What is the difference between DDR 4 and DDR 5?

I am looking to buy a new RAM for my system. I am thinking to purchase a DDR 4 RAM (https://www.allhdd.com/hpe-805347-b21-memory/) but I am not sure if I should buy a DDR 4 or DDR 5 because we usually don’t buy this stuff regularly.

1 Like

If your motherboard supports DDR5, I say go for it! Take advantage of the current tech (more and faster).

Also that link you added shows ECC memory - just double check your motherboard supports that. -EDIT N/M, assuming that you are looking at server memory you’re probably already on top of that

ddr 5 is way more expensive, a little faster

1 Like

How do you define that?

My understanding is that with the same frequency it doubles the bandwidth.

Also, it may be / is more power efficient (runs at 1.1 V, instead of 1.3).

Although, frequency is not the only metrics, as those CAS, etc. may vary a lot, as such can degrade performance a bit (even though frequency is high).

Basically, for a browser, DDR 3 is good enough, for a high-end plotter DDR 5 would be preferred (sure, cost permitting).

1 Like

I am using an old DELL inspiration system. How can I find out if it support DDR 5 or not?

I see you are linking server memory. You are aware that DDR5 is not a drop-in replacement for DDR4? You’ll need a new server/system that supports DDR5 modules.

1 Like

Double-Data rate != double speed.

In a best case scenario, you can almost reach double speed.
However, in practice there are usually countless factors involved (latency, cpu, …)

So that the effective speed bump is mostly marginal:

In our overall performance measurement, DDR5-4800 C40 was 19% and 14% faster than DDR4-2133 C15 and DDR4-3200 C22 , respectively. However, the performance delta decreased when we compared it to performance DDR4. DDR5-4800 C40 was only 9% faster than DDR4-3200 C15.

Note that the first test from Thoms Hardware is irrelevant because its like comparing a new ferrari vs an old fiat.

Second test, when you compare old ferrari vs new ferrari you gain ,10%
Additionally, this is a pure ddr bandwith benchmark. It tells little about the real world gains, but mor on that in the following section:

so you will likely gain something between 1-30% in the final workload (plotting)
That beeing said, I appreciate if someone could do real world plotting benchmarks

If you want to read the full article:


Being an old Dell system it won’t, it’s likely either DDR3 or DDR4.

Either use the Dell service tag ID on the Dell website, or Crucial Advisor tool | Crucial.com to find out what type of memory you require - you can’t fit a different type.


That is exactly what I wrote about those CAS, etc. That C40 is basically the speed killer. Also, I take that that comparison was done early on, when manufacturers were pushing slow memory to high speeds, while keeping the overall speed down with those CAS settings.

Basically, the same applies to any DDR RAM, not just DDR 5. One needs to compare the same frequency with the same CAS settings.


Google your Dell model number for spec sheet, that will have the RAM specs as well.


How chia plotting benefit from mem bandwidth vs penalty of ram latency for ddr4 vs 5 is not clear. as no testing result published yet.
But from the experience from dd3 server vs ddr4 server plotting, seems chia plotting is not too sensitive to ram latency as gaming. Plotting is more like bandwidth dominated application. If cpu core number is not the bottleneck, we can assume 100% mem bandwidth increase will make plotting 60%-80% faster , which is much higher than gaming 's 10-20% gain.