Best Computer Memory: March 2015

[Geef]

I sort of won silicon lottery back in the day.

Corsair CMU16GX4M2C3000C15 DDR4

It is it is able to run at C14 at 3100MHz. Not a huge increase but with 16GB of it, I decided to just keep using until I finally upgrade motherboard to a DDR5 board.

The top pick for DDR5 looks awesome and I will definitely save that for future upgrade plans. Thanks!

 

[dmitche3]

So, I don't see why this article is called best memory for gaming. Are you implying that if you simplying buying the best and fastest memory, whether a single unit or not that it is specifically only the best for gaming? I think not.

 

[InvalidError]

So, I don't see why this article is called best memory for gaming. Are you implying that if you simplying buying the best and fastest memory, whether a single unit or not that it is specifically only the best for gaming? I think not.

Different applications respond differently to various memory parameters. Memory that performs generally best for gaming has a different balance of bandwidth vs latencies than video compression, 3D rendering, data compression, physics simulation, etc.

Since higher bandwidth usually comes at the expense of worse latency, there is no single memory that is simultaneously best at everything and you have to make a compromise: spend a little extra over generic memory to prioritize specific workload categories at the expense or others, spend a large premium on bleeding-edge memory that performs closer to the best across a broader workload spectrum or keep your money by buying something that should be good enough for everything instead of fretting about losing 10-20% performance in some cases.

Within the gaming realm, sticks often trade places on a title-by-title or in some cases, scene-by-scene basis.

 

[phxrider]

Huh.... I bought the #1 rated G.Skill Trident Z5 Neo RGB DDR5-6000 EXPO CAS 30 memory when I built my PC (not based on this article), and it wouldn't even POST on EXPO. I returned it and bought the even lower-latency Corsair 6000 CAS 30 memory, which worked perfectly.

And ironically, it benches faster when clocked at 5600 with 28-34-34-71 timings than at the advertised 6000MHz and 30-36-36-76 timings.