Timings - CL16 @ 3200Mhz = CL18 @ 3600Mhz?

[OpenSource Ghost]

From my research, higher clocks benefit AMD Ryzen systems more than CL value, but what about newer Intel CPU's like Alder Lake and Raptor Lake? I'd like to see some actual benchmark results that demonstrate whether "CL16 @ 3200Mhz = CL18 @ 3600Mhz = CL19 @ 4000Mhz" concept is true or not. Is there an actual formula that calculates theoretical bandwidth based on memory clocks and timings? The latest reviews seem to completely omit comand rate (1T vs 2T) from reported timings. Isn't command rate of more importance to performance than CL values?

 

[dasa]

It will vary from one program to the next depending on its preference for bandwidth vs latency.
If you are buying RAM now for a new Intel platform you should probably just go DDR5.
DDR4 is still a decent option for those that already have a good kit of RAM that they want to keep.

 

[pitingres]

3200CL16 is exactly the same CAS latency as 3600CL18, 10 nanoseconds. CL is listed in clock ticks, and faster transfer rates have higher clocks / shorter clock ticks.

The formula is (1000/(MTs / 2)) * CL gives CAS latency in nanoseconds. MTs is the transfer rate in MT/s, universally mis-labeled as MHz. DDR runs the clock at half the transfer rate, hence the divide by 2. For example, (1000/(3200/2))*16 = (1000/(3600/s))*18 = 10 ns. 4000CL19 works out to 9.5ns CAS latency.

 

[III_Slyflyer_III]

It really depends on use case... within reason, games prefer lower latency vs bandwidth (assuming you have enough bandwidth to begin with). I run my DDR4 on my Ryzen at 3800Mhz CAS 16 and I have had it decoupled from the IF and achieved as high as 4200Mhz CAS 18. Guess what? In gaming benchmarks like Superposition and 3DMark, there was virtually no difference outside the margin of rebooting error. I was actually slightly lower with the 4200Mhz because of the decoupled IF.

All else being equal though, I'd always go for the higher speed as long as latency does not increase much or is at 10ms or lower.