing OCs or takes additional voltage like Hynix
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom