DDR3 1600 vs 2400

Igg

[H]ard|Gawd
Joined
Oct 4, 2002
Messages
1,338
Just upgraded from 2600k on z68 to 4790k 97z

Any thoughts on the reusuing my corsair vengance ram that's 16GB @ 1600? or should I try and spring for the 2400 stuff?

Any thoughts on the speed difference being noticeable on fps?
 

Tsumi

[H]F Junkie
Joined
Mar 18, 2010
Messages
13,581
In a 60 fps game, you might see a 1 fps increase. Maybe 2.

In a CPU dependent game with lots of units (like RTS with massive unit battles), you might see a 2-5% gain.
 

Nenu

[H]ardened
Joined
Apr 28, 2007
Messages
20,223
I moved from 1600 CL6 to 2400 CL11 but am running at 2200MHz due to Sandybridge limitation.
Cant tell the difference at all.
 

evilsofa

[H]F Junkie
Joined
Jan 1, 2007
Messages
10,078
If you are gaming with the graphics on the CPU, it can make a big difference. If you are gaming with a video card, it does not make any difference at all, until you are using very high-end mutliple-card setups.
 

trick0502

Supreme [H]ardness
Joined
Apr 17, 2006
Messages
5,525
I've been wondering this myself. I have 16gb 1600 oced to 1866 in my system now. I also have 8gb 2400 sitting in the box. I haven't really have had time to do testing.
 

defaultluser

[H]F Junkie
Joined
Jan 14, 2006
Messages
14,399
You definitely wouldn't be able to pick it out in a side-by-side unlabeled test drive.

I/O-heavy operations like compressing files see up to 5-10% improvement.

Mixed operations like games and video encoding see less than 5% improvement.
 

munkle

[H]F Junkie
Joined
Jan 16, 2005
Messages
11,800
I went with 2100mhz ram because it was only about $15 more (for 16gb) than 1600mhz and who knows if broadwell will require faster ram so I can always reuse it in the future. You probably wont ever notice a difference so I wouldn't spend big bucks on faster ram.
 

craige4u

Weaksauce
Joined
Sep 3, 2010
Messages
102
There is a small benefit with higher Mhz, but then it doesn't warrant purchase of a new set of Ram when you are already having one.... Just use what you are having as it doesn't make much sense to buy a new one.
 

MorgothPl

2[H]4U
Joined
Oct 13, 2008
Messages
3,020
Yes, in terms of max/average FPs you won't get much difference. But, from the tests I've been reading on various forums, not many pay attention to minimal FPS. It seems that with faster ram, the minimal FPs goes up by a lot, and that does improve gaming. People nowadays are too obsessed with max fps and do not understand that min fps is sometimes more important :)
 

Disposed

Supreme [H]ardness
Joined
Mar 2, 2010
Messages
5,206
Yes, in terms of max/average FPs you won't get much difference. But, from the tests I've been reading on various forums, not many pay attention to minimal FPS. It seems that with faster ram, the minimal FPs goes up by a lot, and that does improve gaming. People nowadays are too obsessed with max fps and do not understand that min fps is sometimes more important :)

No it don't. Maybe a tiny bit like 1 or 2 fps but not a lot. Nobody out there is obsessed with max FPS most are watching the average frame rate.

If your minimum is low enough that a couple frames actually matter you should look to other things to upgrade.
 

Michaelius

Supreme [H]ardness
Joined
Sep 8, 2003
Messages
4,684
DDR3 2400 can give up to 10% performance compared to 1600 so if you didn't have ram already it would be well worth the price premium

But since you already have 16 gigs then it's not really worth it.
 

GoldenTiger

Fully [H]
Joined
Dec 2, 2004
Messages
26,526
Yes, in terms of max/average FPs you won't get much difference. But, from the tests I've been reading on various forums, not many pay attention to minimal FPS. It seems that with faster ram, the minimal FPs goes up by a lot, and that does improve gaming. People nowadays are too obsessed with max fps and do not understand that min fps is sometimes more important :)

Completely true. Who cares if your avg is 70 when it dips during any action to 30 constantly? Especially in cpu heavy games this can be the case where high speed ram helps immensely, while most mass overclockers seem oblivious to said facts.
 

WhoBeDaPlaya

2[H]4U
Joined
Dec 16, 2002
Messages
2,571
Good link. Having said that I'm still on ddr1333 which I run at 1600 but then again I'm on an old x58 system [made new with my new Xeon however]
On X58, you'd want higher speed RAM to achieve 1:1 for your overclocked BCLK, so any performance gains due to RAM a kind of moot.

1600 -> 200 BCLK, which suited me perfectly fine for my old i7 920 D0 :)
 

WhoBeDaPlaya

2[H]4U
Joined
Dec 16, 2002
Messages
2,571
Good link. Having said that I'm still on ddr1333 which I run at 1600 but then again I'm on an old x58 system [made new with my new Xeon however]
On X58, you'd want higher speed RAM to achieve 1:1 for your overclocked BCLK, so any performance gains due to RAM a kind of moot.

1600 -> 200 BCLK, which suited me perfectly fine for my old i7 920 D0 :)
 

Curl

Limp Gawd
Joined
Jul 13, 2004
Messages
231
Like most have said, in gaming you'd be lucky to see a difference
 

Tsumi

[H]F Junkie
Joined
Mar 18, 2010
Messages
13,581
On X58, you'd want higher speed RAM to achieve 1:1 for your overclocked BCLK, so any performance gains due to RAM a kind of moot.

1600 -> 200 BCLK, which suited me perfectly fine for my old i7 920 D0 :)

Huh? 1:1 BCLK meant absolutely nothing after LGA775.
 

Tsumi

[H]F Junkie
Joined
Mar 18, 2010
Messages
13,581
True, but IIRC, Anandtech did an article showing that there were slight performance gains to be had when running 1:1.

Well, if it meant your CPU was running at 4.0 instead of 3.9 ghz, and your RAM at 1600 instead of 1500 mhz, QPI, etc, obviously there's going to be performance gains. But BCLK by itself means absolutely nothing in terms of performance for Nehalem and subsequent architectures.
 

WhoBeDaPlaya

2[H]4U
Joined
Dec 16, 2002
Messages
2,571
Well, if it meant your CPU was running at 4.0 instead of 3.9 ghz, and your RAM at 1600 instead of 1500 mhz, QPI, etc, obviously there's going to be performance gains. But BCLK by itself means absolutely nothing in terms of performance for Nehalem and subsequent architectures.
No, I meant that for a given BCLK and core speed, IIRC it was best to stick to a 1:1 divider.
 

Tsumi

[H]F Junkie
Joined
Mar 18, 2010
Messages
13,581
No, I meant that for a given BCLK and core speed, IIRC it was best to stick to a 1:1 divider.

And again, a 1:1 divider means absolutely nothing for the Nehalem platform. It only meant higher performance for the LGA 775 platform, that's it.
 

GonzoP

2[H]4U
Joined
Jul 3, 2006
Messages
3,660
On X58, you'd want higher speed RAM to achieve 1:1 for your overclocked BCLK, so any performance gains due to RAM a kind of moot.

1600 -> 200 BCLK, which suited me perfectly fine for my old i7 920 D0 :)

Yeah I'm running 200BLCK and 1600 on the ram
 

rhansen5_99

2[H]4U
Joined
Nov 12, 2001
Messages
2,153
Wow those benches are golden, goes to show you that tighter timings on your ram can make up for a frequency increase. I think the same is true when going over dual channel to triple or even quad channel for the 2011 chips, diminishing returns after a certain point on each architecture.
 

sfsilicon

Limp Gawd
Joined
Nov 1, 2005
Messages
505
16GB DDR3 1600 CL8 for $95 or 16GB DDR3 2400 CL11 for $155? 4790K stock. Worth paying 50% more for a 10% performance. Am I better off getting the 1600 kit and OC'ing my 4790K to 4.4GHz?
 

Tsumi

[H]F Junkie
Joined
Mar 18, 2010
Messages
13,581
16GB DDR3 1600 CL8 for $95 or 16GB DDR3 2400 CL11 for $155? 4790K stock. Worth paying 50% more for a 10% performance. Am I better off getting the 1600 kit and OC'ing my 4790K to 4.4GHz?

10%? More like 0-5%, on average maybe 2%.
 

Tsumi

[H]F Junkie
Joined
Mar 18, 2010
Messages
13,581
~5% from 1866 to 2400, while ~10% from 1333 to 1866. Given that 1866 usually costs ~0-5% more than 1333, while 2400 usually costs ~20-30% more than 1866...
 
Joined
Aug 10, 2014
Messages
47
You need to balance the ratio between performance increase and cost. I don't think you'll be benefiting that much considering the price difference ...
 

defaultluser

[H]F Junkie
Joined
Jan 14, 2006
Messages
14,399
Minimum recorded frame rates have no meaning except when you quantify the number of low frames (and set a standard length for the benchmark run), and include the average frame rate for comparison.

This is why sites like Tech Report include counts of low frames, as well as percentile graphs. It really doesn't matter if the number of occurrences of this "lowest" frame rate is in the teens for an entire hour of game-play (you won't notice such rare outliers), so the significance is lost if all you report is the lowest frame rate recorded for the entire test run.

http://techreport.com/review/26735/overclocking-intel-pentium-g3258-anniversary-edition-processor/3
 
Last edited:

MorgothPl

2[H]4U
Joined
Oct 13, 2008
Messages
3,020
Here's my data for 1600 vs 1866 vs 2400. I used my G.Skill CL9 1600mhz, Kingston's Savage 16GB 1866 CL9 and Kingston Savage 8GB 2400 Mhz CL11. I've run about 20 benchmarks of each software and picked the highest. Differences are barely noticeable, and in some of the benchmars my 1600 scored higher than 2400.

1600:
Tomb raider
Min 52
Max 90
Average 69.41

Sleeping dogs
Min 37.2
Max 63.2
Average 54.1

Valley
Min 22.4
Average 54.3
Max 106.9
Score 2271

Kingston 1866

Tomb raider
Min 52
Max 82
Average68.3

Sleeping dogs
Min 38.4
Max 63.7
Average 54.3

Valley
Min 26.8
Max 102.8
Average 54.1
Score 2262

Kingston 2400
Tomb raider
Min 54
Max 82
Average 68.2

Sleeping dogs
Min 38.2
Max 64.6
Average 54.2

Valley
Min 25.9
Max 108.2
Average 54.9
Score 2295
 
Top