Question Effect Of Subtimings

[DefinitelyNotTom]

I can never get a straight answer on this from someone who seems to flat out know the answer.

How much worse is 3600 16-18-18 Hynix than 3600 16-16-16samsung b die when leaving it at those autoamtic settigns?

And how do they compare with 3200 15-15-15 b die and 3200 14-14-14 b die?

I am talking for productivity stuff.

I am selling my 15-15-15 kit due to it not being on qvl and would rather get a little better kit for the trouble and loss of money.

If the Hynix would just be a lateral move, I may not get it, HOWEVER the Hynix kit is $160 and the Samsung one is freakin $280.

 

[Darkbreeze]

There's no secret to the equation. True latency tells you how fast memory is. ANY memory. Here is the formula. (Credit to Computronix for sharing his example)

1 / Frequency (not DDR) x Latency = True Latency (nanoseconds).

Stock 3200 @ 14 is faster than Stock 3600 @ 16:

1 / 1.600GHz x 14 = 8.75nS
1 / 1.800GHz x 16 = 8.89nS

Stable Overclock with 3733 @ 16 is faster:

3733 @ 16 is 1 / 1.867GHz x 16 = 8.57nS

Other primary, secondary and tertiary timings CAN have an effect on latency, but generally speaking the effort is not worth the outcome to make overclocking the secondary and tertiary timings worth messing with unless you are simply doing it as a hobby and are willing to do the VERY EXTENSIVE amount of testing that is required to verify stability anytime you start going outside the configuration assigned by the manufacturer.

 

[DefinitelyNotTom]

There's no secret to the equation. True latency tells you how fast memory is. ANY memory. Here is the formula. (Credit to Computronix for sharing his example)

1 / Frequency (not DDR) x Latency = True Latency (nanoseconds).

Stock 3200 @ 14 is faster than Stock 3600 @ 16:

1 / 1.600GHz x 14 = 8.75nS
1 / 1.800GHz x 16 = 8.89nS

Stable Overclock with 3733 @ 16 is faster:

3733 @ 16 is 1 / 1.867GHz x 16 = 8.57nS

Other primary, secondary and tertiary timings CAN have an effect on latency, but generally speaking the effort is not worth the outcome to make overclocking the secondary and tertiary timings worth messing with unless you are simply doing it as a hobby and are willing to do the VERY EXTENSIVE amount of testing that is required to verify stability anytime you start going outside the configuration assigned by the manufacturer.

There's more to it than that equation, though, because higher frequency still performs better than lower frequency. It's not all down to latency. Ryzen has always been known to be better with higher frequencies, regardless of latency. If 3200 cl14 was better overall than 3600 cl16, then the latter wouldn't cost so much more.

 

[Darkbreeze]

Ok dude. Since you (believe) know so much, I don't know why you even bother posting questions here anymore. In fact, maybe you should just NOT do that anymore. What's the point when you simply tell people they are wrong in practically every thread you create?

People get tired of trying to help you just to hear how it's wrong, or not good enough, or you can't make up your mind. And in ALL of the threads I've looked at by you, I have NOT ONCE, NOT ONE SINGLE TIME, seen you tell anybody "thanks" for any of the help they've tried to extend to you. Don't worry, you won't have to worry about me posting anymore in any thread you create.

Good luck.

And by the way, you're absolutely clueless in this regard. Because CL 14 3200mhz sticks ARE faster than 3600mhz sticks at CL16. But it's obviously a waste of time to try and explain why so I'm not going to. EVERYTHING is down to latency. Always.

 

[DefinitelyNotTom]

Ok dude. Since you (believe) know so much, I don't know why you even bother posting questions here anymore. In fact, maybe you should just NOT do that anymore. What's the point when you simply tell people they are wrong in practically every thread you create?

People get tired of trying to help you just to hear how it's wrong, or not good enough, or you can't make up your mind. And in ALL of the threads I've looked at by you, I have NOT ONCE, NOT ONE SINGLE TIME, seen you tell anybody "thanks" for any of the help they've tried to extend to you. Don't worry, you won't have to worry about me posting anymore in any thread you create.

Good luck.

And by the way, you're absolutely clueless in this regard. Because CL 14 3200mhz sticks ARE faster than 3600mhz sticks at CL16. But it's obviously a waste of time to try and explain why so I'm not going to. EVERYTHING is down to latency. Always.

Think what you want, but G.skill told me that the higher bandwidth of the 3600 makes a big deal for good cpus. Beyond that they didn't elaborate, but I'm going to assume it means once you're doing something intensive the 3600 would become faster regardless of latency because the 3200 bandwidth would be used up quicker.

That's all I said was that it's not "always" only down to latency and you're mad about me stating that truth, Think about it... if 3200 cl14 was better in all ways than 3600 cl16 then why would ANYONE buy 3600, since there is barely any even available with lower latency than 16 and it's insanely expensive if it is? There'd be no point in going above 3200 for anyone, but many do. bandwidth matters.

All I asked was how much do subtimings matter and as far as why ask if I knew.... I asked before g.skill answered. They also said even the main timings are basically for enthusiasts. So basically you're calling me clueless for saying something that G.skill support themselves said. It's not remotely "all" down to latency.

 

[DefinitelyNotTom]

And Crucial also says speed trumps latency, so there's another company you're calling clueless by saying "everything" is down to latency" https://www.crucial.com/usa/en/memory-performance-speed-latency

From that article: "Based on in-depth engineering analysis and extensive testing in the Crucial Performance Lab, the answer to this classic question is speed. In general, as speeds have increased, true latencies have remained approximately the same, meaning faster speeds enable you to achieve a higher level of performance. True latencies haven't necessarily increased, just CAS latencies. And CL ratings are an inaccurate, and often misleading, indicator of true latency (and memory) performance."

 

[DefinitelyNotTom]

what your formula is showing is true latency only, as described with the slightly different formula (but gives same results) here: https://www.techpowerup.com/forums/threads/ddr4-frequency-vs-latency.232240/

which means... the latency is almost the same on the 3200 kit and the 3600 kit before you even consider the speed itself. Speed and latency aren't the same thing and latency being lower doesn't make the 3200 faster.

 

[bignastyid]

And yet you wonder why nobody wants to help you. Since you seem to no longer need assistance with this topic I'm closing the thread.