Question Can I Overclock My RAM Over Its Limit?

[ink123]

Hİ, I have these ram: https://www.amazon.com./RipjawsX-bellek-1600-MHz-240-pin-DDR3-RAM/dp/B007CCV1TI

Is it possible to work them over 1600mhz?

Rest of my rig:

CPU: Intel I7 4790k
GPU: MSI GTX 1070 Ti
Motherboard: MSI GAMING 5 Z97
PSU: GX 750W 80+ Bronze
Hard drives: Samsung EVO 860 500gb m2 (Windows drive), Kioxia-exceria 480gb sata ssd, WD 1TB Caviar Blue 7200 rpm SATA 3.0 hdd

 

[Lutfij]

You could try and perhaps tighten the timings on that ram kit or you could go with a higher frequency and loose timings. Short answer is, you will need to try your hand at overclocking. G.Skill have generally been good/forgiving with overclocking though in retrospect, you're better off buying a ram kit that is advertised to go to DDR3-2133MHz, that's the sweet spot for the 4th Gen i5/i7 platform. Please make sure you're on the latest BIOS version before you attempt overclocking and ofc, that your airflow in the case is more than optimal. What cooler are you working with to cool that processor?

PSU: GX 750W 80+ Bronze
That is not a reliably built PSU.

 

[Karadjgne]

Sweet spot for 4th gen intel was 1866MHz. Which should be easily doable with those sticks at their current timings. 1333MHz/1600MHz is default Intel memory controller speeds, so 1600 was the least stressful and most stable out of the box. 1866MHz is an improvement, but not so big a jump where you needed really to mess with the ram settings and timings.

Timings and speed went hand in hand with Intel ddr3, overall throughput was almost identical for 1600/C9 and 1866/C10 or even 2400/C12. There are very few programs that have a strong reliance purely on ram speeds so getting higher speed ram often doesn't do much other than stress the ram, unless you can tighten the timings also by a notch or 2.

1600/C8 is generally faster overall than 1866/C10, for instance, or 2133/C11. Ppl were getting incredible results with Corsair Dominator sticks, seeing 1600/C6 or 1866/C7, which is seriously fast throughput (time it takes for data to enter, get sorted and shipped out).

 

[Endre]

Hİ, I have these ram: https://www.amazon.com./RipjawsX-bellek-1600-MHz-240-pin-DDR3-RAM/dp/B007CCV1TI

Is it possible to work them over 1600mhz?

Rest of my rig:

CPU: Intel I7 4790k
GPU: MSI GTX 1070 Ti
Motherboard: MSI GAMING 5 Z97
PSU: GX 750W 80+ Bronze
Hard drives: Samsung EVO 860 500gb m2 (Windows drive), Kioxia-exceria 480gb sata ssd, WD 1TB Caviar Blue 7200 rpm SATA 3.0 hdd

Hello!

Yes, you could, but should you?
The benefits are marginal (about 1% increase in real world applications).
Also, you might face future instability issues!
Personally, I wouldn't do it!

 

[ink123]

You could try and perhaps tighten the timings on that ram kit or you could go with a higher frequency and loose timings. Short answer is, you will need to try your hand at overclocking. G.Skill have generally been good/forgiving with overclocking though in retrospect, you're better off buying a ram kit that is advertised to go to DDR3-2133MHz, that's the sweet spot for the 4th Gen i5/i7 platform. Please make sure you're on the latest BIOS version before you attempt overclocking and ofc, that your airflow in the case is more than optimal. What cooler are you working with to cool that processor?

PSU: GX 750W 80+ Bronze
That is not a reliably built PSU.

I have cooler master hyper evo 212 as my CPU cooler. The reason why I want to overclock my ram is to reduce memory bottlenecking. Can I reduce memory bottlenecking in games by overclocking my ram? I am on the latest BIOS version by the way.

 

[Zerk2012]

I have cooler master hyper evo 212 as my CPU cooler. The reason why I want to overclock my ram is to reduce memory bottlenecking. Can I reduce memory bottlenecking in games by overclocking my ram? I am on the latest BIOS version by the way.

What makes you think your memory is limiting your performance?

Manually set your speed to 1866 and the voltage to 1.56 and see if it's stable.

 

[geofelt]

Ram chips are binned and the better chips are used in higher speed products that can sell for more. You can try to oc, you might get lucky.
But, do not count on it.
Other that integrated graphics performance, you are unlikely to be able to see any real improvement.
Here is an older study on DDR3 ram speed scaling:
https://www.anandtech.com/show/7364/memory-scaling-on-haswell/2

How did you determine that memory was an issue?

I might guess that working on overclocking the 4790K might give better results.
My experience with doing that long ago required adjusting the ram speed down.

 

[zx128k]

Its not worth the effort. The amount of testing you would have to do to make sure its stable, is not worth the little gaming performance you would get.

 

[ink123]

Ram chips are binned and the better chips are used in higher speed products that can sell for more. You can try to oc, you might get lucky.
But, do not count on it.
Other that integrated graphics performance, you are unlikely to be able to see any real improvement.
Here is an older study on DDR3 ram speed scaling:
https://www.anandtech.com/show/7364/memory-scaling-on-haswell/2

How did you determine that memory was an issue?

I might guess that working on overclocking the 4790K might give better results.
My experience with doing that long ago required adjusting the ram speed down.

Before making my ram dual channel, both my GPU and CPU usage was lower. Now both my GPU and CPU usage are higher, still they are not at the level where I want them to be. I still don't get a stable GPU usage over 90% in triple a games.

 

[ink123]

What makes you think your memory is limiting your performance?

Manually set your speed to 1866 and the voltage to 1.56 and see if it's stable.

Click to expand...

Before making my ram dual channel, both my GPU and CPU usage was lower. Now both my GPU and CPU usage are higher, still they are not at the level where I want them to be. I still don't get a stable GPU usage over 90% in triple a games.

 

[geofelt]

Before making my ram dual channel, both my GPU and CPU usage was lower. Now both my GPU and CPU usage are higher, still they are not at the level where I want them to be. I still don't get a stable GPU usage over 90% in triple a games.

Nor do you want to.
CPU usage and gpu usage do not operate concurrently.
If you saw either 100% cpu or 100% gpu usage you would conclude that the cpu or the gpu was a limiting factor.(some call this a bottleneck)

 

[Karadjgne]

Before making my ram dual channel, both my GPU and CPU usage was lower. Now both my GPU and CPU usage are higher, still they are not at the level where I want them to be. I still don't get a stable GPU usage over 90% in triple a games.

You misunderstand usage. Usage isn't how much of the cpu/gpu is Used, but how much it Uses. There's a difference.

What's uses is resources. That includes time, bandwidth, size, speed etc. With game code strings, there's gaps, one string doesn't butt up against the next. The cpu core can only actively work on one string at a time, wait for the gap, start next string. There's also only so much bandwidth available per core, and a string might only use 10% of that core, divided by HT, so there's a chunk of bandwidth not being used between the 2 threads.

So a cpu is going to work at 100% ability on the % of resources it needs to get the job done. By switching to dual channel ram, you've increased the availability of resources aimed at the cpu, so the cpu can do more.

The absolute last thing you want is usage getting anywhere near 100%. At 100%, there's nothing left, every resource is absolutely slammed to the gills. So if you are at 100%, and your toon is just running doen the street, when a tank explodes out of the wall, and the cpu has to organize and process all the debris vectors, shadows, lighting etc for 100 chunks of wall flying around. And the cpu has nothing extra to work with. That'll mean the cpu has to make time to work with those things, so fps suffers, a lot.

If the cpu was at 80%, it'd have breathing room to allow for the extra hit, so fps would only drop maybe a little.

 

[ink123]

Nor do you want to.
CPU usage and gpu usage do not operate concurrently.
If you saw either 100% cpu or 100% gpu usage you would conclude that the cpu or the gpu was a limiting factor.(some call this a bottleneck)

You misunderstand usage. Usage isn't how much of the cpu/gpu is Used, but how much it Uses. There's a difference.

What's uses is resources. That includes time, bandwidth, size, speed etc. With game code strings, there's gaps, one string doesn't butt up against the next. The cpu core can only actively work on one string at a time, wait for the gap, start next string. There's also only so much bandwidth available per core, and a string might only use 10% of that core, divided by HT, so there's a chunk of bandwidth not being used between the 2 threads.

So a cpu is going to work at 100% ability on the % of resources it needs to get the job done. By switching to dual channel ram, you've increased the availability of resources aimed at the cpu, so the cpu can do more.

The absolute last thing you want is usage getting anywhere near 100%. At 100%, there's nothing left, every resource is absolutely slammed to the gills. So if you are at 100%, and your toon is just running doen the street, when a tank explodes out of the wall, and the cpu has to organize and process all the debris vectors, shadows, lighting etc for 100 chunks of wall flying around. And the cpu has nothing extra to work with. That'll mean the cpu has to make time to work with those things, so fps suffers, a lot.

If the cpu was at 80%, it'd have breathing room to allow for the extra hit, so fps would only drop maybe a little.

So, having stable GPU usage in games between %95-99 is a good thing or a bad thing? Having a low GPU usage something around %70, does it mean that your system is bottlenecking? Considering you are not limited by v-sync or the refresh rate of your monitor.

 

[geofelt]

There is no such thing as "bottlenecking"
If, by that, you mean that upgrading a cpu or graphics card can
somehow lower your performance or FPS.
A better term might be limiting factor.
That is where adding more cpu or gpu becomes increasingly
less effective.

What counts is how satisfied you are with your gaming performance.

Your $340 I7-4790K was top dog when it was launched some 9 years ago.
No longer.
Today, $250 buys you a I5-13500 that is about twice as strong.
Note that I3/i5/i7/i9 no longer designate the number of cores and the presence of hyperthreading.

Game performance will be limited by the cpu or the gpu.
And, not at the same time.
As a simple test,
Run YOUR games, but lower your resolution and eye candy.
This makes the graphics card loaf a bit.
If your FPS increases, it indicates that your cpu is strong enough to drive a better graphics configuration.
If your FPS stays the same, you are likely more cpu limited.

 

[zx128k]

Higher the resolution the more likely to be GPU bound and the lower the resolution the more likely you are to be cpu bound.

 

[Karadjgne]

Ok, I'll explain it more simply, as I don't feel you are quite grasping the usage concept, which isn't exactly an easy concept to grasp.

You go to hang a picture. That's going to require you put a nail into the wall. When you grab the hammer, it's going to mean that you will use 100% of the muscles in your hand, every tendon, every ligament is used. That's what the gpu does, it uses everything it has, all 100%.

Now you go to swing the hammer, you'll again use every muscle in your wrist and arm.

Usage isn't what's used, it's what uses. So gripping that hammer, swinging that hammer, you'll use every muscle, but you will Not use all the strength you have in either hand nor arm. You do not require a death grip on the hammer, nor require swinging that hammer with all your might. The job does not need or require that you bring to bear All your strength.

That's usage. The amount of resources (strength) you need to use to get the job done. Increasing difficulty, like if you find a stud, may require you to swing harder, use more strength, to get the nail sunk enough (higher complexity frames) . If you were at 100% usage, you could not swing the hammer harder, so the nail needs to be hit more often, that makes the job longer per nail, so pictures hung per hour (fps) goes down.

1% usage or 99% usage or anything in between is the same amount, all that's necessary to get the job done. The Only number that differs is 100% usage, because that's a guarantee that if just 1 small factor, one tiny particle, adds to complexity of the frame, you have Zero strength headroom, fps will drop. You are Far better off with a gpu at 70% usage than a gpu at 99% usage, you have far more headroom for complexity changes, like explosions, going into heavier npc Ai populated areas, fields etc.

Switching to dual channel ram means the ram is shoveling @ 1.5x more data per cycle than single channel. That means the cpu no longer can sit on its butt and take its sweet time per frame, it's now in a hurry to process all that extra data. Usage goes up. Also means it's sending more info per packet to the gpu. Gpu usage goes up. Fps goes up because detail levels are not having as great an affect on the fps packets sent by the cpu. That can mean smoother frame transitions, less stutter, less impact on v-sync...

The way ram channels work is parachuters jumping out of an airplane. Single channel is the guys in a straight line jumping, waiting for the guy in front to clear, next guy jumps. In a straight line, so there has to be a wait period for the guy to clear the tail so you don't hit him.

With dual channel, it's 2x lines and you don't jump straight out, but on an angle, an X, / then \. So right after one guy jumps to the left, another from the other line can jump to the right. So while you still have the wait to clear period, that's only affecting 1 line at a time, the other line is already clear to jump, filling in the gap.

Single channel : °/°/°/°/°
Dual channel : %%%%%