RAM - 1600 MHz vs 1833 MHz

[Harshith Mohan]

Hi!
I'm building a pc and I wanted to know what speed of ram should I use? 1600 or 1833 MHz.. I'll use it for gaming and virtual machines..
Processor - i7-4770k

[hawthorne]

Higher speed memory won't affect you fps, I would go for the cheaper

[flibberdipper]

1600MHz will be fine for you.

[GrantSor+1]

1600 is enough. And CL7 of 1600mhz beats 2133mhz ram

[Nikeshp79]

You can always overclock the 1600MHz RAM if your MOBO allows it. And considering 1833MHz RAM is overclocked 1600MHz RAM just go for the cheapest option.

 

Personally I'm currently in the same situation as you and since the RAM sticks I'm choosing between are similarly priced I'm simply going for the one that looks the best and I will overclock it myself.

[Harshith Mohan]

So, How about Corsair Vengeance?

[DoubleY]

1600mhz would be fine. I overclocked my 1600mhz to 1866mhz

[DJHoro]

I think you meant 1866 Ram. Anyways some games are starting to take advantage of higher ram. When shopping for ram I try to look at a few things. Price, good reviews, reliability, lower cas latency, and if it matches the look of my MoBo. If you see a good reliable set of 1600MHz thats on sale, then get that. If you see an 1866MHz set thats just a few $$$ more then get that. Just try not to pay way more money for a higher frequency ram.

 

1600 is enough. And CL7 of 1600mhz beats 2133mhz ram

I think this is a little misleading. There are certain applications where higher frequency ram is better and timings can always be adjusted.

[BiG StroOnZ]

Problem with people recommending 1600MHz memory is that 2400MHz memory costs about the same and does offer additional performance. There are games that take advantage of higher speed memory as well. Here's 8GB of G.Skill Ripjaws X 2400MHz for $80

 

Also, for the record:

 

1) Faster memory costs the same, there is no difference in price between 1600MHz to 2400MHz. If there is it's give or take a few dollars. Because of this fact getting the slower memory is entirely nonsensical.

2) Going from simply 1600MHz to 2133MHz memory just from doing daily tasks on the desktop has simply felt much faster than 1600MHz memory. This is from startup to shut down. Opening files, folders, programs. Any type of encoding. Extracting .zip or .rar files. Moving files, copying files. All were much faster than 1600MHz. Game FPS went up by a noticeable amount. What is most intriguing is the minimum fps. Meaning when you are gaming sometimes the framerates drop. Having faster memory has increased my minimum framerates by a quite noticeable amount. Meaning when you do drop in framerates it's not nearly as bad as it could be if you were to have slower memory.

3) Did I mention faster memory costs no more than 1600MHz? Yeah, I think I did. So why in the world would you buy a 1600MHz kit today when they cost the same exact price as 2133 or 2400MHz kits. There is no point in doing so.

Faster memory has benefits in games and many daily computing tasks; the difference is quite noticeable too. This idea of faster memory not being beneficial is just heresy from word of mouth that gets dragged on continually from one person to another because of old outdated tests that are irrelevant at this point. If faster memory didn't do anything we would all still be using DDR2 or DDR or even 1333MHz DDR3. We aren't because faster memory has proven to be beneficial. If it wasn't there would be no need to move to the next generation of memory. Why move to DDR4 like what is going to be the next generation of memory movement? Why, because it will have more bandwidth and thus be faster. There are going to be other benefits but mainly it's going to be a great deal faster. Point being, memory evolves and frequency is an important factor.

Here's some tests from other users and a test done by Corsair with BF4:

 

 

 

 

 

 

 

Originally Posted by nleksan 

Here are a few of my results...

System:
- 3930K @ 4.5Ghz (usually 4.8)
- Rampage 4 Extreme
- 16GB G.Skill Ripjaws Z DDR3 2133 9-11-10-27
- EVGA GTX 670 FTW 2GB @ 1326/7460
- Samsung 830 256GB SSD (OS/Sys/Apps)
- WD Blue 320GB (Music/Video)
- HGST 7K500 320GB (Apps)
- WD RE3 1TB (Games)
- Creative Titanium HD
- LG/Hitachi Slim DVD-R/W
- NZXT HALE90 850W PSU
- TRIPP-LITE ISO-BAR-4 ULTRA Line Conditioner/Surge Suppressor/Voltage Regulator

I used the above system for the below tests, with all clocks identical with just the RAM changing. The system is watercooled so no drops in Kepler Boost or anything like that.

*TESTS*

(RAM SPEED + TIMINGS - AVERAGE FPS - MINIMUM FPS - MAXIMUM FPS)
All tests done in 1080p using a Dell P2212Hb connected via DL-DVI-D, all game settings set at maximum unless otherwise noted.

HALF-LIFE 2 EPISODE 2 (CPU @ 3.4Ghz, GPU @ Stock FTW Speeds)
1600 6-7-7-19 - 198.5 - 131.8 - 233
1600 9-9-9-24 - 191.3 - 124.9 - 219
1866 8-8-8-24 - 204.1 - 137.7 - 242
2133 9-11-10-27 - 217 - 147.1 - 259
2133 9-10-9-26 - 226 - 154.3 - 266
2360 9-12-10-29 - 231 - 159.7 - 283

Half-Life 2 and it's countless derivatives (mods, etc) all seem to be CPU Bound at this point, as I see a perfectly linear relationship between a CPU's speed and FPS.


FAR CRY 3 (CPU @ 4.5Ghz, GPU @ 1326/7460)
1600 6-7-7-19 - 52.2 - 26.1 - 93
1600 9-9-9-24 - 50.8 - 24 - 86
1866 8-8-8-24 - 54.1 - 27.9 - 96
2133 9-11-10-27 - 57.4 - 31.1 - 105
2133 9-10-9-26 - 58.7 - 33 - 107
2360 9-12-10-29 - 60.2 - 35.5 - 108

The biggest thing with FC3 is the increase in smoothness. The slower memory feels choppy at times, such as when you get into a firefight. The faster memory never has this problem. Also, with 2133 and above, I get zero texture "pop in", yet it's present with lower memory speeds.


I did recordings of 9 games, 11 benchmarks, and timed start up/shutdown/opening (Firefox with 25tabs/Photoshop/Paint.Net/Chrome with 25tabs,and a half dozen other things), and a few other things.

There is not a single instance in which the 1600 9-9-9-24 didn't come in dead last. The differences ranged from "benchmark-noticeable" to "Wow that's a huge improvement".


I simply don't recommend getting 1600 when the option for faster memory is there, especially if you have an IVB platform. The tests above are X79, and I have done the same tests on a 3770K + GA-Z77X-UP7, 3770K + Maximus 5 Extreme, 3570K + Extreme6 (for Ivy), a 2700K + Maximus 4 Extreme-Z, 2600K + G3.Sniper3, 2500K + Extreme9 (for Sandy), a Phenom II X4 980BE + ASRock 990FX Fatal1ty, 1100T + Crosshair V Formula-Z, 960T + 990FX Sabertooth R2.0, 1090 + M5A99X, 965BE + Gigabyte 990FX UD5(UD7, can't remember), and 8350 + Crosshair V Formula, 8130 + Sabertooth, 6100 + Extreme6, 4100 + Extreme3.

These have been over the course of 16mo, and variables change, not all systems had all tests run (most only had 2-3 games and a few benchmarks), and it's not a controlled experiment. Still, the results are only compared against the results from the same system, so they are perfectly valid.

Every single system wanted the fastest memory possible, although the Phenom II systems had to be controlled for timings by ensuring that the actual latency in ns was better than the prior test (which means most of the Phenom II tests are more about timings for a given speed than speed itself, although 1800 7-8-7-26 was always the fastest, beating 1600 6-7-6-19 by 9.3% on average).

I will try to get the rest of the results all compiled on a single spreadsheet..

 

 

 

Originally Posted by EastCoast 
Here are some results using the old version of that benchmark (because that's what I used with DDR3 1600). One thing I will note is that the results of the DDR3 1600 does look a bit lower then normal but that's the result I got.

Heaven Benchmark

DDR3 1600
Min 11 FPS
Avg 54.7 FPS
Max 117.5 FPS


DDR3 2133
Min 28.6
Avg 55 FPS
Max 117.7 FPS

[DJHoro]

Damn I didn't realize higher frequency ram got so cheap. 

[That Norwegian Guy]

The only RAM combinations that are better than the baseline 1600mhz/ CL9 are:

 

2400 CL10

2133 CL9

1833 CL9

1600 CL8

1600 CL7

 

I swear by Crucial Ballistix Tactical 1600mhz CL8 8GB sticks as they are the most desirable in terms of value, and Crucial is hands down the most rock solid RAM brand in my long experience. (Especially recently) I also wouldn't want to be caught dead next to a stick of G.Skill kiddie edition (combs) RAM, or other immature dumb looking DIMMs.

[patrickjp93]

1600 is enough. And CL7 of 1600mhz beats 2133mhz ram

No it doesn't, although the difference is tiny. Absolute latency is CAS/MHz  

7/1600 = 0.004375

9/2133 = 0.0042194093

Where lower is better. 

 

And since in most cases you can lower the CAS by 1 on your sticks following some simple online guides, you can do even better than this.

I have 16 GB G.Skill Trident X 2133MHz specced at 9 CAS, but you can tighten it to 8 and stay perfectly stable. 

 

Although, in this case, starting with G.Skill Trident X 1600MHz 7 CAS, tightening to 6 produces better results than tightening 2133MHz to 8 CAS.

8/2133 = 0.003750586

6/1600 = 0.00375

And the 1600 sticks use less energy and will thus likely last longer.

 

I did see some whackadoo on extremeoverclocking pull off 2800 at CAS 9 (I believe this was a pair of 4GB A-Data) for an absolute latency of 0.0032142857.

 

But 1600 CAS 7 should be fine for most anything.

[patrickjp93]

The only RAM combinations that are better than the baseline 1600mhz/ CL9 are:

 

2400 CL10

2133 CL9

1833 CL9

1600 CL8

1600 CL7

 

I swear by Crucial Ballistix Tactical 1600mhz CL8 8GB sticks as they are the most desirable in terms of value, and Crucial is hands down the most rock solid RAM brand in my long experience. (Especially recently) I also wouldn't want to be caught dead next to a stick of G.Skill kiddie edition (combs) RAM, or other immature dumb looking DIMMs.

And yet G.Skill actually offers life-time warranties for its DIMMs and has a 95+% 5-star rating on Newegg across all its products. Crucial is in the mid-80s, which is still highly respectable. And the Trident series exists because of people like you who can't deal with the saw-tooth look.

[patrickjp93]

@BiG StroOnZ Interesting... Higher speed can mean higher bandwidth, but that's usually only useful for memory-intense applications. Loads of AI maybe?

[patrickjp93]

Hi!

I'm building a pc and I wanted to know what speed of ram should I use? 1600 or 1833 MHz.. I'll use it for gaming and virtual machines..

Processor - i7-4770k

Buy a pair of 2133 with CAS 9 and try to tighten the CAS to 8 manually. If it can't do that, you can try underclocking down to 1600 CAS 6. The latter is actually better in terms of absolute latency, but buying the 2133 stick gives you more options since the chip is rated for that speed.

[LogicalDrm]

1600 to 1866 is not worth it if there's any difference in pricing. If you want more MHz, go for 2133. Even then difference in benchmarks is about 2% and in gaming situation few fps (tested in BF3).

[That Norwegian Guy]

And yet G.Skill actually offers life-time warranties for its DIMMs and has a 95+% 5-star rating on Newegg across all its products. Crucial is in the mid-80s, which is still highly respectable. And the Trident series exists because of people like you who can't deal with the saw-tooth look.

 

Newegg is not really relevant, Crucials American market DIMMs are made in Puerto Rico, the entire rest of the world gets DIMMs from Taipei.

[-iSynthesis]

1600; unless running an apu you wont see any better performance, and you usually get better cas with lower speed ram.

[patrickjp93]

Newegg is not really relevant, Crucials American market DIMMs are made in Puerto Rico, the entire rest of the world gets DIMMs from Taipei.

Ratings don't matter? If you say so... And who cares about Crucial if they don't even make RAM in the lowest CAS of each speed? Why anyone settles for less than the best-in-class RAM when the price difference is negligible is beyond me. I mean, with DDR4 coming out, getting a value set now to upgrade to a new form factor only 2-3 years from now, fine, makes sense. If you're building any more long-term than that, the lasting performance is more than worth it. As for Tai Pei, you might want to research a bit more. A number of builders moved to Japan to get the better onboard controllers made with the same premium silicon as Samsung back in the mid 2000s.

[patrickjp93]

1600; unless running an apu you wont see any better performance, and you usually get better cas with lower speed ram.

CAS on its own is not how to measure RAM performance, although 1600 is also the sweet spot for performance/watt & performance/dollar. Cas/Speed = absolute latency = absolute performance where low is better. I have the urge to smack people when they say higher RAM speed helps APUs (including Intel's onboard HD & Iris Graphics). It only helps if the absolute latency decreases. Otherwise it's a wash.

[patrickjp93]

This, You can buy a 2133 MHz kit, you'd downclock as he said, but since it's less latency (2133->CL 8.4375 ns vs 1600-> 8.75 ns) you'd stay at that frequency to keep a good memory bandwidth

8.75ns? where did you get that number from? I even did the calculations for you lol.

[That Norwegian Guy]

Ratings don't matter? If you say so... And who cares about Crucial if they don't even make RAM in the lowest CAS of each speed? Why anyone settles for less than the best-in-class RAM when the price difference is negligible is beyond me.

 

Ratings don't matter for a market consisting of 500 million people when compared to the market consisting of 7 billion people. When both markets have two different products with the same name, reputation is attached to the most important, biggest market.

 

Crucial's 1600Mhz CL8 DIMMs are 1.35v, you can set CL7 yourself by upping it to 1.5v, like the already tuned CL7 memory ships at.

[patrickjp93]

Ratings don't matter for a market consisting of 500 million people when compared to the market consisting of 7 billion people. When both markets have two different products with the same name, reputation is attached to the most important, biggest market.

 

Crucial's 1600Mhz CL8 DIMMs are 1.35v, you can set CL7 yourself by upping it to 1.5v, like the already tuned CL7 memory ships at.

Sometimes you can get CL7, but Crucial is known for being far more difficult to tinker with. Also, what does Crucial do well? Cheap, low-performance SSDs?

[BiG StroOnZ]

@BiG StroOnZ Interesting... Higher speed can mean higher bandwidth, but that's usually only useful for memory-intense applications. Loads of AI maybe?

 

Not really only memory intense applications. I've seen noticeable improvements going from 1600MHz to 2133MHz just by doing daily tasks. I've also saw a noticeable increase in minimum framerates in games. 

 

No it doesn't, although the difference is tiny. Absolute latency is CAS/MHz  

7/1600 = 0.004375

9/2133 = 0.0042194093

Where lower is better. 

 

And since in most cases you can lower the CAS by 1 on your sticks following some simple online guides, you can do even better than this.

I have 16 GB G.Skill Trident X 2133MHz specced at 9 CAS, but you can tighten it to 8 and stay perfectly stable. 

 

Although, in this case, starting with G.Skill Trident X 1600MHz 7 CAS, tightening to 6 produces better results than tightening 2133MHz to 8 CAS.

8/2133 = 0.003750586

6/1600 = 0.00375

And the 1600 sticks use less energy and will thus likely last longer.

 

I did see some whackadoo on extremeoverclocking pull off 2800 at CAS 9 (I believe this was a pair of 4GB A-Data) for an absolute latency of 0.0032142857.

 

But 1600 CAS 7 should be fine for most anything.

 

I always mix it up, but I BELIEVE it is: 1000 / Memory Frequency Actual [ie 800 for DDR3-1600] x CAS...

 

so....

 

 

DDR3-2400 CAS 9 7.5ns

DDR3-2400 CAS10 8.33ns

DDR3-2400 CAS11 9.17ns

 

DDR3-2133 CAS9 8.44ns

DDR3-2133 CAS10 9.38ns

DDR3-2133 CAS11 10.32ns

 

DDR3-1600 CAS7 8.75ns

DDR3-1600 CAS8 10.0ns

DDR3-1600 CAS9 11.25ns

[patrickjp93]

 

Not really only memory intense applications. I've seen noticeable improvements going from 1600MHz to 2133MHz just by doing daily tasks. I've also saw a noticeable increase in minimum framerates in games. 

 

 

I always mix it up, but I BELIEVE it is: 1000 / Memory Frequency Actual [ie 800 for DDR3-1600] x CAS...

 

so....

 

 

DDR3-2400 CAS 9 7.5ns

DDR3-2400 CAS10 8.33ns

DDR3-2400 CAS11 9.17ns

 

DDR3-2133 CAS9 8.44ns

DDR3-2133 CAS10 9.38ns

DDR3-2133 CAS11 10.32ns

 

DDR3-1600 CAS7 8.75ns

DDR3-1600 CAS8 10.0ns

DDR3-1600 CAS9 11.25ns

 

Yeah shortly after I fell asleep I remembered in a dream the actual clock rate of RAM is half the listed, so effectively double my posted numbers. And, since the speed is in Megahertz yes, multiplying by 1000 gives you perfect nanoseconds. As long as the main structure (latency/frequency) is correct then anyone can understand what you're saying.