4000mhz cl20 = 3000mhz cl 15?

[Zn1P]

I am looking at 2 ram sets, one is 4000mhz cl 20 and the other one is 3000mhz cl15. Using a ram latency calculator I see, that true ram latency is actually the same - 10ns. So why is the first one 2x more expensive? Are there any other  performance factors here?

[r2724r16]

Just now, Zn1P said:

So why is the first one 2x more expensive?

Maybe to fool potential customers?

[Fasauceome]

Ram speed and latency are 2 different things. The latency timing does tell you what the latency of the ram is using the multiplier, but the ram speed affects things that the latency does not. The kits have the same latency but the 4000MHz kit is still faster.

Edited January 4, 2019 by fasauceome

[Sakkura]

The first kit has 33% higher memory bandwidth.

[mariushm]

The first is more expensive probably because fewer chips are actually capable of that 4000 Mhz.

 

May be a consequence of heavy binning - they may go through loads of memory chips and get just a few that are stable at 4000 mhz.

 

Second reason could be demand ... few people actually want to buy 4000 mhz and consider 3000 or 3200 enough... the performance benefits drop above this threshold anyway. And Ryzen cpus have a hard time going over around 3600-3800, so there goes a large percentage of market....

[LukeSavenije]

the 4000 will actually be faster, but I don't see why anyone would buy it except for money burning or bragging rights

[porina]

I'm actually in the market for a 4000-ish kit, to help with competitive benchmarking/overclocking. #firstworldproblems It certainly isn't on the value curve at that point. Also my other compute interest is finding prime numbers, and that is often ram bandwidth limited so cores can't run at their full potential. A dual channel 4000 kit would be great for quad core Intel CPUs, and maybe be adequate for 6.

 

Most ram I've bought up to this point is around 3000 rating because of cost.

[Jurrunio]

high end stuff tends to drop off in value perspective, that's just how it works.

[Zn1P]

The question was actually not about the price, but about the performance. I thought, that high mhz values are outweighted by also high CL. Thus getting the same performance. Isn't that the case? Or is it like this: it's the same latency, but you get to move more data?

 

[Sakkura]

1 hour ago, Zn1P said:

The question was actually not about the price, but about the performance. I thought, that high mhz values are outweighted by also high CL. Thus getting the same performance. Isn't that the case? Or is it like this: it's the same latency, but you get to move more data?

 

The clocks are a little more important than the timings, because the clocks influence both the (actual) latency and the bandwidth. The timings only affect the latency.

 

In this case, the actual latency is exactly the same, but the bandwidth is higher. Which means when the CPU asks for some data, it has to wait equally long for it to start arriving, but once the data starts to arrive, it comes in significantly faster.

[nick name]

What you will most likely find is that the 4000MHz CL20 kit could be set to run at something close to 3000MHz CL12 whereas the 3000CL15 will almost certainly not.  Those higher speed kits are usually binned better and many folks buy them not only for speed, but their ability to achieve tighter timings.  

 

So what I would expect is that the 4000MHz kit will run 3600MHz at 14-14-14-14 (depending on platform -- Ryzen might be 14-15-14-14) and I would almost guarantee that the 3000MHz kit will not get close to that.  

[xg32]

3 hours ago, Zn1P said:

I am looking at 2 ram sets, one is 4000mhz cl 20 and the other one is 3000mhz cl15. Using a ram latency calculator I see, that true ram latency is actually the same - 10ns. So why is the first one 2x more expensive? Are there any other  performance factors here?

4000/20 and 3000/15 are both equal, but in the case of being equal, the higher frequency set is usually better as a rule of thumb.