Someone having experience with those more expensive but possibly "faster" RAM kits? Pc will be: i5 4690K @4,2 - 4,6 Ghz Z97 Motherboard GTX 560Ti
True, the effect on RAM speed on the performance of games is massively overrated. What will limit your performance there is the graphics card.
As is was already said: you may benchmark differences but in real application it matters zero. So simply get the cheaper option.
The above video is misleading: they're benchmarking games which are obviously GPU bound, probably more specifically by vram bandwidth, thanks to high resolution+ultra details+high AA. When you deem your framerate too low, you'll obviously not play with full effects on and some 64x MSAA at 4K resolution! This is ridiculous, and the numbers shown actually demonstrate it: 26fps and 32fps on average in the 2 games. The cinebench cpu benchmark is obviously made to test the cpu, and not the memory so it can't prove anything, then the openGL cinebench benchmark actually shows a 10% increase between the PC800 and the PC1866. 10% is significant in terms of performance in my book, though i don't know what that benchmark actually does. Then after using settings nobody would use, they conclude their dumb benchmarks by saying that real life applications are going to be virtually unaffected. A significant test would have to benchmark several mainstream games with different 3D engines, and in different circumstances (gibs everywhere, lots of bots, max/mid/low details/lower resolutions). You can't do that in a 4mn video, obviously. Even then, PS2 might be an entirely different animal than the games they'd test: obviously the usual unreal engine games, BF, CoD would make it in such a test...Forgelight? i doubt it.
Faster ram usually helps on the games where you have lots of things going on, like on MMO usually has.. on a low action game like BF 3/4 or CS it doesn't help at all!
Actually BF4 is the game where RAM has the highest impact out of all games. Iirc you could achieve 8% difference in FPS with significantly faster RAM.
Right, written by Corsair employee Dustin Sklavos in a Corsair marketing blog promoting Corsair's uber 1337 RAM and tested on an X79 SLI system which everyone knows is one of the few scenarios where fast RAM actually makes a difference none of which applies to OP at all.
Naah, only memory that matters in most games is the boost u get between 1333mhz vs 1600mhz and on some games with faster memory! All the test i have been looking says the BF 3 or 4 doesn't scale with better memory and is most likely always GPU bound, same fps with almost all intel CPU:s!
That's not what I was referring to. There are a few tests regarding that topic which compared different RAM speeds between 1333MHz and 2100MHz. Well, 1600MHz is quasi standard anyway. I meant higher speeds.
I used 2.4ghz ram when I built this PC just to ensure RAM wasn't an issue. Have no clue if it makes a difference.
If I had the cash I would get 2400Mhz memory simply because I could. However if you're on a budget and the cash you spend on the ram is eating away on another components budget then I would go with the 1600Mhz ram. Even if you do get the higher 2400Mhz ram and you find it makes 0% difference in gaming then you might as well put it to other uses. I would personally underclock it to 1600-1866Mhz then tighten the timings. That should speed up your memory access times which can help when multitasking or moving loads of small files. You'll notice some difference either way.
If they were the same price, then go for higher Hz over a lower CL. Latency decreases as Hz increases, so the lower figure is slightly misleading. You're generally (i.e. within reason) better off spending more money on capacity over speed, otherwise go for the fastest you can afford. As others have noted, memory bandwidth is rarely the bottleneck in games, as much of the high-bandwidth memory work is done by VRAM (textures and the like). Games don't especially tax system memory and there are plenty of benchmarks that illustrate it. That at least, is the world as it is currently. Looking forward, GPU manufacturers are now providing APIs that utilise unified memory, where the graphics card can eat into system memory, in this situation all the talk about not needing bandwidth in memory or PCIe may start to reverse. We'll have to wait and see. There is also the move to DDR4 starting next month with X99. The next round of Intel processors, starting with SkyLake, will be DDR4 too. No doubt there will be some backward-compatibility provided by motherboard manufacturers, but the economies of scale will switch in the next couple of years. So... if you're planning on upgrading in the next couple of years, I'd probably just go with whatever is cheap and save the money for then.
I thought it's the opposite--latency has a direct relationship with clock speed, i.e. clocks go up, latency increases too. Can you explain this please?
The advertised CL latency is not an absolute value but simply the number clock cycles to complete a certain operation. As the frequency goes up, these clock cycles become shorter and so the absolute latency which is measured in nanoseconds becomes lower.
