Consumercarts

Hardware finder

RAM Comparison

Updated October 10, 2026

840 RAM kits from 29 brands.

Consumercarts is reader-supported. Learn more

What these specs mean

Showing 840 RAM found

About RAM

Memory kits listed by the figures that actually differ: DDR generation, rated speed in MT/s, CAS latency, and the true latency in nanoseconds that combines the two.

Generation is not optional. A DDR5 module will not fit a DDR4 slot. Our laptop upgrade checker filters to the modules your specific laptop accepts.

Speed alone does not decide it: read true latency vs speed rating, 32GB vs 64GB and our DDR5 price tracker. Popular comparisons: Corsair Vengeance LPX 16GB vs 32GB, Vengeance LPX 32GB vs Vengeance RGB Pro 16GB and CMK16GX4M2B3200C16 vs CMK16GX4M2E3200C16.

Related categories

Guides for RAM

Popular comparisons

Guides

How to choose a memory kit

Buy a kit, not sticks. Two modules sold together are tested together, and mixing two separately bought pairs is the most common cause of a machine that will not post at its rated speed.

Speed and latency together, never separately

True latency in nanoseconds is the only honest way to compare across speeds: CAS latency multiplied by 2000, divided by the rated speed. DDR5-6000 CL30 is 10.0ns and beats DDR5-6400 CL40 at 12.5ns, even though every retail listing sorts by the bigger number.

Capacity

16GB is the working floor. 32GB is the sensible target for gaming plus browser plus anything else open. 64GB and above is for virtual machines, large timelines and heavy datasets, not for headroom.

The platform decides the sweet spot

On AM5, DDR5-6000 with tight timings is the practical limit for most chips because the memory controller runs 1:1 there. Pushing past it often forces a slower divider and ends up net negative. Intel is more tolerant of higher speeds but gains less from them.

EXPO and XMP are the same idea

Rated speed is an overclock. Out of the box a kit runs at the JEDEC base speed until you enable the profile in BIOS. A kit bought for its rating and left unconfigured is running slower than a cheaper kit that was configured.

Physical fit

Tall heatspreaders collide with large air coolers. Check module height against your cooler before buying, especially with 140mm towers.

Common mistakes

  • Adding two sticks to two existing sticks and expecting the rated speed to hold.
  • Buying DDR5 for a DDR4 board. The generations are not compatible and the slots are keyed differently.
  • Paying for RGB and treating the light show as a performance feature.

FAQs

Will DDR5 work in a DDR4 motherboard?

No. The two use physically different notches and are not interchangeable. The board supports one generation only, and it is stated on the motherboard spec sheet.

Is a lower CAS latency always better?

Not on its own. Latency in nanoseconds is what you feel, and it combines CAS latency with speed. A CL36 DDR5-6000 kit is faster in real terms than a CL30 DDR5-4800 kit.

Do I need two modules or is one enough?

Two, in almost every case. Dual channel roughly doubles memory bandwidth, and a single module leaves performance on the table, particularly with integrated graphics.

What are XMP and EXPO?

Stored profiles that run the kit at its advertised speed. Without enabling one in the BIOS, the memory runs at the slower JEDEC default, whatever the box says.

Does memory height matter?

Yes, if you use a large air cooler. Tall heatspreaders can foul the cooler over the first slot. Check the cooler RAM clearance figure or choose low profile memory.

How do I compare RAM kits?

Match the generation to your motherboard first (DDR4 and DDR5 do not fit each other's slots), then pick capacity, then compare true latency in nanoseconds, which combines speed and CAS latency into one number. Lower true latency is faster.